ProPR Wins Best Enterprise Solution at AI Dev Days Hackathon

We’re excited to announce that ProPR has won the prize for Best Enterprise Solution at the Microsoft AI Dev Days Hackathon! 🏆

This recognition matters to us — not as a milestone to celebrate, but as validation of a specific thesis we’ve been building on since day one: that AI code review done right looks very different from a simple prompt wrapper around a diff.

What ProPR Actually Does

ProPR is a self-hosted, multi-tenant AI code reviewer. It connects to Azure DevOps, GitHub, GitLab, and Forgejo-family platforms, monitors pull requests as they open, and posts structured AI reviews directly on the PR thread — using your team’s own AI provider credentials, on your own infrastructure.

Every file changed in a pull request is analyzed with awareness of the full cross-file context. Every review pass is logged. Every finding is traceable. And critically — the reviewer learns from the decisions your team actually makes.

Why We Won: The Memory Argument

What set ProPR apart in the hackathon wasn’t a clever prompt or a clever UI. It was the architecture we built around reviewer memory.

Most AI code review tools treat each pull request as a blank slate. They re-flag the same patterns, re-raise the same concerns, and produce the same noise your team has already tuned out. That’s not a review tool. That’s a second static analysis pass dressed up with better prose.

ProPR takes a different approach. When a thread is resolved — when a developer responds, explains, or fixes something — that decision is stored as a semantic embedding. When the next PR arrives with a similar pattern, the reviewer doesn’t just re-fire the same finding. It reconsiders it in light of what your team already accepted, rejected, or explained. It suppresses duplicates across incremental updates. It builds a review posture that reflects your team’s actual code quality standards, not generic best practices from a pre-trained model.

That’s the core of the enterprise argument: an AI reviewer that improves with real usage, operates inside your existing security boundary, and never sends your source code to a third-party API endpoint.

Built for Multi-Tenant Enterprise Deployments

The judges also recognized ProPR’s operational model. This is not a SaaS tool — it’s designed to run inside your infrastructure, serving multiple client organizations from a single deployment, each with fully isolated configuration.

Every client configures their own SCM connections, their own AI provider (Azure OpenAI, OpenAI, or LiteLLM), their own embedding model, and their own review behavior. Secrets are encrypted client-side before storage. The admin UI never returns them in plain text. Reviewer identity is scoped per provider so every comment is traceable back to a specific integration, not a shared service account.

The result is a code reviewer that enterprise teams can actually trust with a production codebase — not a demo product that works beautifully until it hits a real organization’s security or compliance requirements.

What’s Coming

The hackathon win comes alongside our most significant release yet — 1.0.0.alpha.0005 — which brought semantic review memory, PR-level review rollups, token governance dashboards, and ProCursor knowledge source auditing to the platform. You can read the full release post for the technical detail.

We want to thank the Microsoft Global Community Hackathon team for the recognition, and our early adopters for the feedback that made this version possible.