App Co-founder Agent

collector → reasoner → actor

every run, in your voice, in your language

​

Get started

• Quickstart — 5 minutes from zero to your first briefing
• How it works — the pipeline under the hood

​

What it does in one sentence

┌──────────┐    ┌──────────┐    ┌────────┐
│ Collector │ →  │ Reasoner │ →  │ Actor  │
└──────────┘    └──────────┘    └────────┘
   data            insights        delivery

​

What it sees today

​

What it produces

:wave: Morning Erencan arica! Here's where we stand with Phosum today:

📊 This period
- Downloads: 13 (+13%)
- Revenue: $0 (-100%)
- Impressions: 3,828
- Conversion: 0%
- ASO Score: 38/100 (Grade D)

🔥 What's on fire
- Conversion is at 0%. We had 3,828 impressions and not a single page view.
- Mexico is driving 25% of downloads with no localization — huge ROI miss.

🎯 Today's focus
1. Metadata overhaul (~1h): rewrite subtitle + first screenshot
2. Localize for MX, IT, GB (~2h): top 3 unlocalized markets
3. Track new keywords (~30 min): "screenshot stitching", "pink possum"

💡 One insight
We're getting real Mexico traffic without localization. Translating could
lift downloads from that storefront alone.

​

What’s special

1. Memory — the agent remembers what worked for your app and stops re-proposing things you ignored.
2. Knowledge Base — profiles, business_context.md, and a markdown Library you edit; the agent reads it on every briefing and can only propose changes. See Knowledge Base.
3. Stale-action follow-ups — if you ignored a critical recommendation 5 days ago, the agent will say so directly.
4. Personality presets — pick a voice that resonates: blunt advisor, friendly mentor, indie hacker, PG analytical, YC partner, or balanced.
5. Multi-language briefings — Turkish, English, German, Spanish, French, Portuguese, Japanese.
6. Competitor move detection — snapshots competitors daily; flags subtitle/keyword/version/pricing changes.
7. Crisis mode — detects review sentiment cliffs (1-star spikes vs your 30-day baseline) and elevates them above the briefing.
8. A/B test memory — when you change your metadata, the agent automatically measures the 14-day before/after impact.
9. Pricing intelligence — compares your IAP/sub price against tracked competitors with percentile ranking.
10. Localization ROI — surfaces countries where you’re getting downloads without local metadata.
11. Public Founder Diary — opt-in: share a sanitized briefing publicly via /cofounder/diary/<your-slug>.
12. Why-agent (RCA) — “why did revenue drop?” gets ranked candidate causes with evidence, not a shrug.
13. Autopilot scorecard — aggregated outcomes of applied actions, so you can grant autonomy based on data.
14. Weekly portfolio briefing — multi-app owners get one cross-app briefing: where to spend the week.
15. Bring your own model — OpenAI, Anthropic, Google, Kimi, or DeepSeek, optionally on your own API key.

​

Capabilities

​

Who it’s for

• Indie developers who want a daily second brain that costs on the order of a cent per run with default models.
• Small teams who want their app intelligence delivered to the channels they already use (Slack, Linear, Notion).
• Power users who want a webhook firehose to plug the agent into their own systems.

​

Quick links

• Quickstart — 5 minutes from zero to your first briefing
• How it works — collector → reasoner → actor pipeline
• Knowledge Base — profiles, Library, patches, and the API behind them
• Integrations — wire up Slack, Linear, GitHub, Notion, Webhook
• API reference — agent runtime, planning, analysis