What is ADRI?
ADRI is a distributed AI architecture built with LangChain, PostgreSQL with pgvector. Each agent has its own pool of LLM models with automatic failover, to maintain availability if a model fails.
Specialised agents
Supervisor
Main orchestrator that coordinates all agents, divides complex tasks and manages data flow. Model: GLM / Kimi
Code Orchestrator
Development, debugging, security, SEO and web search with 20+ tools. Model: Claude / Qwen
Vision
OCR, image, video and audio analysis, TTS and QR codes. Model: GLM / Gemini
Web Search
Web search with SearXNG, Brave and Tavily. Retrieves up-to-date information. Model: GLM / Kimi
OSINT
Open-source intelligence: domains, IPs, emails, usernames, metadata and tech stack. Model: Claude / Qwen
SEO IA
SEO audit 0-100, 5 categories, AI-SEO and Rich Snippets. Model: Claude / Qwen
Security
Pentesting, nmap, nikto, nuclei, OWASP ZAP and forensics. Model: Claude / Gemini
Web Scraper
Fast parallel scraping (5 URLs), content extraction and indexing. Model: Qwen / GLM
Dual Architecture: ADRI + BionicAdri
BionicAdri combines two systems that complement each other perfectly:
- BionicAdri: For structured data, dashboards, scheduled reports and business management.
- ADRI: For flexible processing, conversational AI and intelligent automation.
Supported LLM Models
ADRI supports multiple LLM providers:
- Claude - For complex reasoning and code
- Gemini - For vision and multimodal tasks
- GLM - For quick and general responses
- Qwen - For code and technical analysis
- Kimi - For long context and web search
- DeepSeek - For strategic planning
It can also work with local models on GPU/NPU for companies with strict privacy requirements.
Persistent Memory
ADRI maintains unlimited conversations thanks to a PostgreSQL memory system with auto-summarisation at 80K tokens. Each user has their own context that is preserved between sessions.
Who is ADRI for?
ADRI is designed for:
- Companies that want to automate repetitive processes
- Development teams that need a code assistant available at any time
- Marketing departments that require SEO and automated content
- Startups that cannot afford large teams
- Consultancies that want to scale their delivery capacity
Conclusion
ADRI represents the natural evolution of enterprise AI: from a single chatbot to an ecosystem of specialised agents that collaborate with each other. If you want to see it in action, request a demo.