AI Agent Development

I've shipped AI agents in production. I can build yours.

AI Agent Development

Most people pitching AI development have never shipped an AI agent in production. I have three. Fix-WP processes WordPress emergency repairs automatically. Hunazo runs a marketplace where AI agents handle transactions in USDC. Those aren't demos - they're live, with real users and real money.

I take 1-2 projects per month · Solo consultant, not an agency

3
Live AI projects
200+
Autonomous agent runs
41
Days documented
15+
Years shipping software

What I've already shipped

Before you hire anyone to build an AI agent, check what they've actually deployed. Here's mine.

Live in production

Fix-WP.com

WordPress emergency repair agent. Customer describes the problem, pays $100 flat, and the agent diagnoses and fixes it autonomously. Real paying customers. Real repairs.

Claude API FastAPI Stripe escrow SSE status stream Python
fix-wp.com
Live in production

Hunazo.com

AI agent marketplace. Agents are listed with semantic search, hired via smart contracts, paid in USDC on Base. Dispute resolution is also handled by an AI agent.

pgvector Mistral embeddings USDC on Base Python SDK FastAPI
hunazo.com
Documented experiment

AI Hustler

I gave an AI agent a VPS and a $0 budget and documented what happened. 200+ autonomous runs over 41 days. The full log is public. I document what works and what doesn't - including the failures.

Claude API VPS automation 41 days 200+ runs
Read the experiment

What I build

In my experience, most AI automation projects fall into a handful of categories. Here's what I've built and can build for you.

Claude / LLM API integration
Tool use, structured output, multi-turn conversations, prompt caching. Same API powering Fix-WP in production since January 2026.
RAG pipelines + semantic search
pgvector embeddings, chunking strategies, retrieval-augmented generation. Hunazo's agent discovery runs on this stack.
Autonomous agent workflows
Agents that process jobs, make decisions, take actions, and report back - without a human in the loop for every step.
Web automation (Playwright)
Scraping, form filling, monitoring, data extraction. When there's no API, a browser agent gets the job done.
API integrations
Stripe payments, webhook handling, third-party data sources. Clean FastAPI backends that don't fall over under load.
WordPress + AI combinations
AI features plugged into existing WP sites. Content generation, classification, moderation, recommendation. 15 years of WP internals knowledge helps here.
Agentic task queues with SSE streams
Background job runners that stream live progress to a UI - so users know what the agent is doing, not just whether it finished.
Data processing pipelines
Ingest, classify, transform, and act on large volumes of data with AI in the decision loop. Repeatable, schedulable, auditable.

How it works

Fixed-price projects only. The discovery call is where we figure out if the scope makes sense - for both of us.

1

Discovery call (free, 30 min)

We talk through what you want to build, what you already have, and what success looks like. I ask a lot of questions. No sales pitch.

2

Written proposal with fixed price

You get a document with scope, deliverables, timeline, and a single fixed price. No hourly billing. No surprises.

3

Build + iterate

I build in chunks and show you working software early. If something needs adjusting, we catch it before it's baked in.

4

Handover with docs

Code, deployment guide, and a walkthrough call. You own everything. If you want ongoing support, we can arrange that separately.

Project-based, not hourly

Hourly billing is bad for clients. You can't budget it. You get billed for dead ends. Your incentives and the developer's incentives are not the same.

Fixed price means you know the number before work starts. If I underestimate, that's my problem. If I finish faster, you still pay the agreed amount - but that rarely happens because I scope carefully.

Typical projects start at $3,000 for a focused automation script or pipeline. Full agents with UI, integrations, and multiple components run $8,000-$15,000. The discovery call is how we figure out which bucket you're in.

Typical price ranges
Automation script / pipeline
Scraping, processing, API integration, scheduled jobs
$3K-$5K
AI feature for existing product
LLM integration, RAG, semantic search, classification
$4K-$8K
Full AI agent product
Agent + UI + integrations + deployment
$8K-$15K

Built it. Shipped it. Running it.

There are a lot of "AI consultants" who've read the docs and watched the demos. Here's what I've actually done.

3 live AI projects in production

Fix-WP, Hunazo, and the AI Hustler experiment. Real users, real transactions, real failures I've had to debug at 2am.

🛠

15 years building production software

Ex-CTO at CreativeMinds (2010-2024). 50+ WordPress plugins shipped. I know how to build things that don't fall over.

Toptal top 3% / Upwork Top Rated

95% job success score. Toptal screens the top 3% of applicants globally. Not self-reported - verified by independent platforms.

🔎

I document the failures too

The AI Hustler experiment is fully public. I wrote up what the agent got wrong, not just what it got right. That's the kind of transparency you want in someone building your critical systems.

Common questions

What stack do you use for AI agent development?
Claude API (Anthropic) as the primary LLM. Python and FastAPI for the backend. PostgreSQL with pgvector for semantic search. Playwright for web automation. Stripe for payments when needed. That's the stack Fix-WP and Hunazo run on in production today.
Do you work on non-WordPress projects?
Yeah. WordPress is part of my background but it's not a requirement. Hunazo has nothing to do with WordPress - it's a standalone Python app with on-chain payments. If you have a Python, Node, or plain web backend, I can work with that.
What's the minimum project size?
Typical projects start at $3,000. Below that the scoping and handover overhead doesn't make sense for either of us. If you have something smaller in mind, book the discovery call anyway - I'll tell you honestly if it fits or not.
How long does a typical project take?
Depends on scope. A focused automation pipeline can be done in 2-3 weeks. A full agent with custom UI, integrations, and a status stream takes 4-8 weeks. I put a timeline estimate in the written proposal so you know before you commit.
Do you maintain the agent after delivery?
That's negotiated per project. Some clients take the handover and run with it. Others want ongoing support. I quote maintenance separately so you're not paying for it if you don't need it.
Can you work with our existing codebase?
Usually yes. I'll need to look at it during the discovery phase to understand what we're working with. If the existing code makes integration risky or expensive, I'll tell you that in the proposal - not after we've started.

Let's figure out if this makes sense.

30 minutes. Free. No pitch deck, no NDAs required to talk. Just a conversation about what you want to build and whether I'm the right person to build it.

Book discovery call