Aneta Kahleová

AI Tech Lead · Contractor

I love helping people move forward.

Aneta Kahleová

I build production chatbots and autonomous AI agents for teams that need more than a demo — context engineering, end to end.

Why chatbots

The answer is already there. It’s just hard to find.

A production chatbot isn’t a widget on your homepage — it’s an assistant that knows your material, cites its sources, and gives people their time back. Here’s what that looked like across 8 million articles and 60,000 teachers.

Teachers were spending more time searching for materials than preparing lessons. ScioBot changed that — 8 million articles, one question away.

ScioBot — the AI core I co-developed for Czech teachers.

8M

articles

60,000

teachers

3,000

schools

AI Awards

2024

What chatbots solve, and how I build them

What I do

Need a production chatbot?

I design, build, and deploy custom AI assistants — grounded in your data, with evals, guardrails, and a path to production. Not a slide deck. A bot that runs.

Custom AI agents

Workflow bots and customer-facing assistants that plan, route, and execute — with human gates where it matters.

Enterprise RAG

Hybrid retrieval at scale. Chunking, citation, judge loops — so answers stay honest in production.

Data extraction

Playwright pipelines and clean hand-offs into your knowledge base — the layer your bot actually needs.

Live demo

See what a production assistant feels like.

Ask about custom chatbots, enterprise RAG, or what it would take to ship one for your team — then compare it to the slide deck you have today.

Open the demo

Aneta’s progress over the past year has been incredible — leading and mentoring on her projects, and pushing forward technically. The Agentic Search she brought into our chatbots is a top achievement that puts them ahead of the competition. Enormous talent, and a great future ahead.

Jiří Szewieczek

Engineering Manager · OKlab (OKsystem)

Latest writing

Eval as an Input, Not a Dashboard: Building Self-Healing LLM Systems

Most teams treat evaluation as a scoreboard — a number you glance at and feel good or bad about. The frontier idea is to wire the eval back into the system as an input that rewrites it: traces go to independent judges, fixes get proposed automatically, and one human approves. It's AI all the way down, with a single gate that isn't.

Maximum Recall: Why Your Retriever Should Cast Three Nets, Not One

You compressed your knowledge into clean indexes. Now comes the moment everything hinges on: given a question, can the bot actually find the handful of tables it needs among a thousand? Miss the right one and nothing downstream can save you — so at the retrieval step, recall beats precision.

All insights

Let’s build something.

Book a free 30-minute scoping call — or email anytime.

aneta.kahleova@gmail.com