PACKAGED OFFER · GENAI PROTOTYPE
Generative AI prototyping and evaluation
Use a 30-day engagement to test one bounded generative AI use case against your data, quality criteria and operating workflow. The result is a working prototype, an evaluation record and a recommendation to scale, revise or stop.
- Runs on your data, with grounding so responses cite real sources
- Measured against agreed accuracy and quality criteria
- Integrated to one real workflow, ending in a scale-or-stop call
Designed around production constraints.
How will it be evaluated? What quality threshold makes it useful? How should it respond when the available evidence is incomplete?
Those questions are answered inside the 30 days, so the go or no-go decision at the end rests on evidence. Five deliverables come with it.
A working prototype
One scoped use case, running on your data, ready to put in front of real users.
Grounding and retrieval
Source-linked responses and defined behaviour for questions the available evidence cannot answer.
An evaluation harness
A defined accuracy or quality bar with measured results, so quality is a number, not an opinion.
Real workflow integration
Wired to one real workflow or channel, not left isolated in a notebook.
Scale-or-stop recommendation
Cost-per-task estimates and a production roadmap, so the decision rests on economics you can see.
A bounded build, evaluated before a wider investment.
Product and data leaders evaluating a defined GenAI use case.
For leaders who have a promising GenAI idea and need to know, with evidence, whether it deserves real investment.
The engagement can test a new use case or reassess an existing proof of concept against production constraints. If the use case is not yet defined, start with an [AI readiness assessment](/ai-readiness-assessment). The prototype draws on our scaled GenAI platform capability and applied AI work.
Four weeks, one build, a decision on evidence.
Each week ends with a working output, so the prototype takes shape in front of you.
Lock the use case
Use-case lock, data access, and success criteria, producing a scoped spec and evaluation bar.
First working build
Data prep, grounding, and a first build, producing an end-to-end prototype v1.
Evaluate and integrate
Evaluation, tuning, guardrails, and workflow integration, producing a measured, integrated prototype.
Test and decide
User testing, cost modelling, and readout, producing results, cost-per-task, and a scale recommendation.
Evidence for an investment decision.
Bring one use case and an evaluation question. Review the evidence after 30 days.
If the evidence supports further investment, the evaluation record and production plan provide the starting point for the next phase.