Andon Labs prices autonomy, not tasks

Most AI software is sold as a tool for a specific job. Write the first draft. Answer customer questions. Summarise a meeting. That model is familiar because it fits how companies already buy software, one workflow at a time.

Andon Labs is taking a different view with Pion. Its pitch is not about improving a single task. It is about running much larger parts of a company, which shifts the commercial question from “what does this feature cost?” to “what is it worth to hand over a defined area of responsibility?” The company’s explanation of why it built Pion is worth watching for that reason.

For business owners, this matters because pricing will become a real dividing line in AI. Per-user subscriptions make sense when staff use a tool. Task-based pricing can make sense when usage is predictable. But when an AI system is expected to own outcomes across several workflows, neither model fits neatly. Buyers will want clarity on what the system is accountable for, where human approval remains required, and what happens when it gets something wrong.

Pion is also conceptually close to the direction we see in connected AI operations, where separate agents and workflows are coordinated around how the business actually runs. This is the kind of thing we build into an AI command centre, with visibility and control staying with the operator.

The winner may not be the company with the most impressive demo. It may be the one that can define responsibility clearly enough that a business is comfortable paying for autonomy.