Claude
First met in aicp-c3-basecamp-01 · Chat assistants
ClaudeFirst met in aicp-c3-basecamp-01 · Chat assistants
Anthropic's assistant — the course's pick once work involves your real files, calendar and tools.
The task needs to touch your actual world (connectors) or produce careful structured output — vs ChatGPT for open-ended planning.
Desktop connectors reach your local filesystem and CPU, which browser-based rivals cannot
Skills: reusable .md instruction packages that can also call tools
Connector reliability the trainer scored ~7/10 in daily use, against ~4/10 for ChatGPT apps
Connects Gmail, Calendar, filesystem and custom MCP servers on the desktop app
Hosts skills — a meta prompt, a humanizer, a taste guide — as reusable instruction files
Generates structured artifacts: mind maps, n8n workflow JSON, schemas, system prompts
Runs thinking/adaptive-thinking modes, which are chain-of-thought built in
Doesn't have unlimited free usage — the free-tier daily limit was hit live in BC4
Doesn't always win on structured config — Codex beat it on n8n JSON with native Firecrawl nodes
Doesn't register new connectors without a full app restart
The everyday assistant seat plus the MCP host seat. In this course it's both the subject (BC5) and the tool used to build things in BC3/BC4/BC6.
Chat assistants
The extraction, the generator work and every sitting run on it.
Skills are the underused half: the extraction protocol and the checklist conventions are both natural skill candidates rather than documents to re-read each sitting.
current
chatgpt: different strengths; the trainer runs both
claude-desktop-mcp: connectors are Claude's MCP surface
claude-code: same family; Claude Code is the repo-aware sibling
| Session | Coverage | Moment | What happened |
|---|---|---|---|
| Basecamp 3: Introduction to n8n — I | demonstrated | 1:18:41 | Skill→system-prompt conversion, workflow mind maps, n8n JSON generation (Sonnet 4.6 — missed native Firecrawl nodes) |
| Basecamp 4: Introduction to n8n — II | demonstrated | 0:28:51 | Workflow mind map, JSON schema generation, humanizer-skill → system prompt, Lovable mega-prompt from attached workflow JSON (Sonnet; free-tier daily limit hit live) |
| Basecamp 5: MCP | demonstrated | 1:56:37 | Recommended over Sonnet 4.6 on free plans — faster, cheaper, more tokens; 'Haiku 4.5' cited for retries |
| Basecamp 6: Voice Agents | demonstrated | 2:51:27 | n8n expression generation, IANA timezone lookup, post-call description drafting, and the discovery-form → first-draft-prompt meta workflow |
| Session 6: Building an AI-Powered Lead Gen Machine | demonstrated | 2:16:31 | Drafted the analyst system prompt; preferred for signal analysis ('OpenAI can hallucinate more given lots of data'); live credit-balance failure swapped it out |
| Session 7: The Outreach Engine | 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 |
| Session 8: The Conversion Playbook | demonstrated | 0:50:58 | Wrote the business description (Opus 4.8, with web-scrape), the ICP fields, and the full 4-email sequence with sequence-builder + Humanizer skills; hallucinated an ICP once under long context |
| AI Sprint: AI Evals & Reliability — Day 2 (Testing Systems That Never Answer Twice) | demonstrated | Sonnet 4.5 as the live demo's LLM judge | |
| Basecamp 1: Prompting & RAGs | explained | 1:43:28 | Thinking modes as built-in CoT; skills vs RAG distinction in Q&A (2:30:13); trainer shifted from ChatGPT to Claude for connectors |
| Session 1: Getting your first Client | explained | 2:38:52 | Offer-crafting partner: 'take this session's transcript, dump it into Claude, iterate your offer'; also between-calls solution design from recorded discovery calls |
| Beginner Glossary (30 terms) + I'm Stuck Guides (23) | explained | Workspace terms defined (Project, instructions, knowledge file, artifact, Skill); chat-vs-memory-vs-Project and Skills-vs-instructions guides | |
| Practice Lab: Six Confidence-Building Projects | explained | Project 5's Project build (factored knowledge, XML instructions) and project 1's artifact option | |
| Learning Topics: Sessions 1–5 — Foundations to AI Director | explained | Projects (free alternative to Custom GPTs in the source snapshot), data-analysis dashboards, Research, and Claude Design for presentations | |
| Workbooks 1–5: Guided Builds with Completion Proof | explained | Workbooks 1 & 3: artifacts (three Visuals toggles as hard checkpoint), AI-powered artifacts, Projects with custom instructions | |
| Session 17: Facebook Ads Competitor Research & Replication | mentioned | 2:24:38 | 'Opus 5 just released, and I think it's also really smart' — endorsed as an alternative code generator to ChatGPT |
| Session 21: Advanced Data Extraction from PDFs — Parsing, OCR, and the Token-Free Product | mentioned | Reached for in the handwriting demo when ChatGPT upload stalled | |
| The Five-Level AI Generalist Roadmap (+ Operating System & Monetization) | mentioned | Across levels: Projects (L2 managed RAG), Claude Code (L4/L5), Cowork and Claude in Chrome (L4 consumer agents), the nine-surface ecosystem snapshot | |
| AI Sprint: Claude Code with Antigravity — Day 1 (Taste as the Moat, the Muscle-and-Mirror Setup, and the Model Council Build) | mentioned | The Mythos drop + Carlini/OpenBSD story frames the sprint; Anthropic's $30B run-rate cited as-heard |