Agent Skills: Founder Stage 3 — Launch (prove the *business* deserves to grow)

Founder Stage 3 — Launch (prove the business deserves to grow)

If MVP proved the product deserves to exist, Launch proves the business deserves to grow. Finding PMF was the hard part; now the challenge is keeping it while the organization around the product keeps up. All three capability classes are in full use and compound — coding builds the product, automation builds the company, research operationalizes the knowledge.

Goals

Turn early traction into a repeatable, sustainable growth engine; harden the infrastructure underneath; and build an actual company around the product. The aim isn't to remove the founder — it's to build operational systems that free founder attention for the decisions only a founder can make.

Exit gate — three elements

Advance only when all three hold:

• [ ] Growth is repeatable and channel-driven — you acquire users predictably through specific channels with CAC, LTV, and payback period you know and can defend.
• [ ] The product handles production workloads — infra hardened; security & compliance in order; reliability holds under real conditions (not just tested ones).
• [ ] Operations run without founder bottlenecks — processes + automation exist; you're no longer personally handling support, triage, sprint planning, or reporting.

Failure modes → mitigations

| Failure mode | What it looks like | Mitigation | |---|---|---| | Technical debt comes due | MVP shortcuts now accrue interest under production traffic + new features | Architectural audit → targeted refactor → expand test coverage (exercise A) | | The founder becomes the bottleneck | Decisions that should take an hour now take a week; support piles up because only you know the answer; tasks happen only when you remember | Audit everything you touch; systematize / delegate / keep only founder-judgment work (exercise B) | | Security & compliance no longer deferrable | Real users, real data, enterprise contracts; theoretical risk becomes real exposure | Systematic review before scale; treat findings as required remediation, not suggestions (exercise C) | | Expansion before you're ready | New markets/funding look like growth but introduce variables your product wasn't built for | Resist premature expansion; protect the original user base; expand only with evidence |

Exercises (reusable prompts)

A. Architectural audit + remediation sequencing

"Audit my MVP codebase and produce a prioritized list of structural weaknesses, test-coverage gaps, and refactoring candidates. Then sequence the remediation across sprints: what must be fixed before the next release, what can run in parallel with feature work, and what is acceptable ongoing debt at this stage. Also help me capture the MVP-era architectural decisions that only lived in my head into the CLAUDE.md so future sessions share one mental model."

B. Founder-attention / bottleneck audit

"Run a structured audit of my current operational load: every recurring task, every decision that lands on me, every workflow that only happens because I remember it. Categorize each into: (1) automate entirely, (2) needs a human but not me, (3) genuinely requires founder judgment. For the automation candidates, design the workflow logic — trigger, decision rules, output, and where it goes when done."

C. Security & compliance workstream

"Surface code-level issues that commonly come up in SOC 2 / GDPR / HIPAA audits for my target market. Produce two things: a prioritized remediation sequence, and a list of the documentation + controls (audit logging, access management) an enterprise buyer will ask for before signing. Build this as a continuous workstream in the dev cycle, not a one-time project." (AI scans aid but don't replace a qualified compliance review.)

D. Lightweight PM operating system

"Design a lightweight product-management operating system that runs without me triggering it: a sprint cadence, a minimum spec template (what a spec must include before code touches a feature), a bug-triage decision tree, and a weekly metrics brief that pulls from my actual data sources. Then specify what an automation layer should run on schedule — ceremony scheduling, bug routing, report compilation."

When to leave

• All three exit elements hold → advance to founder-stage-scale.

Guard-rails

• Compliance/security AI output is an aid, not a substitute for qualified human review (especially before signing regulated-industry or enterprise contracts).
• "Free your attention" ≠ "remove yourself" — keep the founder-only decisions (product narrative, key relationships, high-stakes calls).

Attribution

Adapted (process & methodology, original prose) from Anthropic, The Founder's Playbook: Building an AI-Native Startup (2026-05-14) — https://claude.com/blog/the-founders-playbook. No text reproduced verbatim.