EaseAccess24 Automation
Internal AI and developer-productivity work: a Slack + Gemini intelligence system for meeting summaries and Jira sprint digests, a Shopify app, a VivaTech 2026 landing page, and an env-sync CLI.
Shipped an internal Slack + Gemini intelligence system that turns meeting recordings into structured summaries and validated Jira sprint digests, plus a Shopify app, the VivaTech 2026 landing page, and a CLI that replaced manual .env sharing for a 12-person team.
EaseAccess24 is where I do AI and developer-productivity engineering across several internal products. The largest is an internal intelligence system (WorkBrain): a self-hosted Slack + Gemini service - Docker, BullMQ, PostgreSQL, and Redis - that turns meeting recordings into structured summaries and validated Jira sprint digests, surfaced in a Next.js dashboard.
Alongside it I built a Shopify app (Remix, Prisma, App Bridge; 5 of 7 phases, verified on a live dev store), the VivaTech 2026 landing page (Next.js, with a live accessibility checker), and an internal env CLI (Vercel) that replaced manual .env sharing for a 12-person engineering team.
Tech Stack
Key Features
- Internal intelligence system (WorkBrain): self-hosted Slack + Gemini service on Docker, BullMQ, PostgreSQL, and Redis
- Meeting recordings turned into structured summaries and validated Jira sprint digests
- Next.js dashboard surfacing the generated summaries and digests
- Shopify app (Remix, Prisma, App Bridge) - 5 of 7 phases, verified on a live dev store
- VivaTech 2026 landing page (Next.js) with a live accessibility checker
- Internal env CLI (Vercel) replacing manual .env sharing for a 12-person team
Technical Highlights
- Self-hosted service orchestration with Docker and BullMQ job queues
- Gemini-driven summarization with validation before Jira sprint digests
- PostgreSQL and Redis for durable state and background processing
- Next.js dashboard and landing-page delivery with an accessibility checker
- Public-safe framing of private internal tooling
Architecture
Intelligence System
- Slack ingestion of meeting recordings via a self-hosted service
- Gemini processing into structured summaries and Jira sprint digests
- BullMQ queues with PostgreSQL and Redis for durable background jobs
- Next.js dashboard for reviewing generated output
Env CLI
- Pull the correct environment context from Vercel
- Validate required variables locally before run or deploy
- Replace manual .env sharing across a 12-person team
Challenges & Solutions
Turning noisy meeting recordings into summaries teams can trust
Validating AI output before it becomes a Jira sprint digest
Shipping several internal products in parallel without dropping quality
Representing private internal tooling publicly without leaking sensitive details