Intrinsic One — the agency operating system
One tool holding everything an agency knows: every client, the deal pipeline, tasks, invoices, agreements, proposals, email triage, meeting agendas and the case study library. Built to replace a scatter of subscriptions, and now the thing the team actually works in.
STATUS · live, in daily use · SCREENS · ~20 across 7 nav groups · captured from production 3 Sept 2026
The problem
Agency knowledge lived in Asana, spreadsheets, inboxes and heads. Answering "what's the state of this client?" meant opening four apps and trusting memory. Off the shelf CRMs wanted the agency to work their way; the agency wanted a tool that worked its way.
The build
A bespoke web platform where one client record is the spine everything hangs off: pipeline, work, mail, money and documents all attach to it. When it went live, 1,150 tasks came across from Asana — every completed job back to October 2023 plus everything still open — and Asana was switched off.
The whole agency on one screen: recurring revenue, weighted pipeline, overdue work rendered loudly on purpose, and renewals inside 60 days pulled straight from the client spine.
One record per relationship. Each card carries billing, services, renewal dates and last contact, with leads, prospects and lapsed clients filtered off by default.
The deal board, with a weighted forecast and guardrails: a deal can't advance without a value on it, and deals that go quiet for 21 days get flagged in orange.
Tasks grouped by client, the way the team already works. Overdue is loud by design: it is the only way it gets dealt with.
- First draft of every screen, module and database table
- The Asana migration tooling — 1,150 tasks with history intact
- Email sync and matching against client records
- The boring 80%: forms, lists, filters, plumbing
- Decided the client record is the spine everything hangs off
- The guardrails — no deal advances without a value; wins recorded on the deal, not by drag
- Role based access, and an audit trail on every change: who, and why
- Review, security and the call on what not to build
We ran a full UX audit on our own tool three weeks after go live and wrote down everything wrong with it, measured from production: screens too long, no global search, numbers on one scorecard that weren't yet real measurements. That audit is now the improvement backlog.
That's the practice in miniature: build fast with the machine, then let a human be ruthless about what's actually true.