About
A model to understand how people will use AI agents to interact with government, likely government responses, and emerging patterns of exclusion.
Agentic Surfaces is a free, public catalog of design patterns for AI agents in public services. It covers one thing: how citizens use AI agents to deal with government, and how government uses AI agents to support that.
The catalog is organized into nine territories. Each is a domain where AI agents raise a new policy and design problem: proving who really sent a submission, scoping what a citizen lets an agent do, keeping an automated decision answerable. Inside each territory, a set of patterns works on its problem from different angles.
Every pattern has two parts. Assurance is the confidence government needs before it acts, and the policy scope it works within. Access is the duty to keep the path to an outcome open for everyone entitled to it. Both matter equally, for one reason:
every use of AI agents, by citizens or by government, is also a potential exclusion mechanism
A check that makes a service safer to trust can also lock out someone who cannot produce the signal it asks for. So every pattern spells out who that person is, and how they still reach the outcome.
The library is technology-agnostic and precedent-led. Each pattern points to where something like it has already been built, and rates how mature that response is: established, emerging, or frontier. It also takes a position: detecting AI-generated content is not a response. The patterns work instead with the signals that survive automation, such as provenance, intent, delegation, and accountability. Start in the library.
Agentic Surfaces is built by Dan Woods. It is a work in progress: patterns are still being written and sourced, and some are further along than others. Feedback and suggestions are welcome. Reach me on LinkedIn.