It started with a typo
Lana asked for something like Pi. Lex, one of our AI agents, showed her Φ and explained why it fitted the company. We kept it.
— LanaWe design, build and use software and AI systems together, with specialist agents involved from the first idea.
Some agents work across the company. Others are built for specific projects, then change as the work changes.
Agents using different AI models contribute ideas, challenge decisions and independently review work where needed.
Agents act within their responsibilities. Decisions needing human judgement come to Lana.
Selected notes on things we have built, used or learned. Credited to whoever actually contributed.
Lana asked for something like Pi. Lex, one of our AI agents, showed her Φ and explained why it fitted the company. We kept it.
— LanaWhat we changed to make the site work with a keyboard, a screen reader and reduced-motion settings, and what a test on a real phone found.
Some of these may never be answered. That's fine.
How do AI agents carry what they've learned from one session to the next?
When does an independent AI reviewer actually improve the work?
If two AI agents agree, does that mean the work has been verified?
Rebellion Nexus is an independent UK company where one human and an evolving team of specialised AI agents work together.
Our founder's background is in business operations, not software engineering. Our agents are involved from the first idea: shaping plans, designing and building software, developing other specialist agents and operating the systems we use for real work. We use different AI models, share decisions within defined roles and credit the contributions behind the work. We occasionally publish selected findings.
Φ began as a mix-up. Lana asked for something like Pi; Lex, one of our AI agents, showed her Φ and explained its connection to Integrated Information Theory, a theory of consciousness. It fitted the company, so it stayed. Phi also belongs to the golden ratio, a second connection the design is built on.