// services / ai agents
AI agents that run in production, not just demos
Agents, RAG systems and agentic workflows, designed, built and hardened by someone who runs AI in production at his own company. Fixed price after scoping, from $6,000. Most projects land between $15k and $35k.
// what breaks agents
- Tool calls that fail silently
- LLM costs nobody tracks
- Context windows stuffed to the brim
- No evals, so no idea if it got worse
// what i build
Five kinds of AI systems
Autonomous agents
Custom agents that perceive, decide and act. They handle complex multi-step workflows without someone watching every step.
- MCP
- Tool use
- Evals
RAG systems & co-pilots
Retrieval pipelines that give your AI grounded, accurate answers from your own data.
- Vector search
- Vertex AI
- Firestore
Agentic workflow automation
Connects your CRM, databases, APIs and communication tools into pipelines that run themselves.
- n8n
- MCP
- Go
LLM integration
OpenAI, Claude, Gemini or open models added to your existing product, with function calling, tool use and streaming.
- Claude
- OpenAI
- Gemini
- Vertex AI
Multi-agent systems
Specialized agents working in parallel: task decomposition, routing, error recovery and observability.
- Guardrails
- Observability
- Cloud Run
// how i work
I don’t hand off a spec and disappear
I architect, code, review and ship alongside you. Every engagement starts by defining outcomes, not just features.
- 01
Discovery call
Your workflow, data and goals. Free, 30 minutes.
- 02
Scoping document
Architecture, tech choices, timeline and a fixed price.
- 03
Build
Iterative delivery with weekly check-ins.
- 04
Handover
Documented, tested and deployed. You own all of it.
// project investment
Fixed price after scoping
No hourly billing, no surprises.
start here · 1 week
Agent Blueprint $1,500
Before you commit to a build: we map the workflow and data, I design the architecture and pick the models, and you get a written blueprint with a fixed quote. Credited in full toward the build.
| Project type | Scope | Investment |
|---|---|---|
| Single-agent build | One workflow, one or two integrationsSupport bot, document processor, single-task automation | $6,000 – $12,000 |
| Multi-step agentic system | CRM/API integrations, RAG pipeline, evalsKnowledge assistant, lead qualification, workflow orchestration | $15,000 – $35,000 |
| Production multi-agent system | Multiple agents, observability, guardrails, scaleMulti-agent systems, complex integrations, monitoring and cost control | From $35,000 |
| Monthly retainer | Maintenance & iterationMonitoring, improvements, agent tuning, cost control | $2,000 – $5,000/mo |
I take on 2–3 AI agent projects per quarter.
// common use cases
Where agents pay off
- Customer support agent with escalation logic and CRM integration
- Internal knowledge assistant trained on your documentation
- Lead qualification and outreach automation
- Data extraction from unstructured documents
- An AI co-pilot embedded in an existing SaaS product
- An operational agent that replaces a recurring manual workflow
// faq
Questions
How is pricing set?
Every project is fixed-price after a scoping document. No hourly billing, no surprises.
Which models do you use?
I’m LLM-agnostic. I recommend the right model for the use case, and often a cheaper one for routing and classification.
Do you handle the infrastructure around the agent?
Yes: APIs, databases, auth, deployment and monitoring. Most agents fail in that layer, not in the model.
Can we start smaller?
Yes. The one-week Agent Blueprint ($1,500) gives you the architecture and a fixed quote, and it’s credited in full if you go ahead with the build.
How many projects do you take on?
Two or three AI agent projects per quarter. If your timeline is within 60 days, reach out now.
Ready to build something that actually runs?
If your timeline is within the next 60 days, reach out now.