Building an AI team to challenge the status quo
Posted: 20 September 2017 | By James Poyser, CEO of inniAccounts
Ten years ago when setting up a disruptive business, I knew technology would be at the heart of it. The idea was simple, take an industry that relied on paper and calculators and move it online. More than this though, my co-founders and I wanted to be famous for making tax accounting for self-employed people accurate and do it in real-time.
Thinking back, Artificial Intelligence was something only universities like MIT concerned themselves with.
However, sooner than we thought the advent of AI is now here, and it’s a great time for our industry to embrace it.
As I said, we wanted to be known for our tax capability, and we’ve achieved our goal by becoming the only online tool that can do real-time tax accounting. Naturally then, AI has to be about enhancing that reputation further.
This should be a guiding principle for any business wishing to integrate AI into its business model. AI is a very exciting way to differentiate your brand from your competitors if you build it into the right processes.
We quickly realised we needed everyone involved in the AI initiative or we would alienate people. If they see AI as a threat to their job, which it isn’t in our case, then it’s game over
For example, we have applied AI to routine compliance tasks related to tax – automatically categorising things and predicting scenarios. We use the thousands of tax returns we have done in the past to understand the risks people often face and the mistakes they make when using online tools to do their accounts. We can use this insight to develop models that spot the risks in a client’s accounts earlier than they could and provide them with the right guidance at the right time, and as a result steer them in the right direction automatically.
We refer to this approach as ‘a game of inches’ – it’s the sum of a lot of micro services that make a big difference to the user experience. This is especially important because accountancy is all about accuracy. It’s vital that we only let AI take control of a task if the confidence level is high enough, otherwise we will hand off to our specialists.
Every business adopting AI will need to consider how their team’s skills and capacity will change
There are so many other industries, where accuracy is essential and can’t be understated. Take medical science – it’s a life or death situation if you are using AI to make decisions on whether to operate on a tumour. You must understand this balance of risk.
Commercially, most companies like ours will want to use AI to help customers help themselves and then speak to a specialist about the things that will make a difference to how they grow their business.
That’s an important learning too: every business adopting AI will need to consider how their team’s skills and capacity will change. We quickly realised we needed everyone involved in the AI initiative or we would alienate people. If they see AI as a threat to their job, which it isn’t in our case, then it’s game over.
This appreciation for skills has allowed us to identify the very specific roles, and communication channels we need for the programme. Broadly it falls as follows:
- App development skills – These people need a grasp of what AI/ML is and where it can give their product the edge. Developers must know how the product is used. If they work with the product managers who are closer to the customer, they can apply some intelligence to the usage data and learn about people’s habits. That insight can be used to develop a product that offers a smoother experience. But there can be two outcomes to this. You might be able to apply AI to the product as it is and make small tweaks, or you might be able to fundamentally change the product. Being able to develop an app is therefore only part of the skill set – understanding how and why it should be developed is crucial for competitive edge and time to market. We look for these qualities when recruiting.
- Analytical skills: To help with this you need analytical people who are good at pulling apart data but also really understand AI/ML. They need to get into the data and understand it, know how to “ask” the data a question and how to interpret the answer. These data scientists should be able to manipulate datasets, filters, R script, merge data, score their models, and compare and contrast different modelling algorithms without touching any app code (with tools like AzureML which we use).
- Testing skills: Of course there is no point in developing AI apps if they are not fit for purpose. This is where it gets interesting because that ‘magic’ requirement employees or customers demand is very hard to test. So you need people who can see whether the data is used properly and producing the desired outcome. If you don’t have these three types of skills working together then you can’t create a product that solves a person’s problem accurately 90% of the time. This is when AI feels like magic to the person using the app.
- Finally there are the employees, the non-technical people, who need to be included in the process.If the app you build is vital to how your business functions, your own employees can be a great source of change. With a little knowledge of AI automation, suggestions to tweak small things can make a big difference to business performance. In our case, this is where the accountants themselves need an appreciation for technology or at least an enquiring mind that challenges the ‘way it’s always been done’, and to realise that this a way to improve job satisfaction, rather than a threat to their role.
Ultimately we’ve learnt that any industry thinking of building AI-lead services is already ahead of its competitors. But those that think about it in terms of improving efficiency and go on to positively change the status quo – whether their customers realise it or not – will be the real winners.
James Poyser is CEO of inniAccounts, which provides accountancy tailored specifically for independent professionals.