Best AI Product

To enter, a product must:

• Be a live, available product with real users, not a research demonstration or internal experiment.
• Place AI at the core of its value - the product's usefulness depends materially on its AI capabilities, rather than treating them as a minor feature.
• Make responsible, accurate claims about what the AI does, its limitations, and how it handles data, safety, and errors.

Products that build on third-party models are eligible, provided the entry is clear about what the team itself has built and where the value is added. Teams should be prepared to describe their approach to evaluation, safety, and appropriate use.

AI entries should show that the intelligence delivers real, measurable benefit and that quality is managed deliberately. Provide what is relevant:

• Adoption and usage - active users, organisations, or query volume, with growth over time.
• Outcome metrics - quantified improvements the AI enables, such as accuracy gains, tasks completed, time saved, cost reduced, or resolution rates compared with a prior baseline.
• Quality and reliability evidence - how the team measures model performance, including evaluation methods, benchmarks, human-review processes, or error and escalation handling.
• Case studies - accounts of real deployments describing the problem, the AI-driven approach, and the result.
• Validated qualitative evidence - substantiated user statements or structured feedback on the value the AI provides.
• Responsibility measures - how the product addresses data handling, safety, bias, transparency, and human oversight where relevant.

Claims about AI performance should be specific and supportable. Vague statements of capability without evidence weaken an entry.

Entries are scored against the standard weighted framework. For this category, the jury pays particular attention to:

• Innovation and differentiation (20) - a genuinely effective application of AI, judged by the value it creates rather than by novelty alone.
• Evidence of impact and traction (25) - demonstrated, verifiable improvement in real user outcomes.
• Submission clarity and integrity (5) - carrying added weight here, since responsible and accurate representation of AI behaviour and limitations is essential to the category.

Problem significance and market relevance (15), product quality and UX (20), and execution and scalability (15) complete the score. Overstated or unsubstantiated AI claims are treated as a mark against integrity.

Products of the following types would fit this category (described by shape, not by brand):

• A document-analysis assistant that helps legal or insurance teams extract and summarise key terms from contracts.
• A customer-support system that resolves routine enquiries autonomously and routes complex cases to human agents.
• A medical-imaging support tool that helps clinicians flag findings for review.
• A code-generation product that helps developers write, test, and review software.
• A demand-forecasting platform that helps retailers or manufacturers plan inventory from operational data.

The following are out of scope for this category:

• Products that use AI only as a peripheral feature while their core value comes from conventional software - these may fit Best B2B Software Product or Best New Product instead.
• Research prototypes, model demos, or systems without real users.
• Entries that describe AI capabilities they cannot substantiate, or that misrepresent how the system works.
• Products whose primary AI function is generating deceptive, non-consensual, or otherwise harmful content.

A complete entry includes:

• Product name, company, category, and a concise summary of what the product does and for whom.
• A clear description of the role AI plays and why it is central to the product's value.
• A working product link and access for the jury - a live demo, sandbox, guided walkthrough, or credentialed trial account.
• Evidence of impact: usage figures, outcome metrics, and at least one deployment case study or validated user reference.
• An account of how the team measures and maintains AI quality, including evaluation, safety, and human-oversight practices.
• A statement of the AI's known limitations and how the product handles errors and edge cases.
• Declarations confirming the product is live with real users, that AI is central to its value, that the entry is authorised, and that all claims - including AI performance figures - are accurate and can be substantiated on request.

Submissions for the current JIMSO season are open. If your product uses AI to genuinely change what people can do - and you can show it responsibly - enter it for Best AI Product.

---