Decision: discovery niche, not ChatGPT-app agency
Current truth
- The window for building ChatGPT apps as a standalone agency offer has passed (“that ship has sailed” — would have been right 3–4 months ago). No revival of app-building as its own business.
- Refined 2026-06-15: app-building is back on the table — but only as the “build” tier inside the full-service discovery offering (build the connector/app if it’s missing → then optimize/test/discover), under one consolidated brand. Still NOT a standalone ChatGPT-app agency. Vincent: “still convinced building ChatGPT apps is a lucrative offering… we never properly committed.” See 2026-06-15-offering-pricing-fundraise.
- Operating principle adopted: now that we have technology and a niche (AI discoverability), do the same repeatable offer consistently — outreach every month, same offer — and only change what’s demonstrably not working. Horizon: think in terms of a year.
- Context: post-mortem of Jan–Feb (Statista, Mitchells & Butlers, ManpowerGroup era) — capacity, not demand, killed the app-agency angle; Dripify outreach got lukewarm response with poor follow-up.
- Focus boundary set 2026-06-08: ChatGPT + Claude registries/connectors are the niche (Codex on the radar as it gets consumer-bound; Gemini watched, “probably not” GitHub Copilot); web optimization beyond AEO is NOT core — usable only as a free marketing-funnel/audit offering. See agent-discovery-market for the full reasoning vs Ora.
- Reaffirmed 2026-06-17: “agent-ready web” may exist as a supporting page / audit / progression step only if customer demand proves it useful, but it is still not the primary bet. The core offer remains product discoverability inside ChatGPT, Claude, and Gemini via trusted registries/connectors.
Open questions
- Retro still owed on why outreach never became a cadence (named as a sprint-plan goal).
Timeline
- [2026-06-17] (internal discussion — merged Vincent+Elliot “Service offering” notes) Re-tested the scope boundary. Vincent argued for a broader wrapper around agent-ready-web surfaces because it is easier to discuss with more companies; Elliot argued that this risks repeating the team’s old spread-too-thin pattern. Resolution held: broader agent-ready-web can be a supporting page/audit if customers ask, but the actual business bet stays on connectors/apps in trusted registries.
- [2026-06-15] (internal discussion — Vincent + Elliot retro) Refined: the build-then-optimize offering re-opens app builds as a tier within the discovery offer (not a separate agency); consolidated under one brand/narrative. See 2026-06-15-offering-pricing-fundraise.
- [2026-06-10] (internal discussion) Decision affirmed by both founders during stand-up.
- [2026-06-08] (internal discussion) Quarterly session: focus locked to ChatGPT + Claude (“we just forget the rest for now”); web/agent-ready-web framed as free marketing funnel only, pending harness research; assumptions + predictions doc to be completed over the week (target Fri 2026-06-12). Vincent: “we’re the contrarians” vs Ora’s web-first bet — accepted, with the explicit open task of articulating why.
- [2026-06-05] (internal discussion) Precursor debate: Elliot “half bought in” to the consumer/organic vision, wants structured reflection on direction (thesis update as “things that need to be true”); Vincent confident, argues focus on ChatGPT/Codex/Claude clients is itself the smarter way to pivot later, and warns against thinking too big while the market splinters. Agreed: write it down in quarterly planning (session set for Mon 2026-06-08 before Elliot’s surgery); consider a senior advisor (web/mobile platform experience, or a younger person deep in agent discoverability) to challenge their thinking.