Dutch software vendor Unit4 has raised the bar in the accountancy bot race with the launch of its digital assistant ‘Wanda’ at the annual Unit4 Connect Conference.
Later this month users of the vendor’s Business World On! platform will have access to artificial intelligence-powered bot Wanda, with key finance features including automatic expense claim completion, an automated purchasing approvals process and GPS-generated timesheets.
First to market
At last year’s Connect Conference Unit4 was among the first to outline exactly how it was planning to use machine learning and AI in its applications, and since its announcement back in April 2016 the accounting space has been awash with bot talk of varying degrees of functionality, from Sage’s Pegg to Xero’s Facebook messenger chatbot.
However, despite all the hype there has been little in the way of product, and of the larger vendors Unit4’s Wanda is the first to hit the market.
Wanda is billed as delivering the following new functions, which Unit4 calls its ‘assistants’:
Travel assistant – Generates travel requests and manages approvals based on travel patterns and preferences. Auto-population of expense claims using receipt recognition technology
Purchasing assistant – For finding common products and suppliers, generating requisitions and managing approvals
Time assistant – Automatically generates timesheets based on multiple data streams and GPS and beacon location data
HR assistant – Simple completion of HR-related employee self-service tasks like making absence requests and enquiring about holiday balances and pay slips
Approval assistant – Notifies and reminds managers to approve tasks and flags important tasks where deadlines are imminent
One of the main advantages of the Dutch vendor’s product is that it requires little to no user training, as it interacts with users through natural language. Wanda is designed to be embedded in its user’s favourite chat application, whether that is Skype, Slack or Facebook Messenger, and can deal with multiple topics mixed in a single chat thread.
Company product strategist Thomas Staven put the bot through its paces with a demonstration at the Connect Conference. In the case below Staven opened Skype on his smartphone, found Wanda in his contacts list and then asked it to file an expense claim.
As shown above Staven attached a picture of the receipt for a train journey, which Wanda then scanned and processed using OCR receipt recognition technology. The bot has the capability to then attach the receipt to a particular project or projects if necessary, and in the case of taxi journeys ask why public transport was not used.
Staven also operated Wanda using voice capability – in this case with an Amazon Echo – updating a forecast on a project for which he is responsible. Staven acknowledged that may not be the answer for every employee sat at their desk, but could work in the car, for example, or help people with sight impairment. Here's an early demonstration of Alexa in action:
Wanda is powered by Microsoft Azure, the software giant’s cloud division, and benefits from Microsoft’s sophisticated analytics and machine learning technology. Unit4 also uses Microsoft’s Language Understanding Intelligent Service (LUIS), giving Wanda the ability to understand what a person wants from the language they use.
Another round of bot developments
There has been a great deal of bot talk over the past 12 months, so it is almost a relief to finally see a product roll off the shelves. It will be interesting to see if the much-discussed machines can deliver on what they promise, or if there are likely to be teething (or should that be Bluetoothing?) problems that prevent the systems reaching their full potential.
It will also be fascinating to see if, like last year, Unit4’s announcement kicks off another round of bot developments in the accountancy sphere, and what form these will take.
Tom is editor at AccountingWEB, responsible for all editorial content on the site. If you have a story that might interest us or wish to comment on the site's coverage get in touch via the site's private message function or Twitter DM (@AWebTom)