Custom AI agents
Connect language models to approved tools and information for a specific business task.
Services AI agents & automation
Less repetitive work. More room to think.
AI agents and connected workflows that help your team research, organize, generate, and act, with clear boundaries and human review where it matters.
Direct collaboration. Clear scope. Thoughtful delivery.
AI agents & automation
What I can build
Built around
your next move.
Workflow automation
Connect forms, spreadsheets, email, and APIs with n8n or custom Python services.
Research & lead preparation
Collect available business information, remove duplicates, and organize records for review.
Content & reporting systems
Turn structured inputs into drafts, summaries, and recurring performance reports.
Who this is for
A good fit
for your team?
- Teams repeating the same research or reporting tasks
- Agencies connecting multiple client tools
- Businesses exploring a clearly scoped AI use case
Problems I solve
Start with the friction.
Build the right fix.
Manual work keeps repeating.
Identify the repeatable steps and connect them into an observable workflow.
Your tools do not talk to each other.
Use APIs and webhooks to move data between the systems you already use.
AI output needs practical guardrails.
Use structured outputs, validation, error handling, and approval steps where the task calls for them.
What is included
A complete handover.
A clear way forward.
We agree on the exact deliverables before work begins. Here is the foundation we build from.
- Workflow mapping and feasibility review
- Agent or automation implementation
- Agreed API, tool, and data integrations
- Input validation and structured outputs
- Error handling, logging, and test scenarios
- Workflow exports or source code with operating notes
Development process
From first conversation
to a confident handover.
- 01
Map the task
Understand the inputs, decisions, outputs, and current manual work.
- 02
Prove the workflow
Test a small version against representative inputs before expanding it.
- 03
Connect & safeguard
Add integrations, validation, failure handling, and review steps.
- 04
Deploy & hand over
Run agreed scenarios and explain how to operate and monitor the system.
Technologies used
The right tools
for the job.
Selected for your requirements, integrations, and long-term maintainability.
- Python
- FastAPI
- OpenAI Agents SDK
- MCP
- n8n
- Supabase
- Docker
Relevant personal project
AI Lead Generation & Outreach
The AI Lead Generation & Outreach System is a personal automation project that discovers businesses, extracts available contacts, and organizes leads for outreach preparation.
Explore the personal projectPricing
The right starting point.
Room to grow.
Start with one clearly defined workflow or agent task. Pricing depends on the number of integrations, decision steps, and review requirements.
Starter
Automate one repetitive task with a clear outcome.
Starting from
$299USD / project
- One simple automation workflow
- Up to two connected tools
- Basic validation and error handling
- Workflow handover and operating notes
Growth
Most PopularA smarter workflow connecting your tools and data.
Starting from
$899USD / project
- Multi-step workflow or focused AI agent
- Up to four agreed integrations
- Structured outputs and validation
- Human approval steps where needed
- Testing, deployment, and handover
Custom
For complex agents and automation across your business.
Tailored to your scope
Contact for quote
- Multiple agents or connected workflows
- Custom tools and MCP integrations
- Monitoring and evaluation requirements
- Phased delivery and ongoing support options
Indicative starting prices in USD. Final pricing and deliverables are confirmed in your proposal. Model usage, API subscriptions, and workflow hosting are separate operating costs. These are reviewed during scoping.
Before we begin
A few useful answers.
Do I need AI for every automation?
No. Straightforward rules and API integrations are often enough. AI is useful when a step requires interpreting or generating unstructured content.
Can the agent take actions without approval?
That depends on the task. We define the permitted actions together and can require human approval before messages, updates, or other consequential steps.
How do you handle unreliable AI output?
I use representative test cases, output validation, and review or fallback paths. AI can still make mistakes, so the workflow is designed around that limitation.
Can you connect my existing tools?
Usually, if they provide supported APIs, webhooks, or suitable connectors. Access requirements and integration limitations are checked before the scope is agreed.
AI agents & automation
Which repetitive task should we simplify first?
Tell me where you are today and what you want to make possible.
Explore all services