What I build
- AI assistants that answer questions inside your app or platform.
- AI agents that follow a role and a goal. In Her Closing AI, agents play clients with their own objections, so sales teams can practise conversations.
- Search over your own documents (RAG), so answers come from your data, not from the model’s guesses.
- Workflow automation: summaries, classification and drafts that save your team time.
How I approach it
- The right model for each task. Hosted models via API, such as OpenAI, Claude or Gemini, or open-source models on your own servers, chosen by cost, quality and privacy.
- Tested before launch. I check AI features on real examples and set up quality checks, so you know how the feature behaves.
- Your data stays yours. Providers and settings that don’t train on your data, with hosting in the EU on request.
- GDPR and AI Act in mind. I document where data goes, how long it’s kept and what users are told.
When AI isn’t the answer
Not every feature needs a language model. If a simple rule, a search index or a better form solves the problem, I’ll say so. It’s cheaper to run and easier to trust.
Price
AI features are part of a product build or a monthly partnership. Projects start from $3,000, with a written estimate after the first conversation.
