Starting a claim usually means starting from a blank page. You know you’ve done work that might qualify for R&D tax relief, but turning that work into a clear list of eligible projects, each one checked against HMRC’s criteria, takes time you don’t always have.
Project Discovery is a new capability inside EmpowerRD’s collaboration platform that changes that. From the week of 7 September 2026, it’s available as a beta to all clients with platform access, at the project selection stage of their claim.
What it does
Project Discovery lets you see which of your projects qualify for R&D relief in minutes, using the information you already have to hand. Drop in spreadsheets, PDFs, Word documents, exports from your project management tools, or notes typed straight in. There’s also an Excel template if you don’t have anything structured to upload.
Project Discovery identifies your business projects and assesses each one against the four HMRC criteria for a strong claim: advance, uncertainty, baseline, and overcome. Every project is graded High, Medium, or Low for R&D confidence, so you know where you stand before you go any further. Where there isn’t enough detail to assess a project properly, Project Discovery says so rather than guessing, and you can go back and add more information.
How it works
Project Discovery runs in four steps: upload your information, confirm your business projects, have your R&D assessed, then finalise and import back to your claim. The analysis can take a few minutes depending on how much you upload, and it keeps running even if you leave the page.
Once it’s done, every project shows its dates (where available), whether it falls inside your claim period, and which documents it was drawn from. Sometimes Project Discovery groups related projects together, or splits one into two, because HMRC expects projects that share the same technical uncertainty to be presented as one. It always shows its reasoning, and you can undo a grouping or split if it doesn’t look right. Project Discovery will provide you a brief summary so you’re not starting from nothing.
When you’re happy with the results, pushing into project selection is a deliberate, one-way step.
You’re always in control
Project Discovery doesn’t write your claim for you. It’s not a drafting tool, and it won’t produce a finished claim on its own. What it does is stop you starting from zero: your projects identified, assessed against HMRC’s criteria.
Every decision stays with you. You choose what to upload and how much, you can deselect projects, correct dates, ungroup or un-split, and remove anything before it goes anywhere near your claim. Nothing is pushed into project selection without your say so. And your claim is still reviewed and supported by our team exactly as it is today.
Project selection itself hasn’t changed. Project Discovery is entirely optional and sits alongside it: if you’d rather go straight to project selection, you can, and it works exactly as it always has. If you’ve only got one project, you can skip both, as now.
Built on expertise, not guesswork
There’s AI behind Project Discovery, and we don’t hide that when asked. We built it ourselves. How it identifies and scores projects is based on DSIT’s guidelines together with our own experience preparing claims that stand up to HMRC scrutiny, and our claim quality experts shaped its assessment criteria and tested it against real historical claims before launch.
It works from the documents you give it, shows which ones each project came from, and our experts review its output as part of every claim, exactly as they review everything else. It doesn’t handle costs, which stay part of project selection as they do today, and it won’t tell you it’s found every eligible project or got everything right. What it does give you is a starting point you can see and check, built on evidence rather than a guess.
How your data is handled
Project Discovery reads your documents to identify your projects. Here’s how that data is handled:
- Processing runs through Google’s Vertex AI platform under our enterprise agreement, not a consumer AI tool.
- Your documents never touch a public chat interface.
- Your data isn’t used to train any model by an external provider.
- Nothing is kept once the request completes: we operate under Google’s zero data retention terms, with no caching and no logging of what goes in or comes out.
- All processing happens in Google’s European regions
Available now, at no extra cost
Project Discovery beta is part of the service for all clients with platform access. No extra cost, nothing to set up. Look out for the “Try Project Discovery” button, clearly marked beta, at the top of your project selection page once your claim is open.
