IRIS Copilot

Vision Article: IRIS Copilot Prototype - English (etc.) as IRIS Language

Human natural language-driven AI platform for InterSystems IRIS HealthConnect TIE implementation lifecycles.

This project is built for NHS Trust integration delivery: users describe clinical integration requirements in natural language, Copilot designs and generates IRIS artifacts, and deployment is executed only after explicit human approval.

Product Vision

IRIS Copilot should support full lifecycle delivery in one controlled workflow:

1. Capture requirements in human natural language (English and other languages)
2. Discover current IRIS production context
3. Design topology deltas (BS/BP/BO/DTL/rules/lookups)
4. Generate compilable ObjectScript artifacts
5. Human review + approval gate
6. Compile/deploy/update production
7. Run tests and monitor post-deploy health
8. Rollback by version snapshot if required

Current Repository Scope

• IRIS server-side classes (cls/AIAgent/*) for API, UI, engine, model, templates, installer
• Node bridge (bridge/*) for runner orchestration and AI provider adapters
• Deployment tooling (deploy/*) including XML export generator and import helper
• Operational docs and demo lifecycle guides (docs/*)

Architecture

• CSP Chat UI: AIAgent.UI.Chat.cls
• IRIS REST dispatcher: AIAgent.API.Dispatcher
• Orchestrator/engine services in IRIS
• Node bridge adapters for:
  • Claude Agent SDK
  • OpenAI Codex (standard API runner)
  • OpenAI Codex SDK runner
  • Azure OpenAI (optional)

Quick Start

1. Deploy IRIS classes

Use one of:

• docs/DEPLOYMENT-GUIDE.md
• deploy/DEPLOY-README.md

Typical terminal flow:

zn ""
do $system.OBJ.Load("\\deploy\\ImportAll.cls", "ck")
do ##class(AIAgent.Deploy.ImportAll).Run("\\cls\\")
do ##class(AIAgent.Install.Installer).Run()

2. Create IRIS Web Application

In IRIS Management Portal, create/update two web applications:

1. /<namespace>/ for CSP pages (for AIAgent.UI.Chat.cls)
2. /ai for REST APIs (dispatch class AIAgent.API.Dispatcher)

Recommended settings:

• Namespace: your target namespace (for example DEMO2_AI2)
• Dispatch Class (for /ai): AIAgent.API.Dispatcher
• Enable Authentication: Password or delegated SSO per site policy
• Allowed Authentication Methods: align with your enterprise security model

3. Start bridge

cd bridge
npm install
npm run build
npm start

Configure keys and runner settings in:

• bridge/.env (local)
• bridge/.env.example (template)

4. Open Chat UI

http://:52773/csp//AIAgent.UI.Chat.cls

Health checks:

• IRIS API: http://<iris-host>:52773/ai/health
• Bridge: http://localhost:3100/api/health

Key Docs

• Deployment: docs/DEPLOYMENT-GUIDE.md
• User workflows: docs/USER-GUIDE.md
• End-to-end demo scenario: docs/DEMO-LIFECYCLE.md
• Design reference: docs/DESIGN.md
• Test coverage plan: docs/TEST-CASES.md
• Implementation backlog: docs/IMPLEMENTATION-PLAN.md

Safety and Governance

• Human approval required before deployment actions
• Audit logging and version snapshot support
• Intended for controlled healthcare integration environments

Repository Notes

• This repo is standalone under Site/AIAgent
• Secrets are excluded via .gitignore (.env, build artifacts, generated export XML)

License

This project is licensed under the MIT License.
See LICENSE for the full license text.

Copyright (c) Lightweight Integration Limited.