Cover Image for Build Day: Agent Harness & Memory [Build Club x Mem0]
Cover Image for Build Day: Agent Harness & Memory [Build Club x Mem0]
Avatar for Build Club
Presented by
Build Club
The most collaborative AI community in the world (50+ Cities, 30K+ community)

Build Day: Agent Harness & Memory [Build Club x Mem0]

Register to See Address
Registration
Approval Required
Your registration is subject to host approval.
Welcome! To join the event, please register below.
About Event

[invite only, register for approval]


what this is

the bottleneck in agentic AI right now isn't the model- it's the harness and the memory.

the loop the agent runs in, the tools it can reach, the context you feed it on every turn, and whether any of that state survives past a single session or gets thrown away the second the process exits.

SO, lets spend a day on this. the event will run as a few live walkthroughs from people actually building this stuff, then a long sprint where you build your own, then demos.

the theme is agent harnesses and memory, but it's an anchor, not a fence. bring what you're working on in agentic space.

this event is hosted in collaboration of Build Club x Mem0


how the day runs

9:30 – 10:30 → doors + coffee
check in, set up, write what you're building today

10:30 – 11:00 → research yap: agent memory & harnesses in 2026
a fast read on where the field actually is- the layers of a real agent harness, state of memory for long running agents, what's holding up in production. enough shared context that the whole room is building from the same map.

11:00 – 12:15 → build walkthroughs/workshops (live, in code) two walkthroughs from the community

harness: an agent loop from scratch, tool-calling, and context management, nothing hidden behind a framework
memory: bolting a memory layer onto that agent so it persists and recalls across sessions

12:15 – 1:45 → build sprint
the main block, and the actual point of the day. extend a walkthrough, bring your own project, or take a bounty off the board.

1:45 – 2:30 → lunch (keep building if you're in flow)
food and arguments about harness and memory architecture decisions.

2:30 – 3:15 → demos
open floor for attendees. 5 minutes each
either present your progress from the earlier build session or showcase a demo of your project you have been working on (please fill form below)

3:15 – 3:30 → wrap + next steps
and picturesssss!!


call for builders 🛠️

the stage is open for builders to lead these demos and showcase your projects. there are two type of CFP you can apply for:

  1. lead a build walkthrough - WORKSHOP
    take the room through a harness or a memory layer, live, in code. you don't need a polished talk, you need a working build and the ability to explain why it's wired the way it is.

  2. secondary slots:
    - research yap: 10 minutes on the state of agent memory and harnesses in 2026
    - demo: show something you built around agents, harnesses, or memory. doesn't have to be from the day. any sharp build counts.

submit → https://form.typeform.com/to/ShgbP5Ao

[STRICTLY NO SLIDES] (give a walkthrough)


bounty board

if you don't have your own thing to build, take one of these. each is scoped to ship inside the sprint, and each has a catch that makes it actually worth doing:

  1. survive a restart. take a stateless CLI agent and give it memory that persists across process restarts

  2. the harness, by hand. build the agent loop from scratch: model call, tool call, parse, repeat.

  3. memory that actually gets used. store user preferences and retrieve plus apply them in a later session, not just log them.

  4. short vs long-term. run working memory (this session) and persistent memory (across sessions) in one agent and show the handoff between them cleanly.

  5. break it on purpose. poison or corrupt an agent's memory and show exactly what breaks downstream.

  6. context budget. add summarization or compaction so an agent stays coherent well past its context window instead of falling over. measure tokens before and after.

  7. tool router. give an agent 10+ tools and a routing layer so it picks the right one without drowning in schemas. show it degrade gracefully when it's unsure.

  8. prove the memory works. build a tiny eval: a multi-session task an agent can only pass if it remembered something from an earlier session.

just some inspirations, more on the wall on the day.


who's this for

builders.
learners.


bring

- laptop, charged, plus charger your stack of choice
- an LLM API key (BYO, or we'll sort a few on the day)
- a project to build, or an open mind for the bounty board


LET'S BUILDD :)

Avatar for Build Club
Presented by
Build Club
The most collaborative AI community in the world (50+ Cities, 30K+ community)