Portcullis

Architectural enforcement for AI-generated code.Define your architectural rules once.
We enforce them deterministically on every change,
regardless of who or what wrote the code.The governance layer for the agentic development era.

• Learn more

The Problem

AI coding tools generate code faster than teams can review it.
The code compiles. It passes tests.
But it quietly violates architectural boundaries that took years to establish.

97% of engineering teams use AI coding tools.
Only 19% have governance over what those tools produce.

OpenAI calls mechanical enforcement of architectural constraints
an early prerequisite for autonomous agent development.
We productize that enforcement.

The Product

Portcullis is a deterministic infrastructure that maps the dependencies across your entire codebase and checks them against your declared architectural rules.

No AI in the analysis loop. Same code, same rules, same results. Every time.

How it works:

1. Connect_{Point Portcullis at your repo.}

2. Define_{Declare your architectural rules in a simple config (or let us provide a starting point).}

3. Enforce_{Every change is checked against your rules.
Violations surface with exact file, function, line
number, and connection type.}

Current stage: Working analysis engine validated on private and OSS codebases (Anthropic Python SDK, LangChain, etc.) with zero false positives. GitHub Action and governance dashboard in active development. Python support shipping now. TypeScript/JavaScript next.

Why Portcullis?

Deep Codebase Understanding

Understands why code exists, not just what code exists.

Catches What Tests Miss

AI code passes tests but violate architecture.

Works With Any AI Tool

Copilot, Cursor, Claude: Validate code from any source before shipping.

• Check

Tested. Approved. Trusted.

In a controlled benchmark on a Django security vulnerability (CVE-2025-13372), Portcullis-assisted code achieved 83% architectural alignment with a Django core maintainer's fix.Vanilla Copilot: 33%.

• View Demo

Meet the Founding Team

Justin K. — Founder & CEO
Software Engineer with enterprise cloud platform experience.
MS in AI, Georgia Tech.
Background in cloud governance/compliance tooling at world's largest software company.

Horace — Co-Founder & CTO
Senior engineer at FANNG.
Distributed systems and hyperscalar infrastructure.
Built core Portcullis graph analysis engine.

Technical Advisors:

Senior Solutions Architect with graph ontology expertise
Previously built threat detection graphs at a cybersecurity leaderStaff Platform Engineer at Frontier AI LabSenior Frontend Engineer with enterprise SDLC experience
at world's largest asset manager.

Pricing

Free — Architectural visibility on any repository via GitHub Action.
See your codebase structure and dependency hotspots instantly.

Pro — Enforce architectural rules on every PR.
Violations block merges. Full audit trail.

Enterprise — Temporal trend analysis, governance dashboard,
compliance reporting, multi-repo support.

Currently onboarding pilot customers in AI-native startups and regulated industries.

Get Involved

We are currently working with a small group of engineering teams.
Leave your email to get involved.

Portcullis AI

You wouldn't put your server on the internet without Cloudflare.
Why let AI generated code in your codebase without Portcullis?