Peer

An AI teammate that learns how you work and takes tasks off your plate

Role: Product Lead, Fruity AI
Year: 2026

product design
design engineering
prototyping
SaaS
web

As Product Lead at Fruity AI, I designed and built Peer, an AI teammate that learns how you work, connects to the tools you already use, and runs repetitive tasks on demand or on a schedule, all in an isolated, secure environment.

Visit Peer

My role

Product Design
UX Design
Product Strategy
Frontend Development
Cloud Architecture

The problem

Solo founders, freelancers, and small teams lose hours every week to repetitive busywork: summarising metrics, triaging issues, pulling data from one tool and pushing it into another, formatting reports, posting updates. Generic AI chatbots can draft text but can't actually do the work; they don't connect to the apps people live in, can't run on a schedule, and can't be trusted with sensitive credentials. The result is capable people stuck doing the boring parts by hand because there's nobody to hand them to.

The process

As Product Lead at Fruity AI, I designed Peer to feel like a colleague rather than a tool. You describe a task in plain language, no templates or forms, and Peer figures out which tools it needs, sets them up, and runs the task on its own. It connects to the apps people already use across email and calendar, accounting and finance, CRM, project management, communication, document storage, social media, and analytics, essentially any app with an API. It learns new skills on the job, asks smart follow-up questions when something is unclear, and surfaces its work step by step so you can review and approve before anything ships. Recurring jobs run on autopilot on a schedule. The whole system runs on Google Cloud (Cloud Run, Cloud Storage, Secret Manager) with an isolated environment per user, credentials encrypted and injected at runtime, never stored, and it scales to zero so you only pay for compute used.

The outcome

Peer turns one-off requests into standing automations: describe a task once, approve the result, and Peer offers to run it every week without being asked again. By keeping a human in the approval loop while doing the actual fetching, formatting, and posting, it removes the busywork without removing oversight. The isolated, scale-to-zero architecture keeps each user's data private and the running cost low, making a genuinely useful AI teammate viable for solo founders and small teams rather than only large companies.

Connects to any app with an API
Chat with Peer via Native app, Web app, Teams, Slack and Telegram
Runs tasks on demand or on a recurring schedule
Human-in-the-loop review and approve before anything ships
Isolated environment per user, secrets injected at runtime, scales to zero