NeuroBridge

The quality contract

The NeuroBridge Standard

Eight obligations every edition meets before it may use the name. This is what a host is signing up to, what a partner is underwriting, and the reason a score in one city means the same thing in another.
Host an edition
  1. 01

    Two days, in person

    Day one is team formation and building, and ends with the pitch deck deadline. Day two is judged: the first round in the morning, the final and the awards in the afternoon. No remote-only editions — the format depends on the room.

  2. 02

    Teams of one to five

    You can enter alone or bring up to four other people. Teams form before or during day one, and five is the ceiling, not a target.

  3. 03

    The same hundred points

    Value for the user 25, prototype and use of AI 30, quality testing 20, feasibility 15, originality 10. Published before registration opens, identical in every city.

  4. 04

    Two rounds, never one

    Every project is read in a first round against the card, then the finalists are rescored from scratch by a human jury. How the first round is read is the edition’s choice — Baku has GPT and Claude do it live on stage. A single round cannot be compared across cities.

  5. 05

    Mandatory disclosure

    Every submission names the models, data and components it used. Libraries and public templates are allowed when disclosed; previously built products are not eligible.

  6. 06

    Quality testing is scored

    Twenty of the hundred points are for how a team checked its own work. A project with honest failures outscores one with a prettier demo and no evidence.

  7. 07

    Written feedback to every team

    Not only the winners. A judge cannot record a score without writing what was strong and what to change, and the team sees it.

  8. 08

    Conflicts declared, not managed quietly

    A judge connected to a team declares it in one click. The score is excluded from the average and the organiser appoints a replacement.

The card

One hundred points, five criteria, fixed weights. Identical in the group rooms and in the final, and identical in every city.

PointsCriterionWhat judges assess
25Value for the userA specific problem and a clear outcome from solving it.
30Prototype and use of AIA working core scenario, and what AI actually contributes to the result.
20Quality testingTests, examples of failures, and comparison with the current approach.
15FeasibilityData requirements, running costs and a clear next step.
10OriginalityA meaningful difference from common solutions.

Quality testing and feasibility together are worth 35 — more than the prototype itself. That is deliberate. A two-day AI demo is easy to make impressive and hard to make honest, and the gap between the two is where the scoring weight sits.

Three ways an edition runs

The standard is identical in all three. What changes is who delivers it and who pays.

Mode A

Run by Starnest Academy

Who delivers
The network team delivers it end to end.
Use when
Founding editions and any first edition in a new region.

Highest control, highest cost. Does not scale past a handful a year.

Mode B

Licensed to a local host

Who delivers
A university, hub or community runs it, with a lead judge from the network.
Use when
Repeat cities and markets with an existing community.

The mode that makes this a network rather than a touring event.

Mode C

White-label for a client

Who delivers
Delivered for a ministry, corporate or university on their brief.
Use when
Sovereign AI programmes and corporate innovation.

Same card, same rounds. The client brands it; the standard still applies.

What the standard is actually for

So scores travel

A team that places third in Baku and a team that places third in Nairobi were measured by the same five numbers, in two rounds, by judges who could not see each other’s sheets. Without that, a network ranking is decoration.

So licensing does not rot it

Every franchised format has failed the same way: a local edition quietly drops the parts that are inconvenient. The obligations above are the parts that get dropped first, which is exactly why they are the ones written down.