Tools
Open, interactive tools built on the PAC Framework. Free to use, share, and build on.
Potential Scanner In development
An agentic system that researches your organisation, surfaces agentic potential, scores opportunities on business value and blast radius, and maps the capabilities and infrastructure needed. Spawns specialist sub-agents as needed. The same engine powering our agentic screenings.
Coming soonAgent Profiler
Map your agent use cases across six governance dimensions. See where thresholds apply, which autonomy levels they qualify for, and what infrastructure gaps remain.
- Six dimensions, one answer
- Interactive 3D visualization
- 14 example cases with benchmarks
- Shareable via URL or JSON export
Why these tools exist
Most conversations about agentic AI stall on the same question: how much autonomy can we give an agent before it becomes a liability? Opinions differ because the answer depends on things that are rarely made explicit: how reliable the agent is at its task, how much damage it can do when it fails, what it is worth to the business, and which guardrails are actually in place. These tools exist to make those variables visible, so the discussion moves from gut feeling to a shared picture.
Both tools follow the PAC Framework. Potential asks where agents can create value. Accountability asks who answers for the outcome and where the thresholds lie. Control asks which infrastructure earns each level of autonomy. The Potential Scanner covers the first pillar; the Agent Profiler covers the other two.
Who they are for
The Agent Profiler is built for anyone who has to decide whether an agent goes live: product owners, architects, risk and compliance teams, and the leaders who sign off. You describe a use case along six dimensions and the profiler shows which autonomy level it qualifies for and what is missing to go one step further. Fourteen worked examples, backed by public benchmarks, give you a reference point before you profile your own cases.
The Potential Scanner is aimed at organisations that have not yet decided where to start. It researches your company, surfaces candidate processes, and scores each on business value and blast radius, so the first agent you build is the one that is worth it and safe enough.
Open by design
The tools are free to use and the Agent Profiler is published under the MIT licence. Profiles can be shared as a URL or exported as JSON, so a governance discussion can start from a link instead of a slide deck. Nothing you enter leaves your browser. If you want the same methodology applied to your organisation with someone in the room, that is what the workshop and advisory services are for.