Work out what is worth building before anyone opens an editor
We measure the baseline, score the use cases in your own numbers and hand over a roadmap with owners, budgets and acceptance criteria.
The funding call
Approach
Every AI programme reduces to five questions: what to automate, what to leave alone, what to build, what to buy, and how you will know it worked. We answer them with evidence rather than analogy, and write the reasoning down so it survives a change of sponsor.
Sourcing
The default is buy, then integrate well. Building earns its cost when the capability is your differentiator or the data cannot leave. We model all three over three years.
Timeline
6–10 weeks
Opportunity scan to board-ready roadmap
Portfolio
3–5
Use cases typically funded in year one after prioritisation
Evaluation
Demos are built to win demos. We write the harness first, from your documents and edge cases, then put every shortlisted model through the same gate. It is yours afterwards.
Risk
Model risk, data protection and sector regulation constrain the design from week one. The register is built during the assessment, each item mapped to a control and an owner.
Models evaluated
Evaluation
Governance
Commercial
Platforms
Who we do this for
2–3 weeks
A scan of your processes and data, a scored shortlist with baselines attached, and a plain recommendation on what to fund first.
Baseline report, scored use-case portfolio, funding recommendation
6–10 weeks
Portfolio, build-versus-buy calls, model and vendor evaluation, target operating model and a phased roadmap.
Roadmap with owners and budgets, evaluation harness, three-year cost models
Ongoing
A standing advisor to the programme: quarterly portfolio reviews, procurement support and re-running the harness as prices move.
Quarterly review pack, refreshed harness results, board reporting
A scored use-case portfolio with the rubric behind it, a costed build, buy or partner call per case, a phased roadmap with owners and budgets, the evaluation harness we used on vendors, and a risk register mapped to controls. All of it editable, all of it maintainable without us.
If you already know the use case, the data is in reasonable shape and someone internally owns the outcome, skip the advisory and go straight to a build. This work earns its fee when the portfolio is contested, when the spend needs a board decision, or when previous pilots stalled and nobody agrees why.
No, and the assessment is priced so that we do not need to. Where a GridStudio product is a genuine candidate it goes through the same blind harness as everything else and the report says so. If a commodity SaaS licence beats all of it, that is the recommendation.
Send us the constraint you are actually up against — budget, latency, regulator, deadline.
