Skip to content
AI Lab
Contact

AI systems I've built

Working versions of the AI features I build for clients: assistants, agents, retrieval, document processing, voice and automation. Open any card to try it.

Discuss an AI projectOpen Portfolio AI

8 interactive demos. They run in your browser on sample data, so nothing you type leaves the page.

AI Assistant

A chat interface that answers questions about my work, with tables, charts and project references.

Capability: LLM integration and chat UX

  • Streaming
  • Tool calls
  • Structured output
search
read
write

AI Agent

An agent that plans, calls tools and reports each step as it works through a task.

Capability: Planning and tool use

  • Tool calling
  • State machine
  • Human approval

RAG

Ask a question over a document set and see the retrieved passages, scores and a cited answer.

Capability: Retrieval-augmented generation

  • Embeddings
  • Vector search
  • Citations

OCR

Drop an invoice and watch text regions get detected and read, field by field.

Capability: Text recognition

  • Layout detection
  • OCR
  • Post-processing

Voice AI

A voice agent session with live transcript, intent detection and a booking tool call.

Capability: Realtime voice

  • Speech-to-text
  • LLM
  • Text-to-speech

Document Intelligence

Classify a document, extract fields with confidence scores, and route low-confidence fields to review.

Capability: Extraction and validation

  • Classification
  • Extraction
  • Human-in-the-loop

Workflow Automation

A trigger-to-action workflow that runs step by step, with AI deciding the route.

Capability: Event-driven automation

  • Triggers
  • AI routing
  • Retries

AI Analytics

Ask a question in plain English and get the query, a chart, and a written summary.

Capability: Natural language to insight

  • Text-to-SQL
  • Charting
  • Summaries

Have an AI use case?

Documents to read, questions to answer, or a workflow that eats hours each week. Tell me about it and I'll suggest where AI fits and where it doesn't.

Describe your use case

How I ship AI features

What turns a good demo into something you can put in front of customers.

  • Evaluations

    A test set of real questions, scored before and after every prompt or model change.

  • Guardrails

    Schema-checked outputs, scoped tools, and a clear refusal when a request is out of bounds.

  • Observability

    Traces for every model and tool call, with latency, tokens and cost per request.

  • Human in the loop

    Low-confidence results and risky actions wait for a person instead of going straight through.

  • Cost control

    Caching, routing simple requests to smaller models, and token budgets per feature.