Obsidian Research Group
We build AI native applications for the next era of work.
Check out our latest applications and learn about our services below.
Trusted partners & technologies
Custom AI Systems
From autonomous agents to intelligent workflows, we build AI applications optimized for your specific use case.
Learn moreAI WorkflowsSeamless Integrations
Connect AI capabilities to your existing stack. LLMs, vector databases, and ML pipelines that just work.
Learn moreAI InfrastructureEnterprise Scale
High-performance data infrastructure designed for real-time AI inference, training, and analytics.
Learn moreThree pillars of AI excellence
We operate through three distinct pillars, each focused on a critical stage of making AI practical, scalable, and valuable.
Obsidian Arrow
AI Data Structuring
The foundation layer that prepares enterprise data for AI. We structure, encode, and optimise complex datasets so they are ready for analytics, machine learning, and AI applications.
- Data structuring and feature engineering for AI readiness
- Spatial, temporal, and vector encoding for AI workflows
- Scalable data pipelines supporting analytics and inference
Obsidian Development
AI Software & Systems
Custom AI software and systems built for real-world deployment. We design, engineer, and integrate AI solutions tailored to enterprise, infrastructure, and complex operational environments.
- Bespoke AI applications and decision-support tools
- AI-enabled platforms, dashboards, and workflows
- Secure, scalable systems from prototype to production
Obsidian Research
Advanced Analytics & Intelligence
Advanced analytics and multi-domain data science delivering insight, foresight, and strategic intelligence across spatial, economic, and behavioural systems.
- Advanced data science across multiple domains
- Geospatial intelligence and custom analytics
- Scenario modelling, forecasting, and machine learing
Your data, wired into AI
We transform enterprise data into AI-ready knowledge. We design and deploy vector stores, RAG pipelines, and MCP servers that connect your data directly to LLMs, agents, and applications.
From raw data to production-grade AI systems — securely and at scale.
1# rag_agent.py2from obsidian import VectorStore, RAGAgent, MCPServer34# Build a vector store from client data5vector_store = VectorStore.from_sources(6 sources=["data_lake/", "documents/", "geospatial/"],7 embedding_model="obsidian-embed-v1"8)910# Expose data via an MCP server11mcp = MCPServer(12 vector_store=vector_store,13 permissions="read_only"14)1516# Create an AI agent trained on your data17agent = RAGAgent(18 llm="gpt-4.1",19 knowledge_source=mcp,20 tools=["search", "analysis", "reasoning"]21)2223# Embed the agent directly into your application24response = agent.query(25 "Summarise key spatial and behavioural patterns for this site"26)Tools we're building
Beyond client work, we invest in building our own AI-powered products that push the boundaries of what's possible.
Industry leaders trust Obsidian
Obsidian's AI infrastructure allowed us to deploy models 10x faster than our previous solution. Their team truly understands enterprise scale.
Head of Engineering
Fortune 500 Tech Company
The custom AI agents they built transformed how we process data. What took days now happens in real-time.
CTO
Leading FinTech Startup
Working with Obsidian felt like having an elite AI research team in-house. They delivered exactly what we needed.
VP of Product
Global Analytics Firm
Ready to build something?
Whether you need custom AI solutions, data infrastructure, or strategic AI consulting—we're here to help.
Start a Conversation
