FinishLine

Yes. Asymmetri built a hosted Model Context Protocol (MCP) server for FinishLine, its own delivery platform, so Claude, ChatGPT and Cursor can read and act in it as the person who connected them. It includes a full OAuth 2.1 authorization server with PKCE and refresh-token replay detection, 33 tools, per-person access levels, an audit log and a company-wide kill switch. Agent writes run through the same actions as the web app, so permissions and notifications are identical.

Asymmetri keeps row-level security as the only security boundary and changes how it is evaluated. In FinishLine, per-row permission functions were replaced with set-based policies computed once per query, hot functions use generic plans, and read and write policies are split. A Manager's read of all tasks dropped from 160 ms to about 55 ms in testing, with about 3,650 permission assertions proving nothing became less strict.

Asymmetri measures first, then trims every layer. For FinishLine on Vercel that meant verifying sessions locally instead of calling the auth service on every page, lazy-loading heavy libraries, deferring notifications until after the response, and a keep-warm ping. Server load fell from about 1.6 s to 0.4 s in testing, and a 300-row list renders in 503 ms instead of 783 ms.

Yes. FinishLine syncs with Linear both ways using a three-way merge, an outbox and keyset catch-up so no update is lost or echoed; bridges client Slack channels; tracks GitHub pull requests and checks; and keeps Google Calendar current through push notifications that fold bursts of changes into a single sync, with a scheduled reconcile as a safety net.

Asymmetri builds with AI-assisted engineering and holds the output to the same bar as any code. FinishLine's second version was rebuilt in about five and a half weeks, with AI co-authoring most of its 674 commits, alongside about 2,460 unit tests and about 3,650 database permission assertions. The team also runs its own day through AI agents connected to FinishLine.

Asymmetri tests permissions in the database itself. FinishLine has 81 row-level security test files with about 3,650 assertions that sign in as each role (admin, manager, internal, external and client) inside a rolled-back transaction and check exactly what each may read, change or never see, including what AI agents can do on a person's behalf.

Asymmetri is a hard problems technology company. It engineers AI systems, agents and digital products for global brands, across four areas: intelligence, process, experience and scale. Asymmetri works through four practices: AI agentic systems and data, application development, technology consulting, and innovation and R&D.

Every Asymmetri engagement runs one loop with five stages. Frame: find the valuable decision, workflow or experience. Design: shape the interaction, context and control model. Engineer: build the intelligence, agent and tool chain. Prove: evaluate performance, safety and economics. Operate: deploy, observe and improve in the real workflow. The loop repeats, so each pass lands closer to the target. Asymmetri's case studies show each hard problem moving through it.

Send a short brief through asymmetri.in/contact-us or email hey@asymmetri.in: the problem, what's in the way and your timeline. No deck or RFP is needed. Asymmetri replies to set up a framing conversation, then proposes a prototype or pilot with scope, team, timeline and what success looks like.