AI that connects to real tools
Custom MCP servers, agent workflows and API integrations bring approved data and actions into AI tools, with permissions and access boundaries considered from the start.
Available for new projects
I make systems work together.
I'm Tomi Šeregi, an independent software engineer building API integrations, custom MCP servers and workflow automation. I connect your tools, fix what breaks and deliver tested, documented code.
Send a briefWhat we do
build
Broken integrations, AI tools that need the right connections, or search issues you cannot pin down? I take on focused engineering work with a clear scope and handover.
API integrations & webhook repair
I build API integrations and troubleshoot broken webhooks, OAuth flows, failed syncs and unreliable data handoffs. The fix includes error handling, tests and useful logs—not just a connection that works once.
- API and webhook debugging
- Retries and reliable data sync
- Tests and handover notes
Custom MCP servers & AI tool integration
Custom MCP server development for ChatGPT, Claude and other compatible AI clients. I connect approved tools and data with scoped permissions, input validation and checks you can run again.
- MCP tools and resources
- Authentication and access boundaries
- Integration tests and documentation
Technical SEO & AI search audits
I review crawlability, indexing, canonicals, structured data and technical issues holding pages back. For AI search visibility, I examine how content is represented and cited, then prioritize fixes based on evidence.
- Technical SEO audit
- AI search visibility review
- Prioritized fix list
- TypeScript
- Python
- Node.js
- React
- Cloudflare
- Docker
- GitHub
- Vercel
- Django
- Figma
- WordPress
- Grafana
- Prometheus
- Search Console
- Puppeteer
Our mission
Less busywork. More reliable systems.
I build practical AI automation, API integrations and custom MCP servers that remove manual handoffs and keep workflows dependable. When search visibility stalls, I investigate technical SEO issues and turn the findings into tested, maintainable fixes.
Why work with me
AI-first engineering. Built beyond the demo.
I connect AI to the systems people already use, design for real failure cases, and make the result understandable to the people who maintain it.
Built for the unexpected
APIs change, tokens expire and webhooks fail. I address these failure paths with input validation, meaningful errors, tests and recovery strategies suited to the project.
Direct work. Clear handover.
You work with the engineer writing the code. I explain technical decisions, keep scope visible and provide documentation that makes the next change easier.
Frequently asked questions
A few things worth knowing.
Straight answers about AI automation, integrations and how I work.
What business processes can AI automation handle?
Good candidates include repetitive request routing, document triage, data handoffs and research workflows. I look for clear inputs, measurable outcomes and steps that need human review before automating them.
Explore AI automationCan you build custom MCP servers and AI agent integrations?
Yes. I build custom Model Context Protocol (MCP) servers that connect approved tools, data and actions to compatible AI clients. Authentication, access boundaries, validation and testable behaviour are part of the design.
Explore AI integrationsCan AI agents connect securely to our APIs and company data?
That depends on the systems and access available. I design integrations around scoped credentials, input validation, explicit permissions and human approval where needed. We identify sensitive actions and data before enabling them.
Can you help with technical SEO and AI search visibility?
I investigate crawling, indexing, canonicals, structured data and technical barriers, then prioritize fixes that can be verified. I can also review how your content is represented in AI search, without promising rankings or citations.
Explore search visibilityHow much does a custom AI or API integration cost?
It depends on the systems involved, available APIs, access requirements, edge cases and testing. Share a brief and I can review the scope before proposing deliverables and a project price. There is no one-size-fits-all package.
Who builds the solution, and what happens after delivery?
You work directly with the engineer writing the code. The agreed handover can include tests, setup instructions and technical documentation. Any ongoing maintenance is discussed separately rather than assumed.
Can you fix broken webhooks or existing integrations?
Yes. I can trace failed requests, expired credentials, retries, unexpected payloads and unreliable data syncs. The aim is to reproduce the failure, correct its cause and add checks that make regressions easier to catch.
Do I need to replace my current software to use AI automation?
Often not. Existing tools may support APIs, webhooks or exports that let us build around your current workflow. If a system has no practical integration path, I explain the trade-offs before suggesting a replacement.
How long does a typical integration project take?
A small, well-defined connection is different from an agent with multiple systems and approval steps. I estimate the timeline after reviewing the requirements, dependencies and unknowns rather than promising a universal deadline.
What should I include in my project brief?
Describe the problem, the tools involved, what a successful result looks like and any access or security constraints. Examples of failed workflows and an ideal timeframe help, but a short initial message is enough to start.