Skip to content
Talk to us
MenuClose

Build and test one AI product or workflow.

In 12 weeks, we build and test one AI product or workflow using your data. We agree the scope, fixed price and how the results will be judged before we start. You get the working solution, test results and a handover to support your release decision.

For
Teams with one AI use case to build
Engagement
12 weeks · Fixed scope and price
You gain
A working product or workflow, with test results

What you get.

One AI product or workflow, built on your data.
A working solution for the task and scope agreed with you, refined continually as our team and agents learn.
Test results you can judge.
Before building, we agree how to judge the results and what to compare them with: the baseline. Tests on real tasks show how the solution performs and where its limitations remain.
A review and handover.
At week 12, you receive the results, build and handover documentation. We review what the team needs to run it. You decide whether the next step is production, restricted live use or hold.

How the 12 weeks run

  1. Throughout

    Build, test and refine.

    Our team and agents build the product or workflow on your data, test how it performs against the comparison agreed with you and refine it as we learn. We recommend the direction as the evidence comes in.

  2. Week 12

    Review and handover.

    We review the test results, known limitations and what your team needs to run the product or workflow, then hand over the work.

  3. Release

    Your decision.

    Use the results to decide whether to release into production, start with restricted live use or hold for further work.

Use the results to decide what comes next.

A clear way to judge the results.

Test one AI product or workflow on your data, using the tasks, measures and comparison agreed with your team.

Build, test and refine.

Our team and agents build the product or workflow on your data, test how it performs against the comparison agreed with you and refine it as we learn. We recommend the direction as the evidence comes in.

Release is a separate decision.

Use the results to decide whether to release into production, start with restricted live use or hold for further work.

  • Production
  • Restricted live use
  • Hold

How we work with you.

We fix the scope, price and measures of success before the build. Your team provides the relevant data, access to systems and people to review the work. Our specialists build and test the product or workflow alongside them.

Deeplight Foundry gives your team a shared view of the plan, work and test results. As the work develops, we share what we learn and recommend where to take it next. The release decision at the end stays with you. We agree handover responsibilities with the team taking the work forward.

You can start with Accelerate if you have a use case supported by research or testing; we assess the fit before agreeing the scope. If you still need to choose what to build, Discovery can help you decide.

Other ways we can help.

Practitioners

Bring data and AI specialists into an existing programme, from one expert to a full team. We agree the roles, rates and responsibilities, then work alongside your people.

    Our practitioners use Foundry to organise delivery. We agree the working environment and collaboration with your team.

    How we deliver

    Bring us the use case you want to build.

    Tell us what you want to build. We’ll assess the fit for Accelerate and agree the scope and measures of success.