Vyrric

Product intelligence for teams that build from customer evidence.

Comment

Inspiration

Product teams collect a lot of customer evidence: calls, notes, support tickets, interviews, feedback forms, documents, CSVs, and internal discussions. But that evidence usually gets scattered across tools, making it hard to connect what customers are saying with what the team actually builds.

We built Vyrric because product decisions should be grounded in real customer signals, not guesswork or disconnected notes. The goal is to give product teams one place to turn evidence into insights, proposals, and launch plans.

What it does

Vyrric is a product intelligence OS that helps teams turn customer evidence into better product decisions.

Teams can upload or connect product evidence, organize it into a searchable workspace, ask product questions, and generate structured outputs from the evidence. Vyrric helps teams move from raw customer signals to:

  • source-backed insights
  • product proposals
  • implementation briefs
  • launch plans
  • readiness checks
  • success metrics
  • post-launch monitoring context

Instead of just summarizing feedback, Vyrric helps teams decide what to build, why it matters, how to scope it, and how to launch it with clearer context.

How we built it

Vyrric is built as a full-stack web application.

The frontend is built with React, Vite, TypeScript, Bun, Tailwind CSS, Clerk, TanStack Query, TanStack Table, tRPC client, React Router, Recharts, React Markdown, and modern component/UI tooling. The frontend package also includes tools for forms, charts, markdown rendering, command menus, and interactive product workflows.

The backend is a product intelligence monorepo using Fastify and tRPC for the API layer, Prisma for the data model, PostgreSQL on Neon for storage, BullMQ and Redis for background jobs, scheduled cron processes, and modular AI helpers for insight and proposal generation.

The backend is split into separate services for the API, workers, and cron jobs. The API exposes the public product interface, the worker service handles async jobs, and the cron service schedules recurring workflows. The backend also includes packages for database access, API routers, workspace-aware authentication, AI helpers, configuration, queues, and integrations.

For AI and evidence processing, Vyrric uses OpenRouter-compatible chat completion support, LangChain text splitting, Pinecone, Cohere, S3-compatible storage, PDF parsing, Word document parsing, clustering, and background processing pipelines.

Challenges we ran into

The biggest challenge was designing the system around trust. Product teams should not blindly accept AI-generated recommendations, so Vyrric is designed to keep product outputs connected to customer evidence and workspace context.

Another challenge was turning messy, unstructured inputs into structured product workflows. Customer evidence can come from calls, notes, CSVs, PDFs, docs, and other sources. Making that information searchable, useful, and reliable requires ingestion, parsing, normalization, retrieval, and AI-assisted synthesis.

We also had to think carefully about how to move beyond simple summarization. Product teams do not only need summaries. They need proposals, implementation briefs, launch plans, success metrics, and a way to keep learning after shipping.

Accomplishments that we’re proud of

We are proud that Vyrric connects multiple parts of the product workflow into one system: evidence collection, product questioning, insight generation, proposal creation, launch planning, and post-launch monitoring.

We are also proud of the architecture. Vyrric combines a modern React frontend with a modular backend, async workers, scheduled jobs, product intelligence AI helpers, and document-processing infrastructure. This makes the product feel less like a chatbot and more like an operating system for evidence-backed product work.

What we learned

We learned that the most useful AI product workflows are not just about generating text. They are about creating structure from messy information.

A good product intelligence system needs to preserve context, connect recommendations back to evidence, and produce outputs that teams can actually use. That means the AI layer has to work with product workflows, not sit outside them.

We also learned that product teams need more than dashboards. They need a system that helps them move from customer signal to decision, from decision to proposal, and from proposal to launch.

What’s next for Vyrric

Next, we want to expand Vyrric’s integrations with the tools product teams already use, including customer support platforms, research repositories, product analytics tools, project management systems, and engineering workflows.

We also plan to improve source citation, evidence scoring, proposal prioritization, launch monitoring, export workflows, and collaboration features so teams can move from customer insight to execution faster and with more confidence.

Built With

Share this project:

Updates