

Agent Memory Hack Night | Mastra x Elasticsearch
Build AI Agents That Remember
Join Mastra and Elastic for a hands-on hack night exploring agent memory.
Most AI agents start every conversation from scratch. They forget what a user told them, repeat suggestions that already failed, lose track of progress, and sometimes rely on outdated information instead of what happened most recently.
At this hack night, you’ll build an agent that can use information from the past to make better decisions in the present.
Your agent might remember:
A user’s preferences
What has already been tried
Progress toward a goal
Previous decisions and their rationale
Important events or changes
What should happen next
You’ll also explore an important memory challenge: not everything from the past should be treated equally. Recent information may be more relevant than older information, and a new decision may need to replace or outweigh an outdated one.
After a short introduction and demo, we’ll spend the rest of the evening building.
The stack
Mastra
Mastra is an open-source TypeScript framework for building AI agents. It handles the agent loop, tools, memory, and workflows, so you can focus on what your agent should do instead of wiring up LLM calls by hand.
Elasticsearch
Elasticsearch provides the retrieval and memory layer. You can use it as Mastra’s native vector store or explore hybrid search, which combines keyword and semantic matching to help your agent find the right context.
What can you build?
Build anything that becomes more useful when it can access relevant information or remember what happened before.
A few ideas:
A learning coach that remembers a student’s progress
A customer-support assistant that knows what has already been tried
A research agent that tracks sources, findings, and open questions
A travel planner that remembers preferences and past feedback
A project assistant that recalls decisions and next steps
A creative collaborator that maintains continuity across sessions
Bring your own data, use one of our starter datasets, or begin with one of the examples provided.
Developers, data practitioners, designers, product builders, and first-time agent builders are all welcome. No prior experience with Mastra or Elasticsearch is required.
There will be food, drinks, prizes, and plenty of people available to help.
Bring your laptop and your ideas. We’ll provide the tools and starter resources.
How it works
5:00 PM - Doors open
Grab food and drinks, meet other attendees, and get settled.
5:30 PM - Kickoff and demo
We’ll introduce the challenge, walk through the tools, and show a working example.
5:45 PM - Build time
Choose an idea and start building.
Technical? Great.
New to coding or AI agents? Also great. We’ll help you get set up and provide starter projects so you can begin quickly.
8:00 PM - Show and tell
Share what you built with the room, no matter how finished it is. After the demos, we’ll award prizes to the top three projects.
FAQ
Where is the event?
This is an in-person event at the Elastic office in San Francisco:
33 New Montgomery Street, 9th Floor
San Francisco, CA 94108
Who can participate?
Everyone. All skill levels and backgrounds are welcome.
Do I need to bring my own data?
No. You may bring your own dataset, but we’ll also provide starter data and example projects.
Please do not bring confidential, sensitive, or proprietary data unless you are authorized to use it.
Can I start my project before the event?
You may think about ideas beforehand, but all project implementation must take place during the hack night.
Do I have to attend in person?
Yes. This event is in person only. Please RSVP if you plan to attend.