How to implement AI in your practice
As the partner responsible for rolling out the AI programme at Moore Kingston Smith, Becky Shields found sticking with four major principles helped her to get the firm’s team to buy into new ways of working. Here, she shares her learnings.
But as Becky Shields, the partner responsible for the AI programme at Moore Kingston Smith (MKS), is keen to point out, citing the example of Receipt Bank, the receipt-scanning software: “AI has been part of the business for a long time.” So, the key to getting the most from it is just like with any new tech: speedy, detail-focused implementation and then good communication and direction from the top down to ensure everyone uses it in the right way. As Shields’ recent success at MKS shows, four major steps will help.
Start with a plan, be clear and get agreement
As with any big project, launching a new AI platform should start from the position of laying out the rationale for why you want to spend time and money on this. With MKS, the main reason was to improve the quality of its work for clients. And it wasn’t long after the platform launched that AI questions started coming up in the firm’s retendering processes across all sectors.
As a relatively new partner, Shields was asked to report back on MKS’s options. She says even at that early stage they were given “a clear steer on what was wanted from the programme”. This enabled them to come back with a coherent idea of what technology was wanted and how they would use it.
No clarity about these things up front and – even more crucially in a partnership – no explicit agreement on the purpose of the investment can lead to delays and spiralling costs with the project team trying to please too many people.
Get the right team in place; establish a good vendor relationship
Once the plan is signed off, make one person clearly accountable for its implementation. This way it’s clear who is making decisions on behalf of the firm and who people should go to with questions. MKS did exactly this with Shields and she then built a team to support her.
She found people with cross-discipline experience; someone from the MKS technical team also had programming expertise: “He was perfect, in terms of understanding questions like “how should this fit into the audit framework?” or “what does it do in terms of sample size?”
It turned out a qualified senior in the firm was building his own accounting software in his spare time. MKS worked out a deal so he could spend some time on the AI project and some on his own work. Shields had one more full-time member and then pulled in various trainees and others when she needed them.
Shields used her initial analysis to start talking to vendors. One key consideration was the data literacy of employees. “Things like IBM Watson are absolutely fantastic platforms, and they can do this incredible analysis, but you have to have data literacy skills that we didn’t have at the time.” Shields didn’t want to commit the firm to a vendor for too long, given how rapidly things are changing.
“You’d be mad to think you found a solution that will last you three years or more,” she says. They also chose a vendor that they could work closely with. “We wanted somebody that really understood the value of a partnership, rather than just trying to flog us a piece of software.”
MKS was the vendor’s first European client, which meant it partnered closely over demos and referrals, and that when clients wanted something different, Shields was able to say “this will make an excellent new case study for you; it’s worth your time to find a solution for our client”.
Don’t hang about, and focus on putting all your effort into support
The next thing MKS did well is move quickly. It took six months from initial analysis to offering AI capabilities to clients. Speed is important, not only because there’s no return on investment until it is in clients’ hands but also because momentum is maintained across the firm. One reason MKS could move fast was because the initial decision-making was clear. As Shields says: “When we do something, [at MKS] everybody does it.”
Once you’re up and running, be ready to help the rest of the firm – in any way – to use the new capability. Also, find some quick wins to keep showing the value of the new capabilities. Shields and the team prepared standard slides and documentation together so that everyone at the firm told clients the same story. Her team was also rigorous about keeping the tech working: “An audit timetable can be quite restrictive with short deadlines, so we couldn’t have the technology holding up the process at all.”
The biggest challenge was that General Data Protection Regulation legislation came into force a year after the platform launched. The team wrote documentation to answer all general questions and ensured they were on hand to cope with special requests. For example, one client wanted all data to be stored on the premises, rather than in the cloud – a key part of the vendor’s platform. But close custom work with the vendor solved the problem, and the client was happy to pick up the cost.
Record what went well, and prepare for what comes next
Finally, make sure you record lessons learned and keep an eye on what’s next for AI – both in terms of the tech and the skills your staff will need. Shields says they’ve learnt a lot about the importance of data standardisation, and how to ensure people have the right skills to link and analyse all the data from different accounting packages, in particular. “It’s one of the biggest barriers to adoption,” she says.
It’s important to keep evolving the programme too. For MKS, 2020 will see a focus on real-time analysis. “Too much of the profession is almost after the event telling people where they went wrong and it’s just not of value any more,” says Shields. MKS will use its data warehouse and the skills and tech to provide much faster advice to clients at little or no extra cost. Just like a well-managed implementation programme, this would be a win-win for everyone involved.
Words by Tim Stafford
We want to encourage wider debate about the long-term opportunities and challenges for the profession that AI poses. You might also like to read our other related articles:
