Wycro Icon Light
Wycro Icon Dark

Firnot

Decision memory for teams: the meetings and work a team chooses to share become searchable decisions, actions and risks, each linked back to the conversation it came from.

Wycro relationship: Firnot is a Wycro product. Wycro Software Developers designed, engineered and operates it, and firnot.com presents itself as “Firnot by Wycro”.

Platforms
Web application
Production status
Live at firnot.com, with a public landing page, a no-signup interactive demo, published pricing tiers, and an authenticated product behind Wycro Auth.

The opportunity

Teams do not lose their decisions. They lose the reasoning behind them. Six months later everyone remembers that the launch moved, and nobody remembers that it moved because a vendor’s API access slipped — so the decision gets re-litigated from scratch, or worse, quietly reversed by someone who never saw the constraint.

The reasoning is not missing; it is scattered. It is in a Slack thread, a Jira comment, a meeting nobody re-watches and a doc nobody re-opens. No single tool holds it, so no single search finds it.

The product approach

Firnot extracts four things from what a team shares — Decisions, Actions, Risks and the Evidence behind them — and keeps every one of them attached to its original source rather than summarizing it away. An answer is only useful if you can click through to the sentence someone actually said.

Institutional memory is treated as versioned, not static: when a decision is superseded, Firnot keeps the old one, the new one, and the new context that caused the change. That is the difference between a knowledge base and a record.

Capture is opt-in by design, and the product says so on its own landing page: no meeting bot is required, and the team decides what becomes memory. That is a product stance, and it shapes the architecture rather than being bolted on.

Engineering and product challenges

Extraction has to be conservative. A decision-memory product that confidently records a decision nobody made is worse than one that records nothing, so the extraction step is built to keep provenance and to prefer under-claiming.

Search has to answer questions, not match keywords — “why did we choose Postgres?” has to reach the right three decisions across months of unrelated material.

It is a multi-tenant B2B product with multi-level visibility (private, team, project, organization), so every read path has to respect the visibility a user chose, and integrations with Jira, GitHub, Slack and calendar have to honour it too.

The product is bilingual in English and Spanish, and exposes an MCP integration so the same decision record is reachable from an AI assistant rather than only from the product UI.

What was delivered

A multi-tenant SaaS web application with authentication through Wycro Auth, tiered plans, and an onboarding path that works before a team has any data.

A retrieval and extraction pipeline producing D.A.R.E. summaries with source links, decision search, supersession-aware institutional memory, and action tracking.

Integrations with Jira, GitHub, Slack and calendar, an MCP integration for AI assistants, a bilingual EN/ES interface, and a no-signup interactive demo on the public landing page.

  • Firnot dashboard showing decisions with linked sources

What Wycro is responsible for

  • Product design and user experience
  • Multi-tenant SaaS architecture
  • AI extraction and retrieval pipeline
  • Decision search and institutional memory
  • Third-party integrations and MCP surface
  • Authentication and multi-level visibility model
  • Bilingual English/Spanish interface
  • Deployment, operation and observability

What this demonstrates

  • AI product engineering with provenance as a first-class requirement
  • B2B SaaS architecture and multi-tenancy
  • Enterprise and team workflows
  • Integrations with the tools teams already run on
  • Search and knowledge systems over unstructured material
  • Consent-aware capture designed into the product, not added later
  • Bilingual product delivery
  • Running an authenticated product in production

Related Wycro services

Want this kind of delivery on your product?

Tell us what you are building. We will tell you honestly what it takes to ship and operate it.

Talk to Wycro
Firnot Case Study — AI Product & B2B SaaS Engineering | Wycro