Define the task
Workflow, intended users, constraints, baseline process and a measurable definition of a useful result.
A practical first engagementIs there a useful
Is there a useful
project here?
A focused study of one task, your available data and the evidence needed to decide whether to build further.
Discuss a feasibility study ↗The starting pointOne workflow.
One workflow.
A defined question.
Choose one decision to support, one task to automate or one gap in data coverage. We agree on the question and success criteria before choosing a model or generation method.
A study can conclude that the data needs improvement or that a simpler method is sufficient. The deliverable is an informed next step, including a reasoned recommendation to stop where appropriate.
What the study includes.
The proposal identifies exactly which of these steps are included and how they will be evaluated.
Check the data
Formats, labels, coverage, quality, permissions and a suitable evaluation split. Document important gaps.
Establish a baseline
A small model or generation experiment where agreed. Compare against a simple starting point and inspect failures.
Decide what comes next
A review of results, limitations, remaining risks and the scope of a sensible next stage.
What you provideEnough context
Enough context
to scope the work.
- A description of the workflow and the decision or output you need.
- The data format, approximate volume, available labels and how the data is collected.
- The current process or baseline, plus examples of useful and unacceptable outcomes.
- Access, privacy, licensing, compute and integration constraints.
- A person on your team who can explain the workflow and review the result.
Your first enquiry only needs a description. We agree confidentiality, access and a suitable transfer method before you share records.
What you receiveA decision your
A decision your
team can use.
- A documented problem statement and data assessment.
- The agreed baseline or proof-of-concept artifacts, where included.
- An evaluation report with results, error examples, assumptions and limitations.
- A recommendation to proceed, refine the data or stop, with the reasons.
- A handover discussion and a proposed next-stage scope when appropriate.
Production integration, additional datasets and ongoing operation are quoted separately unless explicitly included.
Timing & pricingA written quote.
A written quote.
Clear assumptions.
We review the brief first, then quote a defined study. The proposal states fees, milestones, the expected schedule, client inputs and what happens if those assumptions change.
The timeline depends on data access and preparation, labelling, experiment complexity and your team’s review availability. The schedule starts from the agreed prerequisites, rather than the date of an initial enquiry.
Compute, third-party licences, larger data preparation and integration are itemised where relevant. Any change to scope is agreed before additional work begins.
There is no published starting price or fixed turnaround yet. Send a brief for a scope-specific quote before committing to paid work.
Request a scoped proposal ↗