What if your AI agent could remember a mistake from last week, continue an unfinished task today, and use that history to make a better decision tomorrow?
That idea could win $5,000 in the CockroachDB AWS Hackathon. With $8,750 in total cash prizes, the challenge focuses on one of the harder problems in agentic AI: giving agents persistent memory that remains useful beyond a single session.
And this isn’t about building another chatbot that simply remembers your name.
CockroachDB AWS Hackathon at a Glance
| Detail | Information |
|---|---|
| Hackathon | CockroachDB AWS Hackathon |
| Platform | Devpost |
| Challenge | Build AI agents with persistent memory |
| Total Prize Pool | $8,750 |
| First Prize | $5,000 |
| Second Prize | $2,500 |
| Third Prize | $1,250 |
| Core Technology | CockroachDB + AWS |
| Best For | AI agent and agentic AI developers |
The Real Challenge: Build an Agent That Remembers
Most AI demos work perfectly when everything happens inside one conversation.
Real applications have a harder problem.
Imagine a debugging agent that encounters the same error it solved two weeks ago. Starting the investigation from scratch wastes time. A useful agent should be able to retrieve what happened before and use that information in its next decision.
The same problem appears with project agents. If an agent stops halfway through a workflow, can it return tomorrow knowing what was completed, what failed, and what needs to happen next?
That’s where persistent memory becomes more than a convenient feature.
The Winning Idea May Start With Forgetting
If I were approaching this hackathon, I wouldn’t start by asking, “What can I build with CockroachDB?”
I’d start with:
Where does an AI agent forgetting something actually cost the user?
A support agent could lose customer context. A DevOps agent could repeatedly investigate the same incident. A coding agent could rediscover bugs it has already solved. A business agent could lose track of a multi-step workflow.
Those are stronger starting points because memory becomes essential to the product rather than something added just to satisfy the hackathon requirement.
CockroachDB can serve as the persistent memory layer while AWS supports the broader agent environment. But the interesting part isn’t simply storing information. It’s demonstrating that past information changes what the agent does next.
Why I’d Watch This Hackathon
| Prize Money | |
|---|---|
| First Place | $5,000 |
| Second Place | $2,500 |
| Third Place | $1,250 |
But there’s a bigger reason this challenge caught my attention: agent memory is a real problem, not a demo feature.
If you’re learning agentic AI, this is an opportunity to stop building agents that only respond—and start building agents that remember, continue, and improve their next action.
For more details please check Hackathon Details
