On June 27, 2026, AGI House hosted its Agent Identity Build Day — a strictly in-person hackathon in Hillsborough, California, attended by 256 of the most technically sharp builders in the AI space. No remote participation. No virtual stream. You had to be there, or you were not part of it.
I was not there. But my project was.
What Is Agent Identity Build Day?
The event was organized around a single, urgent question that every serious AI builder eventually confronts:
When your agent acts, is it acting as itself — or as you? Where does its authority come from, and who answers for what it does?
This is not a theoretical question. The moment an AI agent needs to touch a real system — a bank account, a cloud credential, a production database — it needs a real identity. And most systems today solve this badly: a long-lived key sitting in a prompt, an .env file, or the model's context. One bad loop or one injected instruction can turn it loose.
AGI House framed five build tracks around this problem:
- Agents with a wallet — spending authority scoped to a budget it cannot exceed
- Agents that run the stack — deploying, managing resources, holding production credentials without a long-lived key in context
- A personal chief of staff — one agent across inbox, calendar, docs, and CRM, where the hard part is that the same agent reading untrusted email can also send it
- Orgs of agents — when Agent A delegates to Agent B, whose identity and permissions does B inherit?
- …or follow your heart — bring the weird idea you have been putting off
I chose the last one.
Project NaBaB Ai: A Philosophical Filter Agent
My submission — NaBaB Ai — is described as: "A singular philosophical filter agent that halts misalignment by evaluating..."
The full description was cut off in the platform preview, but the idea is complete in my mind. NaBaB Ai is not a productivity tool. It is not an automation layer. It is a moral architecture — an agent whose primary function is to evaluate whether another agent's intended action is aligned with a coherent ethical framework before allowing execution.
In a world where agents are being given wallets, credentials, and the ability to spawn sub-agents, the question of who checks the agent is not optional. NaBaB Ai is my answer to that question.
The Unusual Circumstance
Here is the part that makes this story worth telling.
The event was in-person only. 256 builders were physically present in Hillsborough, CA. Project submissions were meant to be evaluated in the room — builders pitching to judges face to face. The schedule was clear:
| Time (PDT) | Session |
|---|---|
| 10:00 AM | Doors Open |
| 11:00 AM | Keynote Begins |
| 12:00 PM | Project Pitches |
| 12:15 PM | Lunch & Building Begins |
| 4:00 PM | Project Check-ins |
| 7:00 PM | Project Drafts Due |
| 8:00 PM | Finalists Demo & Prizes Awarded |
I am in Bangladesh. When the event was scheduled to begin at 10:00 AM PDT, my local time was already past midnight. I had submitted NaBaB Ai hours before the doors opened in California — possibly the first project in the queue, simply because of where I am on the planet.
The judges — Muhammad Hashmi (Daytona), Audrey Cai (Airbnb AI Eval), Kartik Joshi, Serhii Nechyporchuk (Casca / YC S23), and Zi Teoh (Ramp) — will see the project. The candidate will not be in the room. That is a situation I find genuinely interesting.
Why This Matters Beyond the Event
I want to be clear about something: I am not claiming this is a victory. I am not predicting a prize. The 256 builders who were physically present are extraordinarily capable — many of them have been working on agentic systems for years. They will pitch well. They will demo live. They have every structural advantage.
What I find meaningful is something else entirely.
When I received my ticket approval, I immediately informed the AGI House team and community that I was in Bangladesh and would not be attending in person. That was transparent. What happened next — the project submission feature being available, the ability to submit from anywhere, the fact that the evaluation system is designed to assess the idea and not just the pitch — that is a reflection of something important about how the global AI ecosystem is beginning to function.
Geography used to be a hard wall. You either had a Silicon Valley address or you were outside the conversation. That wall is not gone, but it has cracks in it. And through those cracks, ideas from Bangladesh, from Dhaka, from people who have never set foot in Hillsborough, can enter the evaluation queue alongside the best builders in the world.
The Rashik Connection
Everything I build — including NaBaB Ai — is an expression of the same core philosophy that drives Rashik - The Awakening: Reconnecting Intelligence With The Soul.
A philosophical filter agent that halts misalignment is not just a technical project. It is a statement about what intelligence should be accountable to. It is an argument that the soul of a system — its values, its ethics, its sense of what it should and should not do — is not a feature to be added later. It is the architecture.
That is what I submitted to AGI House. That is what I believe. And whether or not it wins a Mac Mini or a Nintendo Switch, it will continue to be built.
What Comes Next
The event concludes with Finalists Demo & Prizes Awarded at 8:00 PM PDT. I will not know the outcome until after that. But the project is in the system, the judges have access to it, and the idea stands on its own.
If you want to follow the development of NaBaB Ai, or if you are building in the agentic AI space and want to discuss the ethics of agent identity and accountability, reach out. This is a conversation worth having — regardless of timezone.
— G.K.M. Jarif Ur Rahim
Founder, Rashik - The Awakening
Bangladesh · June 27, 2026
