Canonical: https://deeplight.ae/services/programmes
[Reading guide](https://deeplight.ae/llms.txt): company facts, services, published articles and optional visual assets.
Format: Markdown representation of the public HTML page. Interactive controls and diagrams are available in the HTML version.

# Find the right AI use case.  Build and test it.

Discovery helps you decide what is worth building in eight weeks. Accelerate builds and tests one AI product or workflow on your data in 12 weeks. Each programme has a fixed scope and price.

For

Teams choosing or building an AI use case

Engagement

Fixed scope and price

You gain

A delivery plan or a tested AI product or workflow

## Explore the problem.
Build and evaluate.

Discovery helps you explore the options and choose what to build. Accelerate builds and tests it against agreed measures. The stages can overlap as we test ideas and learn.

### [Discovery](https://deeplight.ae/services/discovery)

8 weeks · Fixed scope and price

When you need to choose where AI can help.

1. Discover
2. Define

- Your decision: proceed, rescope or stop.

[Explore Discovery](https://deeplight.ae/services/discovery)

### [Accelerate](https://deeplight.ae/services/accelerate)

12 weeks · Fixed scope and price

When you have one use case to build and evaluate.

1. Develop
2. Deliver

- Your release decision: production, restricted live use or hold.

[Explore Accelerate](https://deeplight.ae/services/accelerate)

- Discovery can inform Accelerate. Each is a separate programme.
- If you already have a use case supported by research or testing, we can assess its fit for Accelerate directly.
- Once live, use feedback and results to decide what to improve next.

## What you get.

### Discovery

#### When it fits

When you need to choose where AI can help.

#### What you leave with

A clear recommendation on what is worth building, and whether to proceed, rescope or stop.

#### Evidence and handover

- A prioritised use case and its expected business value.
- A check of whether the data, technology and controls can support it.
- Success measures and a delivery plan for the next stage.

#### Stage activities

Discover

Map the work, users and data. Test assumptions, using prototypes where useful.

Define

Choose the use case, agree how to judge success and plan the build.

Your decision: proceed, rescope or stop.

#### Duration

8 weeks

#### Scope and price

Fixed before we start

[Explore Discovery](https://deeplight.ae/services/discovery)

### Accelerate

#### When it fits

When you have one use case to build and evaluate.

#### What you leave with

One AI product or workflow, built and tested on your data, with evidence to support your release decision.

#### Evidence and handover

- Test results from real tasks, compared with an agreed starting point.
- Known limitations and what the team needs to run it.
- The build, test results and handover documentation.

#### Stage activities

Develop

Build, test and improve against the agreed success measures.

Deliver

Review the results and limitations. Prepare the handover and release recommendation.

Your release decision: production, restricted live use or hold.

#### Duration

12 weeks

#### Scope and price

Fixed before we start

[Explore Accelerate](https://deeplight.ae/services/accelerate)

## How we work with you.

Experienced people lead the work.

A small senior team brings data and AI engineering expertise and experience deploying agents in production. Our own agents do much of the research, drafting, building and testing under their direction. We explore more options and test assumptions early, with experienced people accountable for the work.

The build starts from what we have learned.

Our practitioners use [Deeplight Foundry](https://deeplight.ae/foundry) to organise findings, requirements, success measures and decisions, with collaboration arranged around your team. Discovery can test early prototypes; Accelerate can revisit the problem as new findings emerge. If you choose to continue, that work carries into the build.

You decide what happens next.

Before building, we agree how to judge the results and what to compare them with. Accelerate is a continual cycle of building, testing and refining, and we recommend the direction as the evidence comes in. At the end, you review the test results, limitations and running needs before a separate release decision: production, restricted live use or hold.

## In practice

[**An energy organisation · Forecasting for energy operations.** A shared data platform supporting demand forecasting, solar forecasting and reserve management, with evaluation built around operational decisions.](https://deeplight.ae/work/energy)

## Thinking behind the work

[Published insight **Agentic workflows in practice** Three real problems agents help solve, from invoice exceptions to new-starter onboarding, traced through their evidence, agent-supported work, human decisions and permitted system actions.](https://deeplight.ae/insights/agentic-workflows-in-practice) [Published insight **Mapping operational workflows: find the steps where agents pay off** An agent’s return differs from step to step. This guide shows how to map a workflow as it really runs, score each step for agent fit and decide who does it.](https://deeplight.ae/insights/mapping-operational-workflows) [Published insight **Simulating agents in a workflow: price each step before you build** Why a faster step doesn’t always make a faster workflow, what a simulation needs, and an interactive simulator that prices agent steps in a hypothetical KYC review.](https://deeplight.ae/insights/simulating-agents-in-workflows)

## Support the next stage.

### [Practitioners](https://deeplight.ae/services/people)

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.

### [Platforms](https://deeplight.ae/services/platforms)

Select, integrate and operate the AI platform that fits your organisation. We work with your technology choices; NodeShift and Muse are among the options we can help you implement.

- [NodeShift partnership](https://deeplight.ae/services/platforms#nodeshift)
- [Muse sovereign AI](https://deeplight.ae/services/platforms#muse)

## Bring us your next AI opportunity.

Tell us what you want to change and what you already know. We’ll help you choose Discovery or Accelerate and agree the scope.

[Discuss your programme](https://deeplight.ae/contact)
