Knowledge base / regulation10.com product guide

Which plan tier fits, and what actually changes between them

The plans are Starter, Professional, Enterprise, or Sovereign. They differ on three things and nothing else: how many AI systems the workspace can hold, how many people can work in it, and which capabilities are switched on. The assessment logic itself is identical at every level. A Starter workspace runs the same classification, the same module questions and the same evidence pipeline as a Sovereign one.

Start with the ceiling, because it is the one limit that can actually stop work. Starter holds up to 1 AI system with 1 included seat. Professional holds up to 5 AI systems with 5 included seats. Enterprise holds up to 25 AI systems. Sovereign has no fixed ceiling on the number of AI systems. Additional seats can be added on the plans where seats are purchasable, so the seat count is a starting allowance rather than a hard wall.

Counting correctly matters more than it sounds. An AI system here means one deployed or planned system with its own purpose and its own accountable owner. It does not mean one model, one vendor or one API call. A customer-facing chatbot is one system. A CV screening tool in human resources is a second. The same underlying model used for both is still two systems, because each has a different purpose, a different population affected and a different owner answering for it.

The capability steps are the second axis. Enterprise identity sign-on becomes available at Enterprise. The agentic assistant that works across your estate becomes available at Enterprise. Live drift monitoring becomes available at Enterprise. The runtime governance gateway becomes available at Enterprise. Self-service export of your full record set becomes available at Enterprise. Single-tenant residency is a Sovereign option. Everything below those lines is present on every plan, including the full evidence pipeline and the audit trail.

The third axis is response commitment. The acknowledgement window for a complaint runs from 240 hours on Starter down to 48 hours on the upper plans. This is an acknowledgement commitment, not a resolution time, and it is stated that way deliberately.

A practical way to choose. Count the AI systems you would register in the next twelve months, not the ones live today, because a system in design is exactly when governance is cheapest to apply. Take the lowest plan whose ceiling covers that count. Then ask one question: is anyone going to inspect this work in the near term? If a regulator, an auditor or an internal assurance function is already asking, the plan that includes the agentic assistant and live drift monitoring is the one that matches how you will actually be working, because you will be answering questions about the estate rather than filling it in.

Two honest caveats. First, the plan does not change how deep any single assessment goes, so buying up will not produce a better answer about one system. It buys coverage of more systems and more operational capability around them. Second, this article deliberately quotes no figures. The current commercial terms are set out on the plan configurator, and the assistant will point you there rather than quote a number it cannot keep current.