Building for the agent era

Confidence in code.
Governance in practice.

Codifide designs programming languages, platforms, and governance frameworks for a world where AI agents write, read, and reason about code alongside humans.

The Thinking — Loop Engineering → The Framework — Stage-Gate-Loop →

Research & Field Notes

Practitioner essays on building governed, secure, agentic systems in production.

AI Governance Thinking

Agentic AI removes the speed constraint on building. It does not remove the constraint on safety. The question isn't "can AI build it?" — it's "should this ship?"

"The danger isn't that AI builds bad software. It's that AI builds software so fast that governance can't keep up."

Execution at machine speed

Loop what can be independently verified. CI/CD, regression tests, security scanning, and policy enforcement scale with agent velocity — they don't wait for humans.

Judgment at human altitude

Gate architecture, domain decisions, security, and consequential tradeoffs. The builder cannot review their own work — adversarial reviewers attack every gate.

Intent must remain intact

Primum monitors whether execution is still accomplishing what humans authorized. Intent Contracts preserve accountability through every agent transformation.

Explore the research → Follow on LinkedIn

Active Projects

Four active projects at the intersection of AI, software, and real-world impact.

AI Governance · Open Source

Agentic Stage-Gate-Loop Governance

Machine-speed execution. Human-grade checks and balances.

AI agents build at machine speed. Governance must operate at three different speeds simultaneously — or it becomes ceremonial. This open-source framework separates human decisions, security judgment, and continuous machine verification into three layers, each running at the cadence it requires.

HUMAN — Intent · Accountability · Decisions
↓
GATE — Architecture · Domain Expertise · Security · Adversarial Judgment
↓
LOOP — Build · Test · Verify · Deploy · Optimize
Primum — Intent Integrity Quill — Organizational Memory
  • Three-layer architecture: Human decisions, Gate judgment, Loop verification
  • Automate repeatable assurance — CI/CD, scanning, policy enforcement scale with agent velocity
  • Three-model adversarial review: A-Team builder, B-Team critic, Zero-Context auditor
  • Forensic code review prompts for reasoning models (Qwen3, o3, DeepSeek-R1)
  • Intent Contracts — preserve what humans authorized through every transformation
  • Domain expert personas for consequential decisions (healthcare, security, compliance)
  • Works with any AI IDE — Kiro, Cursor, Windsurf, GitHub Copilot
"Speed without governance is faster failure. Evidence, not confidence. 'We think it's fine' is not a gate pass."
View on GitHub → Research & field notes →
Open Source · v4.0

The Codifide Programming Language

A contract-and-dispatch language for agent-to-agent trust.

Not a systems language. Not general-purpose. Codifide optimizes for declaring intent and contracts, dispatching among candidates, enforcing effects, and content-addressing compositions — the properties that matter when agents write, read, and compose code alongside each other.

  • Intent preserved as structure — not a comment the compiler discards
  • Effects declared in the type, enforced transitively across the call graph
  • Contracts evaluated by the runtime, not trusted as docstrings
  • Confidence as a first-class dispatch mechanism
  • First-class refusal — bottom is a value, not an exception
  • Content-addressed identity — same hash, same bytes, same trust
Python Rust 461 tests MIT License
Explore the language →
Healthcare · AI · HIPAA

Codifide Health

AI-powered healthcare intelligence, built for compliance from the ground up.

A Tuva-aligned analytics platform designed to support 10M+ patients at scale, with AI clinical decision support, federated machine learning, and multi-region deployment. Built for HIPAA/SOC2 compliance from the ground up — not bolted on.

  • FHIR-native EHR integration with SMART on FHIR OAuth (Epic, Cerner, athenahealth, eClinicalWorks)
  • Fail-closed organization enforcement — no implicit tenant fallbacks
  • GPT-4 powered clinical assistant with audit trails
  • MIPS/HEDIS quality measures with real-time updates
  • Zero-trust security with ABAC policies, HIPAA/SOC2 ready
FHIR Kafka ClickHouse Keycloak Federated ML

Private project — interested in contributing? Get in touch.

iOS · Consumer · AI

decode the sign

parking, interpreted.

A native Swift/SwiftUI iOS app that reads any parking sign and tells you YES, NO, or SOON in seconds. Point your camera, get a plain-language answer — then track your session, get alerts before your time runs out, and navigate back to your car. Built with a time-aware rule engine, AI-powered OCR, and conservative low-confidence gating — it refuses to guess rather than risk being wrong.

  • On-device inference — verdicts in ~750ms via Apple Intelligence, no server needed
  • AI sign reading — every panel, every restriction, every time window
  • YES / NO / SOON verdict with next-change countdown
  • Directional arrow detection — knows which curb side each rule applies to
  • Stale-cache cross-check — detects when server served old data, falls back to on-device
  • Parking session tracking with Live Activity and Dynamic Island
  • Find My Car compass navigation with arrival detection
  • App Intents (Siri) — "Hey Siri, can I park here?"
  • Home Screen Widget with session countdown
  • Crowdsourced sign verification with anti-gaming trust system
  • Conservative low-confidence gating — refuses rather than guesses
Swift SwiftUI Core ML WidgetKit ActivityKit Next.js Supabase Vercel
Visit decodethesign.com →

Private project — interested in contributing? Get in touch.

About Codifide

Codifide Inc. is a software company and research practice building tools for the agent era. The name means the same thing for both the company and the language it stewards: confidence in code.

The work spans language design, healthcare intelligence, consumer AI, and governance frameworks — connected by a single thesis: the properties that matter most in AI-assisted software are the ones that need to be enforced by the system, not trusted to the agent.

Founded and led by Douglas Jones, an engineering leader, hands-on software architect, and AI-native builder whose work focuses on applying agentic systems to consequential production software while preserving security, verification, governance, and human accountability.

Veteran-owned small business

Connect

Get in touch

Interested in the work, a collaboration, or a conversation about AI governance?

Send a message