Whilst HMRC do not have a separate classification for AI and robotics, it is likely that the subject does not sit as a sector vertical but touches any sector that is looking to drive automation and increase productivity. In plain English, these types of development align closely to the R&D tax relief system as the fundamental tests are:
A typical claim is likely to include:
The scheme is broad, and it is likely that if you are working on an AI or robotics project or solely focused on this subject, investigating R&D tax credits is a must.
There are five broad categories which AI and robotics R&D claims fall into – these are staff costs, subcontractors, externally provided works (EPWs), consumables e.g. heat, light and power. More in-depth discussion on each of these can be found below.
The majority of most R&D tax relief claims are in respect of staff costs. This includes people directly involved with the activity and others who are indirectly involved. An example of direct costs would be a software engineer or a robotics engineer – they tend to be defined technical roles who are the doers of the project.
A common misconception is they have to be scientists and academics, this is incorrect. If they are directly contributing to the R&D effort, then there is a good chance their costs can be included.
Indirect staff are people who are also involved in supporting the R&D project such as administration, finance, and personnel activities where they enable the R&D to happen. These costs are usually overlooked but have a rightful place in the claim.
Where you may not have the expertise in house, you may rely on a third party to help with part of the R&D process. These costs are qualifying expenses for the R&D tax relief scheme. For overseas subcontractor costs new rules apply for accounting periods starting on or after 1 April 2024.
Like subcontractors, you may not have directly employed staff but want to have resource for a particular role who act in the same way as employees – i.e. under your care custody and control. These costs can form part of the claim. For overseas EPW’s costs new rules apply for accounting periods starting on or after 1 April 2024.
The proportion of heat, light and power that is used for the R&D project is also qualifying expenditure.
Any software used directly in the R&D is a qualifying cost. For accounting periods on or after 1 April 2023 costs related to datasets and cloud computing can also be claimed.
The SME and RDEC schemes will merge into a single scheme for accounting periods starting on or after 1 April 2024.
With the SME R&D tax credit scheme you can currently claim tax relief of up to 33% on your qualifying R&D expenditure incurred up to 31 March 2023. After this new rates will apply.
Read more
The Research and Development Expenditure Credit (RDEC) rate has increased over the years, companies can get over 10% of their R&D expenditure incurred up to 31 March 2023 refunded. Then this will increase up to 15%.
Read more
A client wanted to reduce the reliance on manual handling in the warehouse. Due to the structure of their products, standard machinery was not suitable so they embarked on a project to see if they could get standard machinery to work by amending their process or create bespoke robotics to get an answer.
After almost two years of trials, testing, failed attempts the client achieved what they wanted to – and received an R&D tax credit over £100,000 each year.
Our client was a start up software company who had become frustrated with the reliance on teams of people to crunch numbers where they felt that this could be coded and continue to learn based on real life events.
By using a number of standard data feeds then creating the linkages into a singular system, in less than six months our client had a prototype of how this could work. This qualified for R&D and created enough cash flow to help push on to the next phase.
Our client then reinvested that money to employ a small team of engineers to work on the AI. They continue to develop the coding and each year benefit from R&D tax relief.
