Agent, Action! We Want to Make AI Agent Action a Long-Term Commitment

Agent, Action! We have decided to make AI Agent “action” a long-term, fundamental focus.

Date: 1.13

Last Sunday, we did something genuinely exciting together.

A group of people who are deeply committed to one question, whether AI Agents can truly “take action,” came together.

This was the first offline event hosted by Qveris Friends × Origin School × Naughty Labs:

Agent, Action! QverisAI Hackathon

A Goal Often Discussed, but Rarely Completed

For this hackathon, we did not set a complex challenge or restrict participants to any specific industry.

There was only one extremely concrete, even slightly demanding goal:

Within 4 hours, run through a truly executable Agent action loop.

(Not a demo, PPT, or concept.)

In other words:

• Not just “analysis”
• Not just “generating recommendations”
• But actually calling tools, actually executing actions, and actually producing results

This is exactly what Qveris has been working on all along.

Nearly 100 developers, product builders, and entrepreneurs came to the event. If you were there, you probably remember scenes like these:

• Someone staring at the screen, waiting for an API response
• Someone repeatedly adjusting an Agent’s execution flow
• Teams engaged in nonstop, heated discussion

All of it pointed to the same question:

Had the Agent actually completed an action?

Models are getting stronger, and demos are becoming more polished.

But Agents that can truly complete an end-to-end action for people in everyday work are still rare.

The issue is not “can it think?”

It is “can it move?”

By the middle of the event, one thing became very clear:

Almost no one was comparing models, prompts, or tricks.

People were spending their time discussing very specific problems:

Why didn’t this step execute?

Was it a permissions issue, or was the tool not connected properly?

Why did the Agent stop here?

One project was stuck for a long time

simply because one action could not be executed successfully.

And it was in these moments

that many people clearly realized for the first time:

When an Agent cannot run,

it is often not because the model is not good enough,

but because the barrier to interacting with the outside world is too high.

At the end of 4 hours, the teams had completed:

10 AI Agent projects

GitHub commits

At least one complete action chain successfully run

All of the projects had one very clear thing in common:

The Agent had already completed something concrete,

instead of staying at the level of discussing what it could do.

With everyone’s consent, all projects have been fully open sourced and placed under the Qveris Friends GitHub:

👇

https://github.com/orgs/QverisFriends/repositories

This was not only to showcase the results,

but also because we hope they can truly be reused and continue running.

Three awards were selected on site:

• The Best PMF: the award for greatest commercial potential
• Just For Fun: the award for the most interesting project
• Qveris Special Award

These projects did not disappear when the event ended.

The code is still there, the logic is still there, and the building continues.

They are more like real samples of Agent action than one-off showcase pieces.

Qveris Friends wants to keep doing this with everyone.

It is not only a community for people to exchange ideas about AI Agents. It is more like a creation camp:

• Failure is allowed
• Not getting things to run is allowed
• But we have to get into “action”, not only the Agent, but people too

If AI Agents are truly going to become the next generation of productivity tools,

then it will not begin with “being better at talking.”

It will begin with “being better at doing.”

Agent, Action.

For us, this is not a slogan.

It is more like a reminder:

Do not rush to define the future.

First, let AI finish one small thing.

Next, we will:

• Break down the real implementations of these 10 Agent projects
• Share where they got stuck and how they eventually ran
• Continue organizing small-scale, hands-on co-creation sessions

If you also care about how Agents can truly be put into practice,

then

Join us now,

See you next time!

Attached: a photo of the founder of Origin School 👆