AI Catalyst C3·Core Sessions - Week 4·2:58:38

Session 7: The Outreach Engine

Harshit Trainer/mentor (Sessions 2-4 trainer, returning) — YC-startup engineer (JustPaid.ai), exited founder (ArtEffect), runs his own consulting practice and AI-avatar studio; claims 500k+ cold emails sent. Theory-first session: fundamentals today, Instantly hands-on tomorrow. · Niharika Cohort manager — open/close, break timing, CSAT poll, Slido launch

▶local recording · LMS lesson

Session map

Engine & offerDomains & deliverabilityCopy & skills

click a node — its card pops up (drag it anywhere, × to close)

Concept

The map reads left to right — engine & offer flow into domains & deliverability, then into copy & skills. Click any node to open that idea here; every timestamp jumps into the recording.

The short version

The outreach engine in one whiteboard: an OFFER (product/service/coaching) plus a LEAD LIST (scraped strangers who fit your ICP) flow through the tech (Instantly) toward one outcome — MEETINGS — with a 'trash can' catching what goes wrong: spam placement, high bounce rate, bad copy. Cold email is NOT email marketing: marketing goes to people who opted in; outreach goes to people who've never heard of you.
Offer construction has a formula — audience + outcome + mechanism + pain avoided — turned live into a reusable skill.md ('I help B2B SaaS founders (50-200 employees) book 15+ qualified demos per month through our done-for-you cold email system, without hiring expensive SDRs').
The golden rule, asterisked on the board: NEVER use your main domain for email outreach. Recipients marking you spam burns the domain — buy cheap outreach/burner domains (consulting-yourname.com) that 301-redirect to the main site, and leave subdomains alone (they still hang off the main domain).
Deliverability is warm-up: Instantly's pool sends emails between its own accounts to build reputation. Pre-warmed accounts are instant but shady-looking (evolveprimesignal.com); done-for-you setup warms YOUR branded domains — give it 2 weeks, then send incrementally (3/day → max 10-15 per burner inbox).
Copy is an art with an anatomy — crisp subject, researched icebreaker, problem, solution, proof, low-friction CTA — governed by rules (short, one idea per paragraph, mobile-first, no fancy formatting), demonstrated with the actual cold emails that booked the VAPI CTO's podcast and got a personal reply from Sam Altman, then packaged into an email-copy skill combining his strategy doc with skills.sh's Humanizer.

The concepts

The outreach engine: offer → lead list → Instantly → meetings

0:11:01

Before any tool, tactic, or template: what are you selling, who's on the list, and what counts as success? Everything else is plumbing.

The engine's four stages are deliberately tool-agnostic. The offer is whatever you package your specialized knowledge into — consulting, a vibe-coded product, an affiliate deal, coaching. The lead list is its mirror: not contacts, but strangers who fit the ideal customer profile, scraped or bought at scale. Between them sits the delivery tech — Instantly here, chosen because it was 'built from ground up for cold emailing' where Apollo and Attio lean CRM — and out the far end comes the only outcome that matters: meetings, because meetings are where selling happens.

The trash can is the honest fifth component. Spam placement, high bounce rates, and bad copy aren't edge cases — they're the default fate of campaigns that skip the fundamentals, which is why this session spends its whole runtime on positioning, domains, and copy BEFORE tomorrow's dashboard deep-dive. The engine framing also explains the session's economics argument: Instantly charges credits per send, so every email fired with weak copy or a cold domain is money into the trash can.

Worked example · from the session

The whiteboard build-out: offer box → lead-list box → Instantly box → 'outcome: meetings' — then the trash can appended with spam / high bounce / bad copy after a learner asked why campaigns fail.

Why it matters

Every later mechanic — burner domains, warm-up, copy anatomy — is justified by which engine stage it protects. Lose the frame and the tactics look like superstition.

People get this wrong

Cold outreach is a set-it-once machine — configure Instantly and it runs.

You can automate the sending, but copy, domains and lists degrade and need continuous tweaking; treated as one-and-done, everything routes to the trash can.

Offer + lead list, through Instantly, aimed at meetings — with the trash can catching spam, bounces and bad copy

Credits and tokens are gonna become the next currency.1:19:29

This idea elsewherebuilds onthe list becomes a machine builds onthe trash can, diagnosed

Go deeper

In one line: Four boxes on a whiteboard: an OFFER (any product, service, affiliate deal, or 1-on-1 coaching — 'if you don't have an offer, there is no outreach campaign'), a LEAD LIST (emails of strangers who fit your ICP), the tech that moves them (the Instantly dashboard), and the OUTCOME — meetings, 'what drives in revenue.' A fifth box, the trash can, collects the failure modes: spam placement, high bounce rate, bad copy.

The offer comes first and must be distilled hard: 'you have to be very, very clear on what your offer is' before any tooling (0:13:03)

A lead list is not people you know — it's scraped strangers who fit your ideal customer profile (0:13:03)

Trainer's credential: 500,000+ cold emails sent across his own campaigns, 'learned a lot with domains' (0:17:10)

It's never one-and-done: copy, domains and subdomains need continuous tweaking or the output lands in the trash can (0:19:11)

▶ Watch this taught: 0:11:01

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What are the four components, and the fifth shadow component?

Offer → lead list → tech (Instantly) → outcome (meetings), plus the trash can: spam placement, high bounces, bad copy.

Why do meetings — not replies or opens — sit in the outcome box?

Meetings are where you connect and sell; opens and replies are intermediate signals. The whole campaign is engineered backward from getting on calls.

Cold outreach ≠ email marketing (the opt-in line)

0:19:11

The difference between a newsletter and a cold email isn't the content — it's whether the recipient asked for it.

The distinction sounds pedantic and is actually load-bearing. Email marketing tools (newsletters, templates, event invites) assume an audience that subscribed — deliverability providers, spam filters, and law all treat that consent as the baseline. The outreach engine deliberately operates without it: you found Praful's email because Praful fits your ICP, and you're sending him a valid offer he never asked for. That changes everything downstream — which domains you dare send from, how many sends per day are safe, and what happens when recipients hit 'mark as spam.'

The graduation path matters too: outreach addresses can feed your marketing list, 'but not immediately' — only once they reply, engage, and opt in. Blasting a scraped list with your newsletter is how both programs die together.

Worked example · from the session

The Outskill homepage newsletter box as the live exhibit of opt-in — versus 'I don't know Praful, but Praful fits my ICP, and I'm sending him an email.'

Why it matters

Confusing the two burns domains and brands: marketing cadence to a cold list is indistinguishable from spam, to Google and to humans.

People get this wrong

Cold email is just spam with better writing.

Spam is irrelevant volume; cold outreach is a researched, valid offer to a defined ICP — the trash can is what happens when you skip the research.

This idea elsewherebuilds oncold vs warm, now by consent

Go deeper

In one line: Email marketing (newsletters, campaign emails, event invites) goes to people who typed their email into your site — they opted in and expect you. The outreach engine emails a lead list of people who've never heard of you but fit your ICP. Same inbox, opposite consent — and the same addresses can graduate from outreach to marketing only after they opt in.

Demonstrated with Outskill's own 'join our newsletter' box: subscribing IS the opt-in that makes later sends marketing, not spam (0:23:14)

Learner Prateek's summary adopted verbatim: 'people opt in to email marketing; there is no opt-in feature in outreach' (0:23:14)

Spam vs cold outreach (Slido): spam is the trash-can content; cold email is a valid offer to a researched ICP who doesn't know you yet (2:44:58)

▶ Watch this taught: 0:19:11

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What single feature separates the two?

Opt-in. Marketing goes to subscribers who asked; outreach goes to ICP-fit strangers who didn't.

When may an outreach contact enter your marketing list?

After they engage and opt in — 'only when they subscribe,' never straight from the scraped list.

Positioning: audience + outcome + mechanism + pain avoided

0:29:21

'I help businesses grow through marketing' says nothing. Fill four slots and the same sentence becomes impossible to misread.

Each slot does distinct work. Audience narrows until the reader recognizes themselves — not 'businesses' but 'B2B SaaS founders, 50-200 employees.' Outcome converts your internal goal (meetings) into their tangible win — quantified, countable, per-week. Mechanism answers the skeptic's 'how' with something concrete and preferably proprietary — a done-for-you system, a custom dashboard. Pain avoided names the alternative they're escaping — hiring expensive SDRs, wasting months testing random tools — which is often the emotional trigger that actually moves them.

The formula's power is its reusability: it becomes the email pitch, the website hero copy, the LinkedIn bio, and the marketing material — 'it's all merging: copy, marketing, sales.' And because it's mechanical, it's automatable: the session's next move is encoding it as a skill.md so any new offer can be constructed by prompt.

Worked example · from the session

The ICP test inside the strong example: 'SDRs' is jargon — if your reader is a founder who knows what an SDR costs, it lands; if your ICP wouldn't know the term, the copy (not the audience) must change.

Why it matters

This single sentence is the atom of the whole engine — it gets pasted into email copy, landing pages, and the offer skill; a weak one poisons everything downstream.

People get this wrong

A broader offer catches more fish — keep it vague so nobody's excluded.

Vague positioning means nothing to everyone. The formula deliberately excludes: the right audience recognizes itself, and everyone else was never going to buy.

Four slots, one sentence — and the weak-vs-strong contrast that shows why each slot earns its place

Cold emailing is an art form. You can try different methods, but you have to keep tweaking things.0:29:21

For your projects

Paul's course-KB pipeline fits the formula almost verbatim: audience (professionals drowning in course content) + outcome (a searchable teaching site per course) + mechanism (the extraction pipeline) + pain avoided (paying for courses you forget). Worth writing down even if it stays personal.

This idea elsewherebuilds onICP fills the audience slot reappears inniche-first positioning again

Go deeper

In one line: The positioning formula: AUDIENCE (who exactly — 'B2B SaaS companies, 10-50 employees') + OUTCOME (the tangible result they get — '3-5 qualified demos per week') + MECHANISM (how you deliver — 'our custom-built AI studio') + PAIN AVOIDED (the problem removed — 'without hiring expensive SDRs'). Weak: 'I help businesses grow through marketing.' Strong: 'I help B2B SaaS founders (50-200 employees) book 15+ qualified demos per month through our done-for-you cold email system, without needing to hire expensive SDRs.'

Audience discipline beats volume: 100k emails to pet-owners about a deep-tech product converts nobody — 'you really have to distill down whom you're selling to' (0:31:23)

Outcome must be tangible and quantified — internally the goal is meetings, but the prospect is promised a countable result ('5 AI reels per week') (0:35:26)

His live worked example: his AI-avatar video studio → founders with no time who want AI content / 5 videos per week / custom-built dashboard / saves time and pushes content (0:37:27)

Learner Lonnie's voice-agent offer assembled on the spot from the same four slots — the formula transfers (0:39:29)

Niche advice (Slido): don't target many industries with vague copy — pick the one you know deepest and send THE offer to them (2:51:04)

▶ Watch this taught: 0:29:21

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

Name the four slots and the strong example's content for each.

Audience (B2B SaaS founders, 50-200 employees), outcome (15+ qualified demos/month), mechanism (done-for-you cold email system), pain avoided (no expensive SDR hires).

Why must the outcome be quantified?

'Get on a meeting with me' is your goal, not their value — a tangible, countable result (5 reels/week, 15 demos/month) is what earns the open and the reply.

Building the offer-construction skill (live, in Codex)

how-to0:45:35

He built the skill on stage in ten minutes — not to show off Codex, but to show that a formula you've internalized once should never be hand-executed twice.

The workflow is deliberately mundane: word-vomit everything you know about offer construction into an agent — the formula, the reasoning, weak and strong examples — and demand a 'hyper-detailed' skill.md built out completely. The output file is a portable asset: attached to a ChatGPT project (or Claude project), it turns any one-line description of a service into a four-slot offer sentence. The demo produced usable pitches for two different service descriptions in under a minute each.

The deeper lesson is the pattern from Session 3 returning in a sales context: knowledge → skill file → project source → repeatable output. He runs his practice this way ('I organize my desktop using Codex itself'), and the session's final act repeats the trick for email copy.

Worked example · from the session

The full prompt structure, on screen: here's the formula → here's a weak example → here's a strong example → 'create a proper skill.md, add it on my desktop, folder offer-construction.'

Do it in this order

01Collect raw material: the formula (audience + outcome + mechanism + pain avoided), your notes, and at least one weak and one strong example
02Prompt the agent: 'fully understand this and create a skill.md based on it… hyper-detailed… build it out completely' — name the output folder (offer-construction)
03Verify the generated skill.md (his had name, structure, workflow, task steps, final-check section)
04In ChatGPT: left sidebar → Projects → + → name it (offer tool) → Sources → add the skill file
05Use it: state your service in one line and ask for the offer; expect a full formula sentence back
06Same file works in Claude Projects — 'use whatever your favorite tool is, whatever is getting the work done'

Gotchas["The skill is an OFFER creator, not a skill creator — one skill covers every future offer; don't generate one per product (0:57:46)", "skills don't have to be literally named skill.md when attached to a project — agents understand any descriptive name as long as it's markdown (2:53:07)", 'He used Codex only because it writes straight to his desktop; the artifact, not the tool, is the point (0:59:48)']

Why it matters

This is the course's compounding move: every framework you learn gets one-time encoded and infinitely reused — and the resources hand you both finished skills so you can skip straight to using them.

People get this wrong

You need a separate skill per offer or per industry.

One offer-construction skill serves every offer — the variable is your one-line service description, not the skill.

This idea elsewherebuilds onskill.md pattern, applied to offers

Go deeper

In one line: The formula gets encoded once as a reusable skill.md: dump the whiteboard notes, the formula, and weak/strong examples into a coding agent (Codex here, 'GPT 5.5 extra high'; Claude works identically), ask for a hyper-detailed skill.md saved to a desktop folder, then attach it to a ChatGPT project's sources — after which 'I do AI consulting and 1-on-1 coaching, help me create the offer' returns a formula-compliant pitch in seconds.

Live output as proof: 'I help service businesses eliminate 20+ hours of manual work per week through custom AI systems, without hiring additional staff or changing existing workflows' (0:55:42)

Second test run: founders / high-ROI AI opportunities / structured implementation roadmap / without wasting months testing random tools — all four slots present (0:57:46)

Positioning skill's real use is internal brainstorming — its output feeds the website copy, marketing material, and the email-copy skill downstream (2:46:59)

▶ Watch this taught: 0:45:35

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What goes into the prompt that generates the skill?

The complete formula, your notes/reasoning, and weak + strong examples — plus explicit instructions to build a hyper-detailed skill.md completely, with output location.

How do you deploy the skill in ChatGPT?

Create a project, add the skill file under Sources, then prompt with your service description — the project context applies the formula automatically.

The golden rule: never outreach from your main domain

1:23:31

Your domain has a credit score, and strangers you cold-email are the ones who get to wreck it.

Email providers score sending domains continuously: bounces, spam marks, engagement. A cold list — by definition people who never asked to hear from you — generates spam marks at some baseline rate no matter how good the copy is. On a disposable domain that's a cost of doing business; on your main domain it's catastrophic, because the reputation damage follows every future email your business sends, including the legitimate ones to clients and subscribers. The website keeps rendering; the domain just quietly stops reaching inboxes.

The architecture that follows: the main domain is the entity (site, official email, brand searches), outreach domains are cheap satellites (consulting-yourname.com, yourname.io) that exist only to send, and every satellite 301-redirects to the main site so a suspicious recipient who checks the domain finds a real business. Subdomains fail this design twice — they're not for outreach (they share the main domain's fate) and they're free to create anyway for their real purpose, fronting apps and dashboards.

Worked example · from the session

Coke.com → redirects to coca-cola.com — the same big-brand pattern: variant domains all funneling to one canonical home.

Why it matters

This is the one unrecoverable mistake in the whole engine. Copy can be rewritten and lists rebuilt; a burned main domain follows your brand around for months.

People get this wrong

Subdomains isolate risk — newsletter.mysite.com protects mysite.com.

A subdomain is still your main domain to reputation systems. Isolation requires a genuinely separate purchased domain.

One pristine main domain; disposable outreach domains 301-redirecting into it; subdomains reserved for apps

Never ever use your main domain for email outreach.1:27:37

For your projects

paulberg.com is the pristine entity here — if the KB or any future offer ever does outbound, the burner-satellite architecture is the day-one setup, not a later optimization.

Go deeper

In one line: Cold recipients don't owe you goodwill — some will mark you spam, and 'the moment they mark your email as spam, you just burnt your domain.' The website stays live, but the domain's sending reputation is gone: Google keeps routing it to spam, and revival is slow and painful. Hence the asterisked board rule: NEVER use your main domain (or your official email) for email outreach.

Burning is asymmetric: seconds to acquire (a few spam marks), months to recover — 'Google has become very, very strict' (1:25:35)

Your main domain is your entity: website, official email (hi@harshit.com), brand searchability — it must stay pristine (1:11:23)

Domain redirect demoed live: harshad.io typed incognito → lands on harshad.com; a learner names it correctly as a 301 redirect (1:23:31)

The rule extends to subdomains: newsletter.yourdomain.com still hangs off the main domain and hurts it — subdomains are for apps (platform.openai.com, aistudio.google.com), not outreach (1:41:48)

▶ Watch this taught: 1:23:31

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What exactly 'burns' and what survives?

Sending reputation burns — mail routes to spam indefinitely. The website, hosting, and brand stay up; they just can't reliably reach inboxes anymore.

Why do outreach domains redirect to the main site?

Legitimacy: a recipient who investigates the sender's domain lands on your real business — the burner gets credibility without the main domain taking sending risk.

Outreach domains: buying, naming, how many

1:15:26

Burner phones for email: cheap, plural, disposable — and every one of them answers to your real address.

Naming follows a recognizability rule: the variant should contain the brand (consultingharshit.org, mentor-harshit.com) so the recipient's quick sniff test passes, while remaining structurally separate so spam marks never touch the canonical domain. TLD variants of your own name (.io, .ai) work identically. The count question dissolves under the warm-up constraint: each inbox safely sends only 10-15/day once warmed, so required volume ÷ per-inbox ceiling ≈ how many domains and inboxes you actually need — start with 3 and grow.

Purchase channel is a genuine choice: Namecheap/GoDaddy/Porkbun for control and price, or Instantly's in-dashboard purchase where the redirect and DNS are handled for you at ~$15/domain — the convenience tax that tomorrow's hands-on session will exercise.

Worked example · from the session

His live search for consultant-variants: consultingharshit.org, meetharshit.com, harshit-hq.com — Instantly generating brandable options on the spot.

Why it matters

This is the capacity planning of cold email: domains × inboxes × daily ceiling = your true send capacity, and it's bought for the price of a few coffees.

People get this wrong

There's an optimal number of outreach domains to buy up front.

Capacity is goal-driven: start with ~3, warm them, and scale domains and inboxes as sending volume requires — the ceiling per inbox (10-15/day) does the math for you.

Go deeper

In one line: Buy cheap variant domains solely for sending: consulting-yourname.com, mentor-yourname.com, with-company.com, or TLD variants (.io, .xyz, .ai). Sources: Namecheap (his default), GoDaddy, Porkbun — or directly inside Instantly (~$15/domain, redirect configured automatically). How many? 'There is no number' — start with 3, warm them up, scale with your goals; 10 domains running 10 daily campaigns is fine.

Domain buying is an online order: search, add to cart, check out — his dashboard shows the portfolio he owns, with harshad.ai next on the wishlist (1:17:27)

Both domains get bought in the pair: the main one you never touch, plus the outreach variants that redirect to it (1:31:39)

Redirects are configurable on Vercel or via DNS — 'honestly, I use Claude to help me out with it' — but Instantly does it automatically for domains bought there (1:29:38)

Main-website TLD advice (Slido): .com first; .ai/.io fine (though .ai is getting expensive); avoid .xyz for your MAIN domain (2:46:59)

Burned-domain detection: campaigns from that domain stop performing, and a test email to yourself lands in spam (1:39:47)

▶ Watch this taught: 1:15:26

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What makes a good outreach domain name?

Brand-recognizable but separately registered: consulting-/mentor-/with- prefixes or TLD variants — passes the recipient's sniff test, isolates the reputation risk.

How do you discover a domain has burned?

Performance collapse in that domain's campaigns plus the self-test: email yourself and watch it land in spam.

Why not just Gmail: the credibility test

1:35:43

Your prospect's first move isn't reading your pitch — it's deciding in half a second whether the sender is real.

The recipient of a cold email runs an instant legitimacy check: who is this? A domain email answers it — the domain resolves to a website, the website names a person, the person has a LinkedIn. A personal Gmail answers nothing, and in 2026's phishing climate, unverifiable equals fake. This is why the burner-domain architecture includes the redirect: even the disposable senders resolve to a real business when checked.

Note the interplay with the golden rule: your official domain email (hi@yourdomain.com) is too legitimate to risk on outreach, and personal Gmail is too illegitimate to convert. Outreach domains thread the needle — branded enough to validate, separate enough to burn.

Worked example · from the session

His own childhood Gmail on screen — birthday-number suffix and all — as the sender no founder would trust with a dashboard update.

Why it matters

Sender identity is the first line of copy the prospect reads. Get it wrong and the subject line never gets its chance.

People get this wrong

Gmail is fine to start — upgrade to a domain when the campaign proves itself.

The campaign can't prove itself from Gmail: credibility failure suppresses the very response rates you'd measure. Domain first, then test.

Go deeper

In one line: Sending your offer from the personal Gmail you made in 6th grade (harshad28octopus@gmail.com) fails the recipient's validation test: there's nothing to look up, it reads as phishing, and 'you lose all my credibility.' A real domain legitimizes: type harshit.com and the website, LinkedIn, and business appear. It's all branding — the sender address is part of the copy.

Learner Victor's framing adopted: from a random Gmail, a genuine offer is indistinguishable from a phishing attack (1:37:44)

The validation asymmetry: a domain email can be verified in one search; a personal Gmail can't be verified at all (1:39:47)

Even the trainer's real, findable Gmail fails the test — the point isn't obscurity, it's the absence of a business entity behind it (1:39:47)

▶ Watch this taught: 1:35:43

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What specific test does a personal Gmail fail?

Validation-by-search: there's no domain to type in, no site, no entity — so the recipient can't distinguish you from phishing.

Why not use the official domain email instead, since it validates perfectly?

It validates AND it's precious — spam marks on it burn the main domain. Outreach domains are the middle path: verifiable but disposable.

Email warm-up: the pool, the fire icon, the two weeks

how-to1:45:52

A brand-new domain emailing 100 strangers on day one looks exactly like what it is to Google — so Instantly spends two weeks manufacturing a respectable past for it.

Spam filters trust senders with history: consistent volume, real engagement, low complaint rates. A fresh domain has none, so warm-up fabricates it benignly — Instantly's pool accounts send each other mail, open it, reply, rescue it from spam folders. Two weeks of that and the domain carries a track record into its first real campaign. The fire icon starts it; the graph and health score let you watch reputation accrue.

The incremental ramp is the same logic continued: even a warmed domain that jumps from 5 to 100 daily sends trips volume anomaly detection. The 3 → 10-15/day-per-inbox ramp keeps the sending curve looking organic, and makes capacity a simple multiplication — which is why the answer to 'how do I email my 2,500 scraped leads?' is more warmed inboxes, never a bigger daily blast."

Worked example · from the session

The launch-date math done in-session: want campaigns live July 1? Domains bought, connected, and fire-icon'd by June 15 — the two weeks are non-negotiable lead time.

Do it in this order

01Add your email account under Instantly → Email accounts (+)
02Click the fire icon on the account row to activate warm-up
03Leave settings at defaults initially (~1 warm-up send/day scaling up automatically)
04Wait 2 weeks — watch the warm-up graph and health score build
05Begin campaigns incrementally: ~3/day → 4-5/day → cap at 10-15/day per burner inbox
06Schedule backward: campaign launching July 1 means domains bought and warming by mid-June

Gotchas["Warm-up isn't optional for self-bought domains — a fresh domain sending 100 emails on day one is a spam-folder guarantee (2:40:56)", 'The 10-15/day ceiling is per inbox, not per campaign — scale volume by adding domains/inboxes, not by pushing one inbox harder (2:40:56)', '2,500 scraped leads ≠ 2,500 sends this week: import the CSV, then let incremental daily limits pace the burn (2:44:58)']

Why it matters

Warm-up is the invisible variable behind every deliverability statistic — the cheapest campaign improvement is patience, and the most common failure is skipping it.

People get this wrong

Warm-up is a setup step — done once, then send at will.

Warm-up buys you a STARTING reputation; the incremental ramp and per-inbox ceilings protect it permanently. Reputation is a continuously spent and earned budget.

Accounts in the pool exchanging mail to build reputation — then incremental real sends under the per-inbox ceiling

Go deeper

In one line: Warm-up is Instantly sending emails between accounts in its own pool to build sender reputation before real campaigns. Connect your inbox, click the fire icon, and give it TWO WEEKS — the dashboard tracks warm-up sends and a health score. Then send incrementally: start ~3 real emails/day per inbox, climb to a ceiling of 10-15/day. Rushing either phase is how fresh domains die.

The pool mechanic: Instantly's accounts email each other and engage, teaching Google's filters the sender is legitimate (1:53:59)

The health score and warm-up graph on each account row are the observability layer — reputation made visible (2:10:10)

Warm-up applies to Instantly-bought and self-bought domains alike; 'there's nothing much to learn — it's automatically done,' the discipline is the waiting (2:12:11)

▶ Watch this taught: 1:45:52

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

Mechanically, what happens during warm-up?

Instantly sends and engages with mail between its own pooled accounts — including yours — building the volume-and-engagement history that spam filters interpret as legitimacy.

You have 2,500 leads and 3 warmed inboxes. What's the send plan?

Import the CSV, then respect the ramp: ~3/day/inbox rising to 10-15/day/inbox — roughly 30-45/day at steady state; add inboxes if that's too slow. Never blast.

Pre-warmed accounts vs done-for-you: the branding trade

1:47:54

Instantly will rent you a trusted-but-ugly identity today, or spend two weeks making YOUR identity trusted — and which one you pick is a branding decision, not a technical one.

Pre-warmed accounts are reputation arbitrage: Instantly bought thousands of domains, aged them in the warm-up pool, and sells the accumulated trust. Deliverability is genuinely excellent — that's the product — but the sender reads as evolveprimesignal.com, which costs you exactly the credibility the Gmail lesson said the sender line must carry. Done-for-you inverts the trade: your brandable domain, Instantly's plumbing (DNS, accounts, pool enrollment), and a two-week wait while reputation builds.

The decision rule that emerged: mass-volume campaigns where the sender is disposable → pre-warmed; anything where the brand IS the offer (consulting, a studio, a product with a name) → done-for-you. Either way the reply flow converges: the prospect's response forwards to your official inbox, and from the first genuine reply onward you correspond as your real, pristine self.

Worked example · from the session

The live vote: 'do you guys like the pre-warmed from Instantly?' — a wall of no's, and Sayed's counter-example sender syed@consultingsyedhassan.org shown as what recipients should see instead.

Why it matters

This is tomorrow's first real decision in the dashboard — and the reply-forwarding flow is the detail that makes burner-based outreach compatible with an honest sales conversation.

People get this wrong

Shady-looking pre-warmed domains must deliver worse — they look like spam.

They deliver BETTER than your fresh domain — reputation is machine-read, not aesthetic. The cost is human trust: what the prospect sees, not what the filter sees.

Go deeper

In one line: Instantly sells two on-ramps. Pre-warmed accounts: instantly usable, high domain reputation, guaranteed inbox placement — but the domains are gibberish (evolveprimesignal.com, 'diana@' addresses; ~$15/domain plus Google provider fees) and 'they look shady.' Done-for-you setup: pick brandable domains (consultingyou.org, meet-you.com), Instantly configures accounts and adds them to the warm-up pool — your branding intact, two weeks slower. The cohort voted decisively for the second.

Why pre-warmed works at all: those domains have been exchanging pool mail for ages under Google as provider — no spam placement, high open rates, campaigns in minutes (1:53:59)

The forwarding-domain field is what redeems shady senders: replies and clicks resolve to YOUR real site (1:49:56)

The reply flow: burner sends → prospect sees the burner address → their reply forwards to your official email → you respond from the official account with the meeting link — 'you're getting closer to your client' (1:58:01)

Cohort verdict on pre-warmed, near-unanimous no: 'I would not wanna dilute my brand just to send out emails' — his own position too, unless you're doing pure mass campaigns (2:02:05)

Done-for-you niceties: sender name configurable (a live glitch showed 'tools.outskill' — fix it before sending), ~5 inboxes max recommended per domain, Google as provider over the untested AirMail (2:04:06)

▶ Watch this taught: 1:47:54

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What are you actually buying with a pre-warmed account?

Aged sending reputation: domains and inboxes that have exchanged pool mail long enough that Google treats them as legitimate — instant deliverability, zero brand equity.

Trace a reply end-to-end in the recommended setup.

Burner inbox sends → prospect replies to the burner → forwarding delivers it to your official email → you answer from the official address with the meeting link.

Anatomy of a cold email (and the billionaire texting style)

2:22:27

The email that got Sam Altman to personally reply was three sentences and read like a WhatsApp message — that's not an accident, it's the whole doctrine.

Each organ has a job. The subject earns the open by being crisp and unexaggerated (and by dodging spam-trigger vocabulary). The icebreaker earns the read by proving you did research — a genuinely common ground, not fake flattery, because recipients pattern-match insincerity instantly. Problem and solution are the offer formula compressed; proof (results, names, numbers) converts claim into credibility; and the CTA deliberately lowers the commitment bar — a conversation-worthiness question, not a calendar demand.

The style layer is the surprise: his samples to actual billionaires read like text messages, because busy people write and read that way. Hence the rules of engagement — short, one idea per paragraph, mobile-first, formatting-free — and the observation that overwrought AI-sounding prose (em-dashes and all) is now itself a spam signal to human readers.

Worked example · from the session

The Sam Altman email on screen: 'Hey Sam — Daniel and Harshit here, massive fans of what you're doing at OpenAI…' — a want-you-on-our-podcast ask in plain human register, and the polite personal decline that proves it landed in primary and got read.

Why it matters

Copy is the decision-and-action stage of the whole engine — warmed domains only buy the inbox slot; these six organs and four rules are what convert it.

People get this wrong

Professional means formal — proper salutations, polished paragraphs, rich formatting.

The highest-status inboxes run on WhatsApp register: short, plain, human. Formality reads as mass-mail; brevity reads as a real person.

Subject → icebreaker → problem → solution → proof → low-friction CTA, under the four rules of engagement

The more I email these billionaires — these people playing at a very different level — they write literally like how we text on WhatsApp.2:20:23

This idea elsewherebuilds onicebreakers, hand-written this time

Go deeper

In one line: Six organs: crisp SUBJECT (no exaggeration) → ICEBREAKER (common ground proving research: 'I'm a mentor at Outskill and I use the VAPI CLI a lot') → PROBLEM (when selling) → SOLUTION you offer → PROOF it works → low-friction CTA. Rules of engagement: keep it short (they owe you nothing), one idea per paragraph, mobile-first, no fancy formatting.

Receipts: his cold DM booked the VAPI CTO for his podcast; his drafted email to Sam Altman got a personal 'really jammed, but thanks for the invite' reply (2:18:20, 2:20:23)

CTA psychology: prefer low-commitment asks — 'is this worth a conversation?' beats 'book a time on my calendar' (2:32:46)

Spam-trigger words in subjects ('free demo inside') get machine-filtered before a human ever sees them (2:32:46)

Icebreaker sincerity rule: no fake flattery — 'we all know which one's written by AI'; em-dashes flagged as the tell (2:32:46)

Mobile-first because prospects triage email on phones: short lines, no formatting that breaks in Gmail or iCloud (2:24:32)

▶ Watch this taught: 2:22:27

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

Name the six organs in order.

Subject → icebreaker → problem → solution → proof → CTA (low-friction).

What makes a CTA low-friction, and why prefer it?

It asks for a signal, not a commitment — 'worth a conversation?' vs 'book my calendar.' Lower activation energy, higher reply rate, and the meeting ask comes naturally next.

The email-copy skill: strategy doc + Humanizer, plus sequences

how-to2:26:35

His copy secret isn't a template — it's a pipeline: a public humanizer skill, welded to a private strategy doc, emitting emails that don't smell like AI.

The build repeats the offer-skill pattern with one upgrade: composition. The Humanizer skill (public, battle-tested, 24k stars on skills.sh) handles register — making prose read like the WhatsApp-style human writing the anatomy demands. His strategy doc contributes the private judgment: sequence timing, CTA psychology, subject-line taboos, icebreaker sincerity rules. The agent merges them into one skill whose output is pre-structured (name, icebreaker, problem, solution, CTA, sign-off) and pre-humanized.

The sequence framework is the operational half: tomorrow's Instantly session wires email 1, the nudge, and the breakup note into an automated sequence with day-gaps — so all three get drafted now, by the skill, in one sitting. The breakup email isn't defeatism; it's the highest-reply email in most sequences because it's the easiest to answer.

Worked example · from the session

The live merge prompt: 'here is the Humanizer skill — combine it with the above pointers and my own consulting email strategy doc — build the email-copy skill, folder email-copy-skill' — then the skill's own subject-line dos/don'ts scrolling on screen.

Do it in this order

01Fetch the Humanizer skill from skills.sh (the highly-starred one — a learner linked the exact skill in chat)
02Paste it into your agent along with the anatomy pointers AND your own strategy doc: 'combine it with the above pointers and my consulting email strategy'
03Ask for an email-copy skill saved to its own folder (email-copy-skill)
04Attach to a fresh ChatGPT/Claude project — separate from the offer project
05Feed it your constructed offer; expect subject options + a structured, humanized body
06Reuse the skill for the follow-up sequence emails, not just email #1

Gotchas["Sequences are where campaigns actually convert: email 1 full framework, email 2 a short nudge after ~2 days of silence, email 3 value-or-breakup ('assuming this isn't a priority right now…') — build all three up front (2:30:44)", 'If a prospect REPLIES, the sequence stops — you respond personally from the official email; never let the automation keep nudging a live conversation (2:53:07)', 'The same skill works for LinkedIn outreach — the anatomy transfers across channels (2:18:20)']

Why it matters

This is the session's compounding lesson applied twice in one night: frameworks become skills, skills compose with other skills, and the resources folder hands you the finished artifacts.

People get this wrong

Follow-ups are the first email resent with 'just bumping this.'

Each sequence step has a distinct job — full pitch, light nudge, value-or-goodbye — and the breakup note often outperforms email #1 on replies.

This idea elsewherereappears incomposing skills from skills

Go deeper

In one line: Second live skill build: take the Humanizer skill from skills.sh (24k+ stars), combine it with his personal cold-email strategy doc and the session's anatomy notes, and have the agent produce an email-copy skill. Output structure: first name → icebreaker → problem → solution → low-friction CTA → sign-off, plus subject-line rules, icebreaker do/don't lists, CTA psychology, and his three-email sequence framework: full pitch → short nudge (2 days later) → value/breakup note.

Skill composition demonstrated: an internet skill (Humanizer) + private strategy + session notes = a personal tool neither source could be alone (2:28:38)

The generated skill included his breakup-email templates verbatim: 'haven't heard back, so I'll assume this is not a priority' and the free-guide value send (2:34:49)

Both finished skills (offer-construction + email-copy) shared in chat and post-session resources — usable without rebuilding (2:34:49)

▶ Watch this taught: 2:26:35

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

What three inputs made the email-copy skill?

The public Humanizer skill (skills.sh), the session's anatomy/rules pointers, and his private consulting email strategy doc.

Describe the three-email sequence and its stop condition.

Full framework → short nudge (~2 days) → value/breakup note. Any reply stops the sequence — the conversation moves to your official email, personally.

The deliverability toolbox: DNS, EasyDMARC, F5Bot, alternatives

2:38:51

The Q&A produced a toolbox worth the session on its own: a reputation checker, a free intent-signal bot, and two escape hatches from Instantly's pricing.

Three tools earn permanent bookmarks. EasyDMARC (and its free-tool siblings) turns the 'is my domain burned?' question from superstition into a report. F5Bot watches Reddit and Hacker News for your chosen keywords — company names, competitor names, problem statements — and emails you each mention: free, always-on intent signals that slot directly into the signal-reading discipline from Sessions 5-6. And for the cost-sensitive, the Instantly alternatives ladder: Apollo.io as the commercial rival, Listmonk as the self-hosted newsletter engine he 'hacked with Claude Code' into a cold-email sender, and Twenty as the self-hosted CRM behind his own pipeline.

The DNS demystification matters for tomorrow: 'advanced DNS' in your registrar is just the records panel — where subdomains get created and where Instantly's done-for-you setup writes the entries that make your outreach domains deliverable."

Worked example · from the session

F5Bot configured with your product's problem statement — every founder complaining about it on Reddit becomes a hot, timestamped lead in your inbox.

Why it matters

These are the free/cheap 20% that covers 80% of deliverability hygiene and lead intelligence — and the self-hosted options matter once credit costs bite.

People get this wrong

Deliverability tooling is only for big senders — small campaigns can skip the hygiene.

Small senders have the LEAST reputation margin — one bad blast from a fresh domain is fatal, so the free checkers and warm-up discipline matter most at small scale.

This idea elsewherereappears inintent signals, a free tool

Go deeper

In one line: Slido-driven grab bag with real tools: DNS is just the domain-name system (advanced-DNS panel is where subdomains and records live); EasyDMARC checks a domain's sending reputation; F5Bot emails you whenever your company or problem-statement is mentioned on Reddit/Hacker News (free intent signals); alternatives to Instantly include Apollo.io and self-hosted Listmonk ('I hacked it with Claude Code for cold email'); his CRM is self-hosted Twenty.

Domain reputation check: EasyDMARC (email-gated report) and similar free tools answer 'is my domain burned?' with data instead of vibes (2:46:59)

F5Bot as poor-man's intent engine: mention-alerts for your product or problem space, feeding exactly the signal-reading Cameron and Akhil taught (2:55:10)

Study workflow endorsed: dump each session's transcript + resources into NotebookLM, generate the interactive deep-conversation audio, and 'have a podcast with me' (2:49:02)

AI-overwhelm answer, candid: anxiety is real (he cited the Fable cutoff), the hedge is skills + open source + 'community first — no matter how good AI gets, we come back to human connection' (2:51:04)

Provider notes: Google as inbox provider over GoDaddy's Microsoft 365 bundling; .com.au-style ccTLDs likely need the done-for-you route, not pre-warmed (2:53:07)

▶ Watch this taught: 2:38:51

Check yourself

Answer from memory first — the recall attempt is what makes it stick. Then reveal.

Your open rates cratered. Which tool answers whether the domain burned?

EasyDMARC (or similar free reputation checkers) — plus the self-test of emailing your own inbox and checking placement.

What does F5Bot give you, and which earlier discipline does it feed?

Email alerts for Reddit/HN mentions of your keywords — free intent signals feeding the signal-reading → context → narrative drill from Sessions 5-6.

Every concept, three clicks deep

The same concepts as a quick reference: the closed row is the glance, open is the study card, and every timestamp jumps into the recording.

01The outreach engine: offer → lead list → Instantly → meetingsFour boxes on a whiteboard: an OFFER (any product, service, affiliate deal, or 1-on-1 coaching — 'if you do…0:11:01›

Four boxes on a whiteboard: an OFFER (any product, service, affiliate deal, or 1-on-1 coaching — 'if you don't have an offer, there is no outreach campaign'), a LEAD LIST (emails of strangers who fit your ICP), the tech that moves them (the Instantly dashboard), and the OUTCOME — meetings, 'what drives in revenue.' A fifth box, the trash can, collects the failure modes: spam placement, high bounce rate, bad copy.

The offer comes first and must be distilled hard: 'you have to be very, very clear on what your offer is' before any tooling (0:13:03)

A lead list is not people you know — it's scraped strangers who fit your ideal customer profile (0:13:03)

Trainer's credential: 500,000+ cold emails sent across his own campaigns, 'learned a lot with domains' (0:17:10)

It's never one-and-done: copy, domains and subdomains need continuous tweaking or the output lands in the trash can (0:19:11)

02Cold outreach ≠ email marketing (the opt-in line)Email marketing (newsletters, campaign emails, event invites) goes to people who typed their email into you…0:19:11›

Email marketing (newsletters, campaign emails, event invites) goes to people who typed their email into your site — they opted in and expect you. The outreach engine emails a lead list of people who've never heard of you but fit your ICP. Same inbox, opposite consent — and the same addresses can graduate from outreach to marketing only after they opt in.

Demonstrated with Outskill's own 'join our newsletter' box: subscribing IS the opt-in that makes later sends marketing, not spam (0:23:14)

Learner Prateek's summary adopted verbatim: 'people opt in to email marketing; there is no opt-in feature in outreach' (0:23:14)

Spam vs cold outreach (Slido): spam is the trash-can content; cold email is a valid offer to a researched ICP who doesn't know you yet (2:44:58)

03Positioning: audience + outcome + mechanism + pain avoidedThe positioning formula: AUDIENCE (who exactly — 'B2B SaaS companies, 10-50 employees') + OUTCOME (the tang…0:29:21›

The positioning formula: AUDIENCE (who exactly — 'B2B SaaS companies, 10-50 employees') + OUTCOME (the tangible result they get — '3-5 qualified demos per week') + MECHANISM (how you deliver — 'our custom-built AI studio') + PAIN AVOIDED (the problem removed — 'without hiring expensive SDRs'). Weak: 'I help businesses grow through marketing.' Strong: 'I help B2B SaaS founders (50-200 employees) book 15+ qualified demos per month through our done-for-you cold email system, without needing to hire expensive SDRs.'

Audience discipline beats volume: 100k emails to pet-owners about a deep-tech product converts nobody — 'you really have to distill down whom you're selling to' (0:31:23)

Outcome must be tangible and quantified — internally the goal is meetings, but the prospect is promised a countable result ('5 AI reels per week') (0:35:26)

His live worked example: his AI-avatar video studio → founders with no time who want AI content / 5 videos per week / custom-built dashboard / saves time and pushes content (0:37:27)

Learner Lonnie's voice-agent offer assembled on the spot from the same four slots — the formula transfers (0:39:29)

Niche advice (Slido): don't target many industries with vague copy — pick the one you know deepest and send THE offer to them (2:51:04)

04Building the offer-construction skill (live, in Codex)The formula gets encoded once as a reusable skill.md: dump the whiteboard notes, the formula, and weak/stro…0:45:35›

The formula gets encoded once as a reusable skill.md: dump the whiteboard notes, the formula, and weak/strong examples into a coding agent (Codex here, 'GPT 5.5 extra high'; Claude works identically), ask for a hyper-detailed skill.md saved to a desktop folder, then attach it to a ChatGPT project's sources — after which 'I do AI consulting and 1-on-1 coaching, help me create the offer' returns a formula-compliant pitch in seconds.

Live output as proof: 'I help service businesses eliminate 20+ hours of manual work per week through custom AI systems, without hiring additional staff or changing existing workflows' (0:55:42)

Second test run: founders / high-ROI AI opportunities / structured implementation roadmap / without wasting months testing random tools — all four slots present (0:57:46)

Positioning skill's real use is internal brainstorming — its output feeds the website copy, marketing material, and the email-copy skill downstream (2:46:59)

05The golden rule: never outreach from your main domainCold recipients don't owe you goodwill — some will mark you spam, and 'the moment they mark your email as s…1:23:31›

Cold recipients don't owe you goodwill — some will mark you spam, and 'the moment they mark your email as spam, you just burnt your domain.' The website stays live, but the domain's sending reputation is gone: Google keeps routing it to spam, and revival is slow and painful. Hence the asterisked board rule: NEVER use your main domain (or your official email) for email outreach.

Burning is asymmetric: seconds to acquire (a few spam marks), months to recover — 'Google has become very, very strict' (1:25:35)

Your main domain is your entity: website, official email (hi@harshit.com), brand searchability — it must stay pristine (1:11:23)

Domain redirect demoed live: harshad.io typed incognito → lands on harshad.com; a learner names it correctly as a 301 redirect (1:23:31)

The rule extends to subdomains: newsletter.yourdomain.com still hangs off the main domain and hurts it — subdomains are for apps (platform.openai.com, aistudio.google.com), not outreach (1:41:48)

06Outreach domains: buying, naming, how manyBuy cheap variant domains solely for sending: consulting-yourname.com, mentor-yourname.com, with-company.co…1:15:26›

Buy cheap variant domains solely for sending: consulting-yourname.com, mentor-yourname.com, with-company.com, or TLD variants (.io, .xyz, .ai). Sources: Namecheap (his default), GoDaddy, Porkbun — or directly inside Instantly (~$15/domain, redirect configured automatically). How many? 'There is no number' — start with 3, warm them up, scale with your goals; 10 domains running 10 daily campaigns is fine.

Domain buying is an online order: search, add to cart, check out — his dashboard shows the portfolio he owns, with harshad.ai next on the wishlist (1:17:27)

Both domains get bought in the pair: the main one you never touch, plus the outreach variants that redirect to it (1:31:39)

Redirects are configurable on Vercel or via DNS — 'honestly, I use Claude to help me out with it' — but Instantly does it automatically for domains bought there (1:29:38)

Main-website TLD advice (Slido): .com first; .ai/.io fine (though .ai is getting expensive); avoid .xyz for your MAIN domain (2:46:59)

Burned-domain detection: campaigns from that domain stop performing, and a test email to yourself lands in spam (1:39:47)

07Why not just Gmail: the credibility testSending your offer from the personal Gmail you made in 6th grade (harshad28octopus@gmail.com) fails the rec…1:35:43›

Sending your offer from the personal Gmail you made in 6th grade (harshad28octopus@gmail.com) fails the recipient's validation test: there's nothing to look up, it reads as phishing, and 'you lose all my credibility.' A real domain legitimizes: type harshit.com and the website, LinkedIn, and business appear. It's all branding — the sender address is part of the copy.

Learner Victor's framing adopted: from a random Gmail, a genuine offer is indistinguishable from a phishing attack (1:37:44)

The validation asymmetry: a domain email can be verified in one search; a personal Gmail can't be verified at all (1:39:47)

Even the trainer's real, findable Gmail fails the test — the point isn't obscurity, it's the absence of a business entity behind it (1:39:47)

08Email warm-up: the pool, the fire icon, the two weeksWarm-up is Instantly sending emails between accounts in its own pool to build sender reputation before real…1:45:52›

Warm-up is Instantly sending emails between accounts in its own pool to build sender reputation before real campaigns. Connect your inbox, click the fire icon, and give it TWO WEEKS — the dashboard tracks warm-up sends and a health score. Then send incrementally: start ~3 real emails/day per inbox, climb to a ceiling of 10-15/day. Rushing either phase is how fresh domains die.

The pool mechanic: Instantly's accounts email each other and engage, teaching Google's filters the sender is legitimate (1:53:59)

The health score and warm-up graph on each account row are the observability layer — reputation made visible (2:10:10)

Warm-up applies to Instantly-bought and self-bought domains alike; 'there's nothing much to learn — it's automatically done,' the discipline is the waiting (2:12:11)

09Pre-warmed accounts vs done-for-you: the branding tradeInstantly sells two on-ramps.1:47:54›

Instantly sells two on-ramps. Pre-warmed accounts: instantly usable, high domain reputation, guaranteed inbox placement — but the domains are gibberish (evolveprimesignal.com, 'diana@' addresses; ~$15/domain plus Google provider fees) and 'they look shady.' Done-for-you setup: pick brandable domains (consultingyou.org, meet-you.com), Instantly configures accounts and adds them to the warm-up pool — your branding intact, two weeks slower. The cohort voted decisively for the second.

Why pre-warmed works at all: those domains have been exchanging pool mail for ages under Google as provider — no spam placement, high open rates, campaigns in minutes (1:53:59)

The forwarding-domain field is what redeems shady senders: replies and clicks resolve to YOUR real site (1:49:56)

Cohort verdict on pre-warmed, near-unanimous no: 'I would not wanna dilute my brand just to send out emails' — his own position too, unless you're doing pure mass campaigns (2:02:05)

10Anatomy of a cold email (and the billionaire texting style)Six organs: crisp SUBJECT (no exaggeration) → ICEBREAKER (common ground proving research: 'I'm a mentor at…2:22:27›

Six organs: crisp SUBJECT (no exaggeration) → ICEBREAKER (common ground proving research: 'I'm a mentor at Outskill and I use the VAPI CLI a lot') → PROBLEM (when selling) → SOLUTION you offer → PROOF it works → low-friction CTA. Rules of engagement: keep it short (they owe you nothing), one idea per paragraph, mobile-first, no fancy formatting.

Receipts: his cold DM booked the VAPI CTO for his podcast; his drafted email to Sam Altman got a personal 'really jammed, but thanks for the invite' reply (2:18:20, 2:20:23)

CTA psychology: prefer low-commitment asks — 'is this worth a conversation?' beats 'book a time on my calendar' (2:32:46)

Spam-trigger words in subjects ('free demo inside') get machine-filtered before a human ever sees them (2:32:46)

Icebreaker sincerity rule: no fake flattery — 'we all know which one's written by AI'; em-dashes flagged as the tell (2:32:46)

Mobile-first because prospects triage email on phones: short lines, no formatting that breaks in Gmail or iCloud (2:24:32)

11The email-copy skill: strategy doc + Humanizer, plus sequencesSecond live skill build: take the Humanizer skill from skills.sh (24k+ stars), combine it with his personal…2:26:35›

Second live skill build: take the Humanizer skill from skills.sh (24k+ stars), combine it with his personal cold-email strategy doc and the session's anatomy notes, and have the agent produce an email-copy skill. Output structure: first name → icebreaker → problem → solution → low-friction CTA → sign-off, plus subject-line rules, icebreaker do/don't lists, CTA psychology, and his three-email sequence framework: full pitch → short nudge (2 days later) → value/breakup note.

Skill composition demonstrated: an internet skill (Humanizer) + private strategy + session notes = a personal tool neither source could be alone (2:28:38)

The generated skill included his breakup-email templates verbatim: 'haven't heard back, so I'll assume this is not a priority' and the free-guide value send (2:34:49)

Both finished skills (offer-construction + email-copy) shared in chat and post-session resources — usable without rebuilding (2:34:49)

12The deliverability toolbox: DNS, EasyDMARC, F5Bot, alternativesSlido-driven grab bag with real tools: DNS is just the domain-name system (advanced-DNS panel is where subd…2:38:51›

Slido-driven grab bag with real tools: DNS is just the domain-name system (advanced-DNS panel is where subdomains and records live); EasyDMARC checks a domain's sending reputation; F5Bot emails you whenever your company or problem-statement is mentioned on Reddit/Hacker News (free intent signals); alternatives to Instantly include Apollo.io and self-hosted Listmonk ('I hacked it with Claude Code for cold email'); his CRM is self-hosted Twenty.

Domain reputation check: EasyDMARC (email-gated report) and similar free tools answer 'is my domain burned?' with data instead of vibes (2:46:59)

F5Bot as poor-man's intent engine: mention-alerts for your product or problem space, feeding exactly the signal-reading Cameron and Akhil taught (2:55:10)

Study workflow endorsed: dump each session's transcript + resources into NotebookLM, generate the interactive deep-conversation audio, and 'have a podcast with me' (2:49:02)

Provider notes: Google as inbox provider over GoDaddy's Microsoft 365 bundling; .com.au-style ccTLDs likely need the done-for-you route, not pre-warmed (2:53:07)

Tools referenced

Tool	Coverage	Moment	Context
Instantly	demonstrated	0:25:16	The session's platform — dashboard tour, pre-warmed accounts, done-for-you setup, warm-up pool, fire icon, lead lists; deep-dive scheduled for the following session
Codex	demonstrated	0:45:35	Built both skills live ('GPT 5.5 extra high'); chosen because it writes files straight to his desktop — also organizes his desktop weekly
ChatGPT	demonstrated	0:55:42	Projects + Sources as the skill runtime: attach skill.md, prompt with your service, get the offer
Claude	demonstrated	0:59:48	Same skill-creation flow shown in Claude for learners who asked; also his helper for DNS redirects and the Listmonk cold-email hack
skills.sh	demonstrated	2:26:35	Skill marketplace (698k+ skills); source of the Humanizer skill (24k+ stars) composed into the email-copy skill
Namecheap	demonstrated	1:15:26	His domain broker of choice; portfolio dashboard shown; GoDaddy and Porkbun named as alternatives
EasyDMARC	demonstrated	2:46:59	Domain-reputation checker for the 'is my domain burned?' question (email-gated report)
F5Bot	demonstrated	2:55:10	Free Reddit/Hacker-News mention alerts — his intent-signal source
Slido	demonstrated	2:36:50	20-minute closing Q&A (his first time driving it)
Apollo.io	mentioned	2:42:57	Commercial Instantly alternative; also carries lead lists; earlier dismissed as more CRM-shaped than cold-email-native
Listmonk	mentioned	2:42:57	Self-hosted newsletter manager he 'hacked with Claude Code' into a cold-email sender — the open-source escape hatch
Twenty CRM	mentioned	2:14:14	Self-hosted CRM behind his consulting pipeline (crm.harshad.vc)
NotebookLM	mentioned	2:49:02	Recommended study loop: transcript + resources in, interactive deep-conversation audio out — 'have a podcast with me'
VAPI	mentioned	2:18:20	The cold-DM success story: his email-copy skill booked VAPI's CTO for his podcast
Vercel	mentioned	1:29:38	Where he configures domain redirects (with Claude's help)

Session materials

Archived locally on V: — click to open. Companion pages link to the LMS.

vtt

GMT20260613-135736_Recording.transcript.vtt

224 KB

Action items

Fill in the offer cheat-sheet: run YOUR service through audience + outcome + mechanism + pain avoided before the next session (0:43:32)
Create the two projects (offer tool, email copy) in ChatGPT or Claude and attach the shared skills — test each with your real service (0:55:42)
Buy 1-3 outreach domains (brand-variant names) and set their redirects to your main site — never plan outreach from the main domain (1:31:39)
Sign up for Instantly, connect the domains, and hit the fire icon NOW — the two-week warm-up clock gates your first campaign (2:10:10)
Draft your three-email sequence (full pitch / nudge / breakup) with the email-copy skill, ready to wire into tomorrow's campaign build (2:30:44)
Set up F5Bot alerts for your product's problem statement and run EasyDMARC on any domain you plan to send from (2:55:10)

Resources mentioned

Resources

docOffer-construction skill (skill.md) — built live, shared in chat and post-session feeds 0:53:41
docEmail-copy skill (skill.md) — Humanizer + his strategy doc, shared in chat and post-session feeds 2:32:46
docOffer cheat-sheet exercise (strong/weak worksheet) — complete before next session 0:43:32
docThe session Excalidraw board + his cold-email strategy document — shared post-session 1:09:22
docHumanizer skill on skills.sh (the 24k-star one, linked in chat by learner Syed) 2:26:35

Extraction notes

This page was built from an auto-generated transcript, which garbles product and people's names. Those were corrected silently in everything above and logged here for transparency. The warnings flag claims that were true on the recording day but change fast.

Transcript corrections applied

The transcript says	The trainer actually means
Harshad / Arshad / Harshet / Harshed / hardship dot com / Hirsat	Harshit (trainer; registry spelling) and harshad.com-family domains as heard
ChargeGPT / Chargegbt / charge g p t	ChatGPT
Claw / cloud / plot / Claude Cord	Claude / Claude Code
x Cali draw / x sky clock / x carry draw / XCalendar / x sky clock	Excalidraw
Instinctly / Inslee / Instant Lease / instant least	Instantly (instantly.ai)
easy demark	EasyDMARC
list monk / ListMonk	Listmonk
20 CRM	Twenty CRM
pork bun	Porkbun
f 5 bot / f bot / alt	F5Bot
fabble / Fabul / Fabol / Fabl / a rib Fable	Fable (Claude model he lost access to mid-build)
Versal	Vercel
Atio	Attio
just paid dot a I / JustPay	JustPaid.ai (his YC startup)
3 0 3 redirect	301 redirect (learner comment; permanent redirect intended)
maximum 5 minutes per domain	maximum 5 inboxes/mailboxes per domain (likely; Instantly's standard guidance)
session 15	Outskill-internal numbering for what the LMS lists as Session 7
g b t 5.5 extra high	GPT 5.5 at extra-high reasoning (as heard, in Codex)
AirMail	Instantly's own inbox provider option (new; untested by trainer)
Truffle / Praful	learner Praful
28 octopus	the digits in his childhood Gmail address (birthday-based, as bantered)
evolve, prime, signal dot com	example pre-warmed domain names as heard (gibberish by design)
with company dot com, the company dot com	outreach-domain naming patterns: withcompany.com / thecompany.com
human a e AI / Humanae AI	learner-mentioned ecommerce email tool (deferred to next session, unresolved)

True on recording day — verify before relying

The Fable lament is dated color: he built harshad.com with Claude's Fable model, lost access mid-second-page ('there's an issue with the selected model'), and repeats 'I miss Fable' — plus a garbled 'US banning Fable' aside. Point-in-time model-availability drama, amusingly extracted BY Fable.
Instantly pricing/UI is of-the-day: ~$15/domain pre-warmed, the $585 bundle quote, AirMail as a new untested provider, sender-name glitches during the live demo.
Sam Altman and VAPI-CTO anecdotes are personal receipts as told; the VAPI podcast episode was recorded but unreleased at session time.
skills.sh star counts (Humanizer 24k+, 698k skills) and .ai-domain price commentary are point-in-time.
Cameron's intent-signals session cross-reference left open — 'ask in the LMS'; F5Bot offered as his personal method meanwhile.