Skip to main content
Main Logo Black The AI Learning and Innovation Hub helps public-sector teams use Blueprint to move from a mission opportunity to a responsibly tested, reusable AI capability. It is a practical space to learn by doing, share lessons across agencies, and build tools that improve service delivery for constituents. Start with a concrete service or operations challenge. Then use Blueprint to discover useful starting points, build a focused prototype, test it with appropriate data and people, and decide what to improve, govern, or scale.

​
The government innovation cycle

​
1. Start with mission, people, and outcomes

Define the public need, the people the work is intended to serve, and the outcome that would make the effort worthwhile. Identify policy, privacy, accessibility, procurement, and operational requirements early. Be clear about what a tool may recommend, what must remain a human decision, and how people can raise a concern or receive help.

​
2. Discover a responsible starting point

Explore Public Libraries and your organization’s approved libraries before building from scratch. Review an asset’s purpose, inputs, connected tools, data handling, permissions, and expected outputs. Reuse or fork only when it is appropriate for your agency’s context and governance expectations.

​
3. Build the right kind of AI tool

  • Prompts turn a recurring task into a clear instruction with defined inputs and expected outputs. Create a Prompt.
  • Blueprints connect AI, data, logic, knowledge, tools, and human review into an inspectable workflow. Create a Blueprint.
  • Chatbots provide a guided conversational experience for staff, partners, or constituents when that interaction is appropriate. Create a Chatbot.
  • Agents can bring together a defined purpose, approved tools, knowledge, and operating boundaries for a reusable assistant.

​
4. Test in a safe, representative setting

Use de-identified, synthetic, or otherwise approved test data. Test the normal case, incomplete inputs, and edge cases. Check whether outputs are useful, accurate enough for the intended task, understandable to the people using them, and routed to a person whenever judgment, eligibility, rights, benefits, or safety are involved.

​
5. Govern before delivery

Keep work private while it is being tested. Before sharing or deploying, confirm the accountable owner, access permissions, approved data and tool connections, human-review points, escalation path, recordkeeping needs, and communication expectations. Make sure the tool is accessible and fits the service experience you want to provide.

​
6. Share what you learn

Document what worked, what did not, and what needs more evidence. Preserve useful versions, share reusable patterns with approved organization libraries, and let lessons from one team improve the next build. This turns isolated experimentation into lasting public-sector capacity.

​
Learn together across agencies

The Hub is designed for hands-on testing, ethical framework development, AI literacy and capacity building, and shared learning across government teams. Cohorts can compare approaches, surface common risks, and refine practical patterns for responsible use in public service.

​
Start a government-ready build in Blueprint

  1. Open the Blueprint Launchpad and identify the mission problem you want to explore.
  2. Search public and organization libraries for a reusable starting point.
  3. Create a small Prompt, Blueprint, Chatbot, or Agent with an explicit purpose and a named owner.
  4. Test with appropriate data and reviewers, including people who understand the service, policy, and user experience.
  5. Review governance, decide whether to revise, share internally, or prepare a controlled deployment, and capture what the team learned.

​
Additional learning