Products we built

These are Phi’s own products, separate from our client work. Phi built and runs them with AI agent fleets: agents write the code, open the PRs, run QA and help operate the sites. Numbers below are measured from GitHub, not estimated.

Measured 2026-10-24.

Personal context layer for AIJul 2026 – present

Dayze

A personal context layer for AI, built and operated by an agent fleet

People want ChatGPT, Claude and Gemini to know their life (who they met, what they planned, what happened) without handing over raw data or trusting a black box. The brief was open-ended: make personal context usable by any assistant while keeping it private, source-linked and correctable.

MCPOAuthNext.jsSupabase / PostgresOpenAI
Read case study
EdTech · Exam preparationAug 2026 – present

CFA Mock Exam

An exam-prep business run day to day by AI agents

CFA candidates need realistic, timed mock exams that feel like the real computer-based test. The business question was whether a small team plus agents could run a paid exam product (content, payments, analytics, security) without a full-time engineering staff.

ReactFirebase / FirestoreNetlify FunctionsStripeGoogle Analytics
Read case study
Crypto trading platformSep 2026 – present

Coincess

A mobile-first crypto app for swaps, perps and predictions

Crypto users juggle separate apps for swaps, bridges, perpetuals and prediction markets. Coincess puts them in one installable app with on-chain signing and real-time positions.

Next.jsHyperliquid APILI.FIPrivyviem
Read case study
AI agent operationsLive

Phi’s AI agent fleet, live

Live numbers from the AI coding agents that build our own products

Ship several live products with a small team, without lowering the bar for what reaches production.

CodexClaude CodeGitHub ActionsSupabase
Read case study