Frequently asked questions

Kreoh is a configurable R&D tax operating system for advisory firms, tailored to each firm’s methodology and supporting the claim from client interviews and evidence review to report production and UK AIF preparation. Built for R&D tax advisers and innovation incentives teams, Kreoh’s R&D Engine brings specialist AI agents together under the direction of your consultants, with dedicated agents for the UK, Ireland, Australia, the US and the UAE. Behind it is a team of AI software engineers and experienced R&D consultants. They adapt the Engine to your firm’s methods, keep refining it as those methods develop and build additional capabilities where you need them. We also work with client-partners to develop support for further jurisdictions.

Kreoh carries your firm’s methodology through every stage of a claim, so each consultant works from the same templates, terminology and evidence standards, whoever is writing. During setup, your templates, benchmark examples and terminology are built into the Engine, so a newer writer starts from the same foundations as your most experienced consultant. From there, Kreoh prepares the ground for each interview, separates material covering several projects and researches the technological baseline. It then drafts in your firm’s voice. Kreoh’s Reviewer flags gaps and risks before work reaches your senior reviewers. Finished reports export as Word or PDF documents in your firm’s style, and additional agents extend the same support to live client calls and the UK AIF.

Kreoh is more than a report generator: configured around the way your firm already works, it supports your consultants from researching the client’s technology and leading the interview, through drafting and review, to UK AIF preparation. In practice, the work unfolds like this: Before you meet the client: Kreoh’s Research Agent studies the technology behind the project in advance, so you arrive at the interview fully briefed and ready to probe the work with confidence. In the meeting: the Passive Screener joins your call on Teams, Zoom or Google Meet and follows the conversation with you, pointing out gaps, red flags and follow-up questions while the client is still there to answer them. Once the call ends: your recordings and documents are gathered in one place and separated by project, and anything still missing is flagged for follow-up, so you begin writing with the full picture. As you write: you build the narrative one section at a time with the supporting sources beside you, so every interpretation in the report is your own. At review: your reviewers step into the same environment to refine the work with you, and the finished report leaves in your firm’s branded Word template, ready for your client’s feedback. For UK claims: the AIF Agent fills the additional information form in HMRC’s online service from your report, then hands you over to HMRC’s website to review and submit it yourself. At every stage, your firm’s own review and approval process stays exactly as it is.

Kreoh works with your client evidence as it already exists, from interview recordings, transcripts and notes to project documents in all major formats, so your team never has to reshape material to suit the software. Integrations let you draw on material already held in your firm’s systems, and the Passive Screener captures recordings of your client calls on Teams, Zoom and Google Meet. Everything you bring becomes a source Kreoh can cite, so each statement in a draft traces back to the evidence behind it. For deeper evidence, the early-stage Discovery Agent can analyse work-management software and code repositories, with connections tested on GitHub, Jira and Slack. Where your evidence includes specialist formats or other systems, our engineers scope and add support in your deployment, agreeing the formats, connections and any additional development with you.

Yes, Kreoh’s R&D Engine integrates with SharePoint, OneDrive, Dropbox, Google Drive and Box, and exports finished reports into your firm’s own Word template. During setup, we agree which sources and folders your workflow needs, then configure and test your template’s structure, branding and report sections. Your team can then review and edit the exported document in its usual format. If you use another system, our team can assess the connection and scope any additional development with you.

Kreoh restricts each narrative to the evidence for its own project, separating multi-project material before drafting and checking citations against the source, so every draft stays tied to what your client actually said and provided. When an interview or document covers several projects, Kreoh separates the material first, so one project’s scope or evidence doesn’t spill into another. Your adviser confirms the project boundaries before drafting begins. Every statement can then be traced back to its own evidence, giving your team a clear trail if HMRC asks questions. This respect for source material runs through all of Kreoh’s engineering.

Kreoh’s configurable Research Agent carries out deep research into the technological baseline across hundreds of sites and can bring the findings together in minutes, with clear citations for your draft. The scope and time required depend on the research task. Your advisers inspect sources, assess their relevance and question the findings. Research can also surface evidence that challenges an initial interpretation or needs further discussion with the client.

Yes, every statement in a Kreoh draft is referenced to its precise location in the source, rather than only to the file it came from, so your reviewers can check each claim against its evidence. Your report template can place references alongside the text or in supporting reference sections, and advisers can inspect the sources, challenge an interpretation and amend the wording before anything is approved.

Kreoh’s Drafter pauses generation, prompting a request for more information when required evidence is missing, and its screening and review tools flag the gaps to take back to the client. Your team supplies or clarifies the evidence and resolves uncertainties before relying on the output. AI can still make mistakes, so expert review remains essential.

No, Kreoh is built to be directed by R&D experts, taking on the groundwork so your advisers’ time goes into judgement rather than rework. Every output Kreoh produces is designed to be interrogated and reviewed by your experts at every stage. The agents surface source material, draft text and review findings, and the Reviewer can assess individual sections or a complete report for strengths, gaps, risks and suggested edits. Your advisers interpret the evidence, decide which changes to accept and approve the final claim position. This human review approach reflects the oversight described in HMRC’s guidance on generative AI software.

Kreoh uses a combination of leading models from providers including OpenAI, Anthropic and Google, evaluating which suits each workflow and agent. We run evaluations of the latest models to determine which models are best suited for Kreoh’s different capabilities and update our selections as models develop. In the Kreoh Agent, your team can choose which supported model to work with and communicate with multiple models simultaneously. Your evidence, instructions and expert review remain central whichever models you use.

No, Kreoh is configured around your consultancy’s existing methodology, templates, terminology and evidence standards, and turns your firm’s way of working into repeatable AI workflows. It begins with your own work. In private sessions, which we call AI clinics, our engineers work alongside your advisers to understand your process, study the reports your firm is proudest of and agree the standards Kreoh’s outputs should meet. The same sessions look beyond the report. Together, we find where your team’s time goes and where work is held up, then shape the Kreoh Agent around the outcome you need: researching prospects against your firm’s criteria, turning payroll and invoice data into structured cost workbooks, or preparing drafts for review against your internal checklist. Once a workflow is established, your team can reuse it for every client that follows, with your advisers retaining judgement and approval. The sessions continue as your methods and priorities evolve, so Kreoh grows into a configurable operating system for your practice, applying your expertise consistently across business development, R&D delivery and review.

Kreoh has a dedicated AI agent for each jurisdiction it serves, the UK, Ireland, Australia, the US and the UAE, each built around that regime’s own claim structure and guidance. In Australia, for example, the agent works to the scheme’s own structure of core and supporting R&D activities, rather than adapting a UK-style narrative. Discovery Agent coverage is confirmed for the UK and Ireland, with expansion elsewhere available on request. For other markets, we welcome partners who want to combine their local expertise with our AI engineering team.

Kreoh’s Passive Screener joins client calls on Microsoft Teams, Zoom and Google Meet and highlights information gaps, eligibility red flags and useful follow-up questions live, while the client is still there to answer them. The Passive Screener supports you at each point of the interview: During the call: it follows the conversation as it happens and points out gaps, red flags and follow-up questions, so you can explore them while the client’s competent professional is still there to answer. When the call ends: the recording is already captured within the same workflow, so the conversation moves straight into evidence review and drafting. Across your team: recordings and examples help newer interviewers learn from your strongest ones, building a consistent standard for how your firm runs an R&D interview.

Kreoh’s AIF Agent uses your completed R&D report and supporting information to fill the AIF in HMRC’s online service, then directs you to HMRC’s website to review the completed form and submit it yourself. Kreoh keeps the final submission with your adviser deliberately, in line with its approach of compliance by design rather than compliance by correction. The agent fills the form through a secure, internally isolated browser instance, and Kreoh does not submit the AIF on your behalf or save your HMRC login credentials on its servers during this process. The AIF Agent is an add-on module for UK claims.

In a head-to-head test run by Kreoh’s R&D team, Kreoh’s AIF Agent filled the additional information form (AIF) in HMRC’s online service about 35% faster than OpenAI’s Codex, at roughly 80% lower estimated model cost, and carried 2.6 times as much content from the completed R&D report into the form. Carrying more of the report into the form means less of the evidence your advisers prepared is left out of what HMRC receives. Both agents entered all nine checked financial figures correctly. The cost difference comes from Kreoh selecting the right model for each step of the task. As with every AIF that Kreoh completes, your team reviews the form on HMRC’s website and submits it.

Yes, R&D cost-preparation workflows are currently supported within the Kreoh Agent. Your team can work through cost information in conversation with the Kreoh Agent, alongside the rest of the claim, rather than moving between separate tools.

No, Kreoh does not use your client information to train public AI models, and access to it is controlled within a separate client deployment of the R&D Engine. We provide data-protection documentation so your firm can review exactly how its information is processed and the terms that apply to its deployment. Your IT or risk team can review this, along with our deployment information, in Kreoh’s Trust Centre.

Kreoh’s Trust Centre holds our security and data-protection information, and our security pack covers ISO 27001 certification and SOC 2 Type II assurance. Request the applicable certificate, audit report and scope for your IT or risk team. Some documents may require access approval or an NDA. Visit the Kreoh Trust Centre.

Yes, Kreoh carries your firm’s branding, templates and writing standards within a separate client deployment. The platform and the reports it produces carry your firm’s identity rather than ours. We agree on branding, deployment and any additional customisation with you during setup.

Your firm’s system admins control access in Kreoh, assigning owner, editor, reviewer, viewer and super-user roles and managing seats and integrations. Designated super-users can maintain grounding and templates. We agree on the role setup with your firm so that the people preparing reports, reviewing work and managing the platform have appropriate access.

Kreoh confirms the storage location, access and retention arrangements for your service before setup, with data-residency options to match your requirements. Alternative hosting or residency arrangements are chargeable and scoped with your firm. Your IT or risk team can review supporting security documentation in Kreoh’s Trust Centre.

Within two weeks, your team can be screening, researching and drafting through Kreoh’s standard onboarding programme. Those first weeks are spent with your own material. Your templates, example reports and instructions shape how Kreoh drafts, so the work begins to sound like your firm from the outset. The timetable is agreed around your materials, access and any additional integration or security requirements.

Kreoh’s onboarding training comes at no additional cost, followed by complementary customer-success check-ins weekly, fortnightly or monthly. Onboarding is shaped around the work your advisers actually do. Afterwards, our customer-success team checks in at whichever rhythm suits you, to review how Kreoh is being used, answer questions and refine the setup. Paid training modules are available for teams who want to go further, and your proposal sets out what is included and what is optional.

Yes, Kreoh plans the move with you, starting from a scoped pilot that uses your existing templates, example reports and source material. The pilot lets you see Kreoh working on your own material before anything changes. From there, a migration plan covers the material to bring across, access to existing systems, team training and review stages. Any data transfer or additional integration work is agreed in scope before the move.

Kreoh is sold as usage-based packages calibrated to your team’s actual work, with consumption you can monitor as you go. Contact our sales team to discuss the right package, included usage, setup costs and rates for additional usage, so your team can plan for additional capacity.

Yes, Kreoh’s AI software engineers work directly with your team through forward deployed engineering to build tools, connect systems and adapt workflows beyond R&D tax. The same engineers who tailor the R&D Engine to your methods can turn to problems elsewhere in your business, shaping each tool around the way you operate. The Kreoh Agent can also work with your firm’s in-house rules and knowledge beyond technical narrative preparation. We scope the requirements, delivery and commercial terms before development begins.

Yes, Kreoh’s Discovery Agent analyses existing work records to find potential R&D projects, including smaller or less visible activities, and gathers cited evidence for adviser review. Rather than relying on what a client remembers in an interview, the Discovery Agent works through their existing records and prepares questions for the client’s competent professionals, so each conversation starts from evidence. It is an early-stage product, offered through scoped evaluations and adaptation, and not yet fully integrated into the wider R&D Engine. We confirm the supported sources, jurisdiction and handover to your claim-preparation process.

Kreoh’s Discovery Agent analyses code, project-management records, documents and interview transcripts, within the sources, claim period and jurisdiction you select. We’ve tested connections including GitHub, Jira and Slack, and documents and transcripts can be uploaded when direct system access is unavailable. We confirm the connections and any adaptation your client’s systems need during setup.

Your client’s system owners authorise every Discovery Agent connection, scoped to the relevant repositories, folders or sites, and access is checked against workspace membership. Contributors connect only the workspaces they are authorised to access, and you control what is shared with each one, so evidence can be assembled across teams within the agreed boundaries.