cd /news/ai-agents/my-ai-agent-rejected-90-of-its-own-i… · home › topics › ai-agents › article
[ARTICLE · art-173495] src=dev.to ↗ pub= topic=ai-agents verified=true sentiment=↑ positive

My AI Agent Rejected 90% of Its Own Ideas, and That's Why It Won 🏆

A developer at TAMU Datathon 2026 built a self-improving agent that learns plain-English if-then rules from examples and only keeps a change after it proves itself on unseen data, finishing 3rd at the event. The agent runs validation in rounds, freezes itself when a round stops paying off, and routes its own uncertain failures to a teacher model that writes reusable "skills" — more than half of which made results worse and were automatically discarded. No model weights were modified; the agent's knowledge stays as readable text and settings, with teammates Nidhi, Mihir and Ravi credited.

recommendation

Validate each self-generated rule or skill on unseen data before keeping it, and auto-discard any that don't measurably improve results.

more in playbook: Evals and testing for LLM apps
by read2 min views1 publishedNov 20, 2026

This is a submission for the MLH x DEV Writing Challenge Everyone wants AI that improves itself. Almost nobody talks about the scary part: an agent that changes itself can just as easily make itself worse, and confidently tell you it got better.

At TAMU Datathon 2026, I built an agent around one rule:

No change is kept until it proves itself on data it has never seen.

It finished on the podium, taking 3rd place. Here's how it works.

Instead of hiding everything inside a black box, the agent learns plain-English if-then rules from examples. Then it puts every rule on trial:

It runs this in rounds, and here's my favorite part: it decides when to stop. Once a round no longer earns its keep, the agent freezes itself. No babysitting.

The result is a rulebook a human can actually open, read and argue with.

Some decisions need real reading comprehension, not pattern-matching. So the agent looks at where its own rules are unsure and keep failing, and hands those mistakes to a teacher model. The teacher writes a short, reusable "skill," and the base model uses that skill to double-check only the shaky answers.

Then the same trial happens again: a skill survives only where it measurably helps. In more than half of the places I tried, the AI made things worse, so the agent threw those skills away automatically.

No model weights were touched. Everything the agent "knows" is readable text and settings it learned and tested itself.

TAMU Datathon 2026 was pure energy from the opening minute: whiteboards filling up with half-baked ideas, people huddled over laptops sketching architectures, and that moment of "wait… what if the agent checks its own work?" that turned into this whole project. The brainstorming was easily the best part: chasing an idea, watching it fail, and coming back with a better one a few hours later.

Then came the late-night chaos: Hearing "3rd place" at the end made every one of those 3am bugs worth it. 🏆

Kudos to my teammates Nidhi, Mihir and Ravi, and huge thanks to the TAMU Datathon organizers for an amazing event!

── more in #ai-agents 4 stories · sorted by recency
── more on @tamu datathon 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/my-ai-agent-rejected…] indexed:0 read:2min 2026-11-20 · —