AI that actually fits your business —
built by someone who ships.
I design, build, and deploy AI for small businesses: Microsoft Copilot in your tenant, Claude inside your workflows, and autonomous agents that turn your data into decisions.
Four ways we plug AI into your business
Pick a starting point — we deliver the first working version in under three weeks.
MS Copilot Integration
Custom Copilot agents, Graph connectors, and Power Platform plumbing — deployed inside your Microsoft 365 tenant.
See Copilot workClaude Integration
Production Claude apps with tool-use, agent SDK, prompt caching, and a hardened evaluation harness — not a chat demo.
Build with ClaudeAgentic Stock Analyst
🎁 Included with every engagementAutonomous market intelligence — live data, technical analysis, and an LLM that explains every call. Clients get 6 months free; prospects get 1 month to explore.
An AI analyst that reads the tape — included with every engagement.
The LumenGuide stock analyst is a Claude-powered tool I built and maintain. It's not a product I sell separately — it's something clients and prospective clients get access to as part of working with me.
- Live data + multi-timeframe technical analysis
- Logged calls with auto-PnL and win-rate tracking
- Conversational interface — ask "why" and get an answer
- Self-hostable; your data never leaves your tenant
Embed Microsoft Copilot and Claude in days, not quarters.
We build Copilot Studio agents for your existing M365 stack and ship Claude-powered apps with tool use, agent loops, and prompt caching. Same team, two frontiers — chosen per use-case, not per vendor pitch.
- Copilot Studio agents on your tenant data
- Graph + Power Platform connectors
- Claude Agent SDK with retries, budgets, and audit
- Model routing — pick Opus, Sonnet, or Haiku per task
• Initech — $22k · last touch 16d
• Soylent — $61k · last touch 14d
Strategy in week one. Production by week three.
A tight loop, scoped to a single high-leverage workflow at a time.
Discover
30-min call to map your highest-leverage workflow and the data behind it. No deck, no fluff.
Architect
We choose the model (Copilot, Claude, or open-source) that fits cost, latency, and privacy targets.
Build
2-week sprint: working agent, evals, guardrails, and a deployable artifact in your environment.
Operate
Monitoring, weekly tune-ups, and a handover that lets your team own it — or we run it for you.