All insights

Public Sector3 November 20261 min read

Prototype before you procure

AI has collapsed the cost of building working software. Public buyers can now ask suppliers to show, not tell, and suppliers can afford to.

For decades, public procurement of digital services has rested on written promises. Suppliers describe how they would meet each requirement in 750 or 1,000 words, evaluators score the prose, and the first working software appears months after contract award, by which time changing course is expensive.

That made sense when building software was slow and costly. It makes much less sense now.

The cost of a working prototype has collapsed

Using Forge, we recently built a complete resident engagement platform at bid stage: registration with multi-factor authentication, surveys and consultations, idea boards, events, moderation queues, consent management, live insight dashboards, report templates for boards and regulators, CSV/Excel/PDF exports, an API, SCIM provisioning and an MCP server for AI analysis. Evaluators can click through it end to end.

A few years ago, that would have been an alpha phase in its own right. Today it is something a small senior team can produce to support a bid.

What buyers can do differently

The Procurement Act 2023 gives contracting authorities more flexibility in how they run competitions. Used well, that flexibility lets buyers:

  • Ask for demonstrations, not just descriptions, at shortlist stage.
  • Score working software against real user journeys and accessibility tests.
  • Reduce risk by seeing a supplier's actual engineering quality before award.
  • Level the playing field for SMEs, whose working software can outshine larger suppliers' bid-writing teams.

A caution

A prototype is not a production service. Security accreditation, data migration, integration, support and operational resilience still take real work, and a good supplier should be explicit about which parts of a demo are real and which are simulated. Our prototypes ship with a written list of assumptions and limitations, for exactly that reason.

But the direction is clear. When working software is cheap, the most honest answer to "how would you meet this requirement?" is "here it is, try it."

Keep reading

Related insights