Show Notes
Parker lays out the core skills AI coders need to avoid being NGMI in 2025, sharing a concrete vibe-coding framework, plus updates on the daily and main channels, community accountability, and future projects.
Core thesis: the missing skills for AI coders in 2025
- It’s not enough to know how to tinker with models. You need a plan-to-deliver workflow that combines planning, multiple expert roles, testing, and a repeatable prompt-driven process.
- The goal: build real value fast by turning experiments into products through disciplined, repetitive practice and public accountability.
The vibe coding blueprint
- Roles (three core pillars)
- R1/R2: Researchers – gather context, data, and domain knowledge
- F1: Technical PM – planning, requirements, and coordinating the flow
- F2: Solutions Architect – design of the approach, architectures, and integrations
- F3: Execution Engineer – the hands-on implementation and outputs
- Flow and outputs
- Start with manual iteration to define outputs, then scale with prompts
- Outputs feed into a central tracking system (e.g., a sheet) for evaluation and iteration
- Each step is a prompt; you chain prompts to form a multi-step agent workflow
- Testing and QA
- T1: Unit tests between steps
- Q1: End-to-end QA checks
- CI/CD workflow and GitHub integration (G1/MCP-type bits) to move from test to live
- Data flow and tooling
- Build an ETL-like engine to normalize incoming data (bank data, APIs, etc.)
- Create downstream tools (e.g., type generation) to bridge front-end and back-end
- Plan for live deployment and feature flags to test in production safely
Planning before vibe coding
- Don’t rush to ship. The big wins come from planning, not just speed.
- Do multiple “reps” of the process: plan, execute, test, refine, repeat.
- If you blip and skip planning, you’ll waste hours down the line fixing preventable issues.
Practical tips and tools discussed
- Data scraping and data collection
- Instant Data Scraper (Chrome extension) for quick scraping jobs
- Apify for scalable scrapers and automation
- Nick Surve as a reference for scalable scraping approaches
- Model and platform choices
- Gemini 2.5 Pro and raw client usage vs. Cursor; raw tends to be more reliable when you know your flow
- Vertex AI and related tooling for large-scale experiments (as discussed; follow setup guides if you pursue)
- Architecture and ecosystems
- Next.js with TRPC11; preference for clean API surfaces and cross-platform logic
- Hono as a lightweight server option (but evaluate what fits your stack)
- ETL and data harmonization concepts (engine to normalize differing provider data)
- Generating types and bridging front-end to back-end to DB types (a practical dev-time helper)
Projects and open-source direction
- Map app revival and open-sourcing the project to attract contributors
- Multiplatform approach: iOS native vs Expo; satellite/companion apps with deliberate constraints (e.g., avoiding commerce in certain contexts)
- “Vibe TV” and Pool Suite inspiration to create a visual, MTV-like dev network
- Cross-pollination with the community to drive new use cases and SAS opportunities
Building in public and the flywheel
- Public accountability drives consistency and feedback: your audience becomes a quality control and production coach
- Community fuel: feedback, use cases, and diverse viewpoints accelerate learning and product ideas
- The long-term plan: turn these patterns into a SAS built from community-driven use cases; grow a guest lineup (Mount Rushmore-style) for Vibe TV
1K per month: a concise path (from the chat)
- Find a real problem in your life and solve it with tech
- Build something you’d use, then offer free value publicly (open-source can help)
- Talk about what you built online to attract attention and potential buyers
- Turn the repeatable process into paid services or products once you’ve proven value
- Expect to do it through “sets and reps” across your life: improve technical depth, build, and iterate
Community notes and belt concept
- The belt idea is a playful way to recognize strong comment contributions and practical ideas
- The belt and avatar offer are incentives for constructive feedback and engagement
Closing notes and what’s next
- Expect three videos per week on the main channel starting next week; daily updates continue
- The plan includes more builds, QA, and strategy content; open-sourcing aspects like Map
- Ongoing focus on practical, not hype-driven, AI progress; keep projects fun and sustainable
Links
- Instant Data Scraper (Chrome extension)
- Apify
- Midday and related cross-platform patterns
- Next.js and tRPC (architecture references)
- Hono (server option)
- Vertex AI and Gemini 2.5 Pro discussions
- Apple developer docs (scraping context and learning approach)
- Poolsuite – window-manager-inspired UI concepts
If you want the belt, drop a comment with the word motor. Also, look for three new videos next week on the main channel, with the daily updates continuing on this channel.