The idea of automation fills many with instinctive dread, conjuring up images of awkwardly interacting with an emotionally void bot.
So Kriti Sharma, Sage’s global head of mobile, knows the bot she is creating will be a hard sell. Dressed informally and speaking with non-corporate plainness, Sharma is an incongruous figure to be perched so high in a major corporation.
Sharma’s bot has a simple goal: it will help small business owners and freelancers capture and manage invoices and expenses. Sharma accepts that people still harbour powerful preconceptions around AI but, she doesn’t waver: “There used to be people who said, I would never do banking on my mobile,” she says.
There’s the famous (probably apocryphal) quote from the automotive giant Henry Ford: “If I had asked people what they wanted, they would have said faster horses.” The big innovations, Sharma says, don’t come from users explicitly asking for it.
“As a product visionary, you sometimes have to create the product that users will love although they might not be asking for it,” says Sharma, closely echoing Ford’s words.
Sharma’s quest to make AI an everyday fixture is pretty high level, but the problem she and Sage face in convincing small businesses and entrepreneurs to take a punt on a new product is a universal one.
“When you’re explaining a new idea, it depends on what audience you’re speaking to,” says Sharma. “If you speak to consumers, it’s a different way of communicating. It’s so easy to lapse into jargon. I explain Sage’s bot as: ‘oOur bot is going to turn into a personal trainer for your business’.
“It’s about the utility and the problem it will solve. That’s the best way to describe any new product.”
In a recent piece for Wired, Gaurav Jain and Joe Flaherty argued that we’re entering an era known as ‘big software’.
“[T]he epitomes of ‘Big Food’ and ‘Big Pharma’ were once humble startups… Over the next hundred years, we could see the same thing happen with the most high-minded tech of tech companies. Google, Amazon, Apple, and Facebook could grow to dominate the market in the same way Pfizer and Kellogg’s have dominated theirs.”
This is the environment that entrepreneurs like Sharma and others now operate within.
Sage is the embodiment of big software. The company was an early innovator in accounting software and now casts a formidable shadow over the accounting software industry, particularly in The UK. For a lone entrepreneur like Sharma, it’s easy to be intimidated but, instead, she saw it as an opportunity.
Sharma built her first bot at age 18 - it dispatched chocolates from a snack bar - and only graduated from St Andrews in 2011. The opportunity to accelerate her vision and to do what she loves.
In return for her innovation, Sharma says Sage has granted her complete control. The attraction for her is obvious: She can now integrate her bot into a large portfolio of products that are well used.
That, according to Sharma, is the big promise of online platforms and patronage by larger firms: audience. “There are 2.5 billion people that have a messaging app, online messaging and interaction is a social context we’re used to now,” Sharma says. “All entrepreneurs should analyse how their creations can plug into existing platforms.”
The internet boom has settled to an extent and the big players have settled into their hegemony. A key for them now is looking for new ways to augment their offerings. It was that, which drew Sage to Sharma.
Keep it contained
Decades of sci-fi literature has deeply inflected how we view concepts like artificial intelligence, bots and deep learning.
More so than most, it would be very easy for Sharma to lapse into the old entrepreneurial mistake of biting off more than she can chew. But Sharma studiously avoids over reaching.
Sharma was lured by Sage’s offer because she views accounting as an excellent use case for artificial intelligence. Sage not only offered Sharma large resources to pursue her vision, but a way to channel her work into a neat, pre-ordained space.
“A business has to create an expense. The problem we’re solving is quite contained,” she says. “It’s not solving a philosophical question; the meaning of life and the universe or something like that.”
Sharma’s bot – it has a name, but Sage isn’t ready to release it – is focused. “Sage’s bot says: ‘I’m here to solve problems with accounting, this is my domain’.”
The best way to think of Sage’s bot is something akin to Apple’s Siri. The user interface is very plain, housed on your smartphone screen. The user speaks in clear, natural language and the bot captures and notes it. The bot will also analyse a business’s data and provide KPIs and reminders.
A bot is a bot
The traditional standard for artificial intelligence is the Turing test. Developed by the Alan Turing in 1950, his test would place an evaluator in a conversation with two unseen partners, one human and one machine. The machine passes the Turing test if the evaluator cannot distinguish between its responses and the other human’s.
The Turing test has been the stock standard since then, but it’s an idea that Sharma sees as outdated. For her, a bot is a bot – and users that have tested it are absolutely fine with that. “A good bot doesn’t need to pass a Turing test,” argues Sharma. “It needs to pass the beer test.” By this Sharma means, would I have a beer with this bot if it were a real person?
The beer test sounds whimsical, but it has remarkable implications for our age of office automation. The application of the beer test to AI was introduced by Slack’s Amir Shevat. In a post on Medium., Shevat wrote: “[I]t is time to take the step forward, provide your service as a bot - not an impersonator bot, but a friendly and approachable bot.”
Sharma agrees: there is no need any longer to conceal the fact that a bot is a bot.
This requires a cognitive shift among web and app startups in particular. The hunt for a machine’s humanity is a futile one, argues Sharma.
Instead, Sharma chooses to embrace her bot for what it is. “Interacting with a machine is becoming more acceptable,” says Sharma. “You don’t have to hide the fact that it’s a bot. It’s okay to not pretend to be a human. A good bot is one that introduces itself as a machine and clearly sticks to a predefined role.”
Interested in reading more about how startups evolve into scale-ups? Sign-up here to get three emails a week with the best content for entrepreneurs.