An analyst with a navigator’s instincts.
A boutique, AI-first data & analytics assurance practice — and a named human who signs off on every number.
The firm
We are deliberately small, and built around one conviction: the crest of the AI capability wave is where the value is. Models, harnesses, skills, plugins, agents, the landscape reinvents itself monthly, and the techniques for extracting real value from token work moves just as fast. Our practice is riding that crest: staying informed, testing every new capability against real analytical problems, and folding what works into how we deliver so our clients don’t have to chase it themselves.
We sell solutions, expertise and guidance, and a standing commitment: align with AI·BI and you operate at the cutting edge of AI capability for knowledge work as it develops, not as it stood the day you signed.
Our deep expertise is the Anthropic Claude platform. Claude.ai, Claude Cowork and Claude Code: agents, skills, plugins, MCP integrations, multi-agent orchestration and autonomous, semi-supervised execution through the Claude subscription (rather than the API) — utilising routines, loops, skills, context engineering, persistent memory, remote control and dispatch to orchestrate your team of Claude agents to optimise productivity and capability.
What an agent knows the moment it starts is an engineering problem, not a prompt — and it’s where most of our development effort goes. Most users point a model at a question; we provision it for one.
The right context, at the right time. Provisioning the agent with the context it needs to operate reliably and accurately. Most agent mistakes are assumptions made to cover missing context, or an ask disproportionate to the available token budget — without appropriate headroom, and without staging that applies real thinking at each phase of the work, shortcuts get taken. Context engineering is providing the conditions for the model to operate at its best, at all times.
The principal
James Furness
PrincipalReads the work, applies judgement where the data is ambiguous, and puts his name on the figure that reaches you.
- Enterprise ERP transformation
- Lakehouse platform build
- Executive & board reporting
- Demand forecasting & MRP tooling
- Data governance
- Power BI
- SQL
- Python
- Databricks
- Azure Data Factory
- Excel — Power Query & VBA
- Claude Code
- MCP integration
- Plugin architecture
- AI governance
- Workflow automation
- Steering-committee delivery
- Stakeholder engagement
- Change management
- Team leadership & mentoring
Rob — the AI analyst
Rob
AI analystBuilt for deep, multi-session work, utilising the best-practice techniques to conduct research and analysis at the cutting edge of AI capability. Rob does the rigour at a pace a human team can’t match; every figure still crosses the principal’s desk before it reaches yours.
- Multi-session engagements
- Multi-agent research sweeps
- Warehouse-direct SQL analysis
- Brand-as-code deliverables
- Bespoke agent & plugin builds
- Two-tier persistent memory
- Decision & learning registers
- Domain dictionaries as skills
- Scoped briefs per subagent
- Session-to-session handovers
- Adversarial QA gate
- Human sign-off on every figure
- Claude (Anthropic)
- Claude Code
- MCP connectors
- File-based persistent memory
Bring us your messy numbers.
Tell us the decision you’re trying to make. We’ll tell you what the data can — and can’t — support.
Book a 20-minute call →