Custom AI Agents
Claude-powered agents that reason, call your tools, and take real action. RAG assistants, support triage, research workers. Built on your data and wired into production.
Typical: 1–3 weeks
AI engineering studio · solo founder
AI agents you can inspect before you pay
I build custom AI agents and automation workflows, RAG assistants, support triage, lead routing, wired into your existing stack with Claude, n8n, and Supabase. Every build is open-source. Read the code and watch it work before you ever book a call. You deal with the engineer writing it, not an account manager.
Available for select projects · 2026 · usually replies same day
3 demos
open-source repos on GitHub
1–3 wks
kickoff to deployment
Direct
founder access, no middlemen
About
The engineer who builds it, on demand
Vertrez is a one-person studio that builds AI agents and automation workflows that actually ship to production. Not demos, not decks, just working systems wired into the tools you already run. You talk directly to the engineer doing the work, every step. Most builds go from kickoff to deployment in 1 to 3 weeks.
Capabilities
Anthropic Claude APIOpenAI APIMulti-agent systems
RAG & vector search (pgvector)Tool-use orchestrationNode.js & TypeScript
n8n workflow automationSupabase & PostgresSlack & Telegram bots
Webhook & CRM integrationsAsync background jobsVercel deployment
What I build
Built on your data, shipped into your stack
Workflow Automation
Kill the manual busywork. Webhook-to-CRM routing, lead enrichment, Slack and Telegram alerts. Automations that run reliably in the background so your team doesn't have to.
Typical: 3–10 days
Internal AI Tools
Private tools for your team: knowledge assistants over your docs, AI-assisted dashboards, and operations bots that plug into your existing systems. Not another tab to babysit.
Typical: 1–2 weeks
AI Engineering Retainer
Ongoing capacity to ship, tune, and extend your AI systems. Direct founder access, fast turnaround, and a roadmap that keeps moving. No agency overhead.
From 1 month, ongoing
Most projects land between $1.5k and $4k, fixed-quoted in a written proposal after a free intro call. You see the exact scope and price before you commit anything. No hourly meter, no surprise invoices.
Selected work
A look at what I ship
Three real builds, each open-source with a working demo. Click a card to expand it, play the demo and read the code.
How it works
From first call to handoff
- 01
Intro call
15 minutes to scope your use case and confirm fit.
- 02
Proposal
Within 48 hours, a written proposal with approach, timeline, and price.
- 03
Build sprint
Async daily updates while I ship. You see progress, not slides.
- 04
Handoff
Deploy, document, hand over. Optional retainer for iteration.
No big logos yet. Just receipts.
Trust the code, not the pitch
Vertrez is a new, solo studio, so I don't ask you to trust a wall of logos. I ask you to read the source, watch the demos, and judge the work directly. Every case study ships as an inspectable repo, and you talk to the engineer who wrote it from the first message to deployment.
Open-source demos you can inspect
The DocsBot, Triage Agent, and Lead Router case studies are real, runnable repositories on GitHub. Clone them, read the orchestration code, run them locally. Nothing behind a curtain.
Working demo videos, not slideware
Each build comes with a real screen-recorded demo of the system doing the actual job: RAG answering grounded questions, the agent classifying and routing a live ticket. You see the behavior before you book.
You talk to the engineer who builds it
No account managers, no handoffs, no offshore team you never meet. The person on your intro call is the person writing your code and deploying it. Direct access from kickoff to production.
Production focus, fast response
Most builds go from kickoff to deployed in 1–3 weeks, integrated into your existing stack: Node, TypeScript, Supabase, n8n, Vercel. Async by default, quick to reply, honest about scope.
What you can verify right now
No client testimonials yet, so instead, here's proof you can check yourself today. Open the repos, read the code, watch the demos.
Before you book
Straight answers to what you're probably wondering
Because no two builds are the same, and a one-size-fits-all price would either overcharge simple projects or under-scope complex ones. After our intro call I scope the work and send you a written proposal with a fixed quote for that specific build, usually somewhere in the $1,500–$4,000 range. You see the number and what it covers before you commit a cent. No hourly meter, no surprise invoices.
Most agencies sell you a senior engineer in the pitch and quietly hand the build to a junior. With me, the person you talk to is the person who writes every line. No account managers, no telephone game, faster decisions.
And because Vertrez is new, I don't ask you to take my word for anything. My demos are open-source on GitHub, so you can read the actual code before we ever talk. When the build is done, you own the code, fully documented, so you're never locked to me or anyone else. Solo doesn't mean fragile. It means you keep full control.
You do, completely. On delivery the code, prompts, workflows, and config are pushed to your repo under your account, and the system runs on your own accounts and API keys, not mine. Nothing in the live path belongs to me. No license to keep paying for, no platform only I can touch. It's handed over fully documented, so if we ever part ways you walk away with a system any competent engineer can pick up and maintain.
Your keys live in your accounts and your own environment variables, never hardcoded, never stored on my side. I build inside your stack (your Supabase, your Vercel, your cloud), and when the agent calls an LLM it runs through your own OpenAI or Anthropic account, so customer data stays under your vendor agreement and isn't used to train anyone's models. During the build I work against test data, never your real customer records, and any access I'm given is scoped and revoked at handover. I'll sign an NDA up front, and a DPA if you're handling personal data, and we scope exactly what the agent can and can't touch.
Most builds run one to three weeks, from kickoff to a tested, deployed system, not just something that runs once in a demo. That speed comes from focus and tight scoping up front, not from cutting corners. The exact timeline depends on scope and how fast I get access and feedback from your side, and I work on only a few projects at once so yours actually moves instead of sitting in a queue. After the intro call you'll get a realistic timeline tied to your specific build, and I stay on it through the first weeks of real use, not gone the moment it goes live.
Every build ships with a proper handoff: documentation, a walkthrough, and 30 days of post-launch support to fix anything that isn't working the way we scoped it. Changes or new features beyond that scope are a separate fixed quote, so you always know what you're paying for. Because you own the code and it lives in your stack, you're never hostage to me. If you want me back later for new work or iteration, you hire me for it, fixed-quoted like everything else. Your option, never a dependency I engineer in.
Then I'll tell you on the call, and I won't take the project. The call is a scoping conversation, not a sales pitch. If your problem is better solved by an off-the-shelf tool or another specialist, I'll say so and point you that way. I'd rather pass on a bad-fit build than ship something that doesn't earn its cost.
Get started
Book your 15-minute intro call
Pick a date in the calendar and the available times appear once you choose a day. We scope your use case and confirm fit. No pitch deck, nothing to buy.
Opening the booking calendar…
Open cal.com/vertrez ↗Prefer async?
Send a quick note instead
A few sentences is enough. I'll reply, usually the same day.