Two years after we first started hearing about assistants and bots, the race is truly on to put effective artificial intelligence (AI) tools in the hands of accountants.
US accounting software giant Intuit made the latest play in this arena when it revealed QuickBooks Assistant at its QB Connect event in California event last week.
The QuickBooks bot is not the first into the market: UNIT4 laid down the first marker early in 2016 and Sage emerged as the pack leader in 2017, with Software Excellence Innovation Award for its Pegg chatbot in October.
From the QB Connect demo, it looks like QuickBooks has caught up fast. Conversing with the bot on stage at San Jose, Sasan Goodarzi, general manager of Intuit QuickBooks, explored his current financial standing.
“What did I make in October?” he asked.
Within a couple of beats, the bot answered. “$5,424.”
There were gasps of awe from the 2,000-strong audience, but this is pretty standard stuff for accounting bots. But when Goodarzi followed up with the question, “What did I spend my money on?” the bot responded with a simple bar chart showing the different expense categories.
Beneath the cart was a set of text bubbles with suggestions such as “Last month,” “Last quarter” and “Estimated taxes”. When he clicked the latter, Goodarzi discovered he had a $5,000 liability and asked, “Do I have enough to pay for that?”
On discovering the balance was $9,000, he served up one of the best quips of the four-day event: “I expected a lot more than this in the Intuit bank, but we had to pay a lot of money for this conference.”
What gives QB Assistant the edge at this point is the way the developers have linked it to the underlying accounting tools and enhanced the interface to make it more relevant and friendly to users. The project has been closely aligned with the QuickBooks Self Employed edition and tailored to cater for the kinds of questions those customers ask. (See AccountingWEB.com for on the spot report)
From a UK perspective, it is easy to envisage running a similar query on income and expenses for the past quarter, and then asking QB Assistant, “That looks OK, could you file those figures with HMRC, please?”
Cassie Divine, the business segment leader for self-employed at Intuit, has an explanation for the company’s strength in this area: “Nobody innovates on what causes cuctomers pain like we do to make it the best it can be to be useful. We’re really good at listening to clients.”
The text bubbles reflect this approach, she continued, “They’re suggesting what might be the next piece of information the user would want to know.
“We had hunches, but had to get [QB Assistant] in front of real users to see what they really wanted to know. We learned so much, so fast.”
Because of that customer focus, the size of its user base (including 390,000 QB Self Employed users) and the critical business scenarios that QuickBooks handles, Divine says the AI platform developers at Google and Amazon are keen to collaborate.
A lot of sophisticated machine learning programming goes into bots like QB Assistant, much of which is being enabled by the underlying cloud platforms. But real world applications of AI depend on people Divine building conversational frameworks around typical user interactions.
Having conceptualised those pathways at a high level, QB Assistant can link to other apps and their bots. “That’s why I would bet on Intuit,” said Divine. “We can plug into anything and are building everything with a broad eye on the Intuit ecosystem and the wider world.”
Well funded, well researched and well designed, QuickBooks Assistant will pose a big challenge to rivals such as Sage, Xero and Exact. The latter, for example, rebranded its assistant as Exact Go to help users navigate around data held on their accounting system, while the Xero chatbot currently concentrates on helping businesses to find a suitable accountant.
Sage Pegg made the early running, but will need to get a move on over the winter to match QB Assistant’s functionality and integration.
Until recently, accounting bots have been fun toys to experiment with. If you’ve experienced the frustration of dealing with customer service bots that can’t handle anything more complex than standard queries, you’ll be aware of some of the limitations of accounting’s early experiments.
But the pace of change is so fast in this area that we’re now getting to the point where bots are starting to offer experiences that ordinary businesspeople and accountants might actually want to use.
About John Stokdyk
John Stokdyk is the global editor of AccountingWEB UK and AccountingWEB.com.