You can just build
Personal ASI is coming soon. We realized our own product can now one-shot things I cannot build myself.
The Singularity Has Been Crossed
Starting in November 2025, we felt clearly that the singularity of software engineering had been crossed.
Before that, even the best engineers still treated AI as an “AI typewriter”. People were still responsible for architecture design and business implementation. AI only bridged the “garbage time” of writing code.
But after that, especially after the later release of GPT-5.4, even the most demanding architects no longer needed to keep opening editors over and over. Intent and done were bridged.
We witnessed the coding harness becoming complete:
From autocomplete for a snippet of code, to handling an entire code file, to handling every file across a whole repository. We used to need complex RAG and indexing strategies just to make the model barely work. Now, coding agents can dynamically understand any codebase, and they have taken over requirements analysis, testing, deployment, and CI fixes.
As a result, coding assistants and coding copilots have become digital engineers.
Only end-to-end work can show the real value of AI. As long as there is still a missing link, you need several times more attention to find gaps and patch them. We once hoped that agentic AI could also make the repetitive, boring, tedious parts of our work disappear. We should be putting our attention on what truly matters. But for a long time, across the network of general human tasks, neither human nor AI attention could be allocated correctly.
But now, the singularity has moved beyond coding. Coding has a clean, elegant harness: Git, code file diffs, API-driven build pipelines, and a credential-based permission system make it a perfect fit for AI. What you are seeing is not a specialized AI, nor infrastructure specialized for AI. Instead, when AI is smart enough, you can see it doing any work in every step and in every tool.
Introducing Bridge
We want to accelerate the arrival of truly general personal ASI. We need agentic AI that can use human tools and deliver results in human workflows. If ASI means anything, it first has to be able to free our hands from the keyboard.
So we started a company called Afk .Inc (Away from keyboard). Our first product is called Bridge.
It has its own computer, and it can also use any of your devices. Bridge connects scattered tools and context to complete any task. Whether it is monitoring new AI updates and stock prices online, or digging through clues across your computer, cloud drive, Slack chat history, and meeting notes, it can handle it all. Humans can only work serially, but it can pay attention to every step and every source at the same time. Screenshots, files, email, messages, all handled.
It keeps aligning with what you need:
It proactively senses and helps with your emotions and impulses. It uses long-term memory to keep learning and evolving, aligning itself with your work environment and habits. After completing complex tasks, it summarizes what it learned and reports back, forming your personal skill set. Every deliverable it produces for you is another round of alignment training.
Complete tool set:
Humans can only effectively control one mouse and keyboard, but AI does not have that limitation. If the tools you use do not have a CLI or an API, that is fine. With our Computer Use, AI can collaborate with you to finish tasks, or handle them quietly in the background on its own. Also, using only one app at a time is a human limitation. Bridge can use multiple sets of mouse and keyboard to operate multiple apps simultaneously.
Just build and ship:
Bridge is a lead agent. Multimodal inputs and sub-agents are integrated automatically. You do not need to learn a thousand different AI tools. Instead, end-to-end image generation, code building, video understanding, and speech understanding are automatically orchestrated by Bridge to produce deliverables you can ship.
With a single sentence, the best multimodal models work behind the scenes to turn your imagination into an app anyone can use.
You can even use Bridge to build other agents.
Save your attention for what you need
“Attention Is All You Need,” and the Transformer it introduced, gave us a way to compress massive neural networks by focusing on what matters.
We want our human attention to be reserved for the decisions that truly matter and the steps that are truly critical. We want everyone to have the power to make things happen simply by saying so.
But over the past two years, agentic AI has instead made our workflows more bloated and more complicated. The cost of aligning and training AI seems to have gone up.
Attention became scarce. Until now, let Bridge take care of things.
Bridge, a new source of attention. Your personal ASI.