CognitionX founders reveal how they’ve built an AI

Artificial intelligence has become big business, but many of the investments in the technology don’t pay off. Gartner expects the global business value derived from AI to reach $3.9 trillion in 2022, but the analyst firm also estimates that 85 percent of big data projects fail.

CognitionX founders Charlie Muirhead and Tabitha Goldstaub believe their startup can change that. The company provides an AI advice platform that connects specific queries to specialists who may be able to solve them.

“We felt that there was a need to democratise access to good advice on technology adoption with AI as a focus,” Muirhead tells Techworld from a corner of CognitionX’s headquarters at the Wayra start-up incubator in Piccadilly, central London.

“Not just to make sure that there’s good advice, but to make sure that that advice results in safe and responsible deployment of these very powerful technologies, and that the benefits get shared amongst everybody and don’t end up manifesting with a very small group of people.”

The system functions as an advice marketplace that matches questions to domain experts in real-time, a little bit like Quora, who then answer them in private and provide written responses and links to resources. Inquirers can speak to multiple experts and talk off-network whenever they want.

The basic version of the product is entirely free, but experts can use their advice as lead generation, while an enterprise version that creates a custom-built internal expert network for clients is available as a paid-for subscription.

Queries are matched to respondents based on their expertise, availability, and preferences, whether they’re a small startup looking for potential buyers, to pre-sales technical employees whose sole job is answering leads.

“The experts get to decide if they’re going to answer it for free because it is a good sales lead, or if they’re going to charge for it,” explains Goldstaub. “You can then start to create a business model for these experts, whether they’re academics who don’t get paid for their expertise, or they’re data scientists or consultants who could get additional income.

“You’re creating a whole new way for these experts to monetise their knowledge, and that in itself is really important if we’re moving into a knowledge economy where people care so much about insight and information – more so than the products in many cases.”

Founder stories

At first glance, Muirhead and Goldstaub seem an unlikely partnership, but their track record suggests that they make a complementary combination.

Muirhead, 43, is a serial digital entrepreneur dressed in a crisp white shirt, while Goldstaub, ten years younger, is a dynamic networker and communicator dressed in a novelty Christmas jumper.

Both attended the illustrious Bedales School in Hampshire, but their subsequent careers followed different paths. An 18-year-old Muirhead helped buyout Music Bank, an equipment rental business, before joining Imperial College to study computer science.

He dropped out after a year to start Orchestream, which floated on the London Stock Exchange and Nasdaq in 2000 with a market cap of £1 billion. He went on to found data communications company NexAgent, web companies incubator iGabriel, and online video network t5m, which is where the partnership with Goldstaub began.

Goldstaub had studied graphic design and advertising at the University of Arts London, and was introduced to Muirhead by her mother Jane Procter, the former editor of Tatler magazine. Goldstaub went on work for Muirhead as a video syndication director at t5m, where she noticed a need to simplify the syndication and monetisation of live and on-demand video.

Just two years later, the duo cofounded Rightster, a video rights network now known as Brave Bison, which is where the idea for CognitionX emerged. They wanted to build data science tools for the service, but couldn’t make them work despite having the support of 100 software engineers employed by the company.

Some self-described experts offered help but much of it was snake oil.

“People were charging us £10,000 a month for stuff and we didn’t even have the data lake set up,” remembers Goldstaub. “If I could have messaged someone asking should I use this product, I’m pretty sure someone would have said no. But I didn’t have anyone to ask.”

The duo tried to learn how to operate by inviting experts to dinners where they’d discuss the state of AI. They organised 100 such events over the next two years, which evolved into CognitionX: an AI advice platform and curated news service.

“We built something for ourselves and now everyone else wants it, which is great,” says Goldstaub.

Internal affairs

Many of the enterprises they spoke with said that a more urgent problem than accessing external advice was finding the right person in the company to answer their questions.

“There’s a quote from the 1980s by Lou Platt when he ran HP: ‘If we only knew what we knew, we’d be three times more productive.’ That was out of frustration that at one of the most amazing tech companies in the world you could never find out who to go to if you had a specific question,” says Muirhead.

“In the enterprise version of our project, we create an internal expert network for the client, and when they put a question in it finds the best fit internal expert with a percentage match and the best fit external expert simultaneously. And they can pick one, the other or both to get advice.”

That product costs £10 per month per active user, a business model that recently attracted $4.1 million of funding from investors including Unilever Ventures and the CXO Fund.

Early customers include consumer goods giant Henkel, which will roll out the enterprise edition of CognitionX to access a Wiki of AI products, the company’s existing internal AI knowledge, and a network of external AI experts, all through a single platform. Henkel will also syndicate curated AI news from CognitionX’s handpicked updates.

HSBC is another early partner. The global bank is working with CognitionX to understand how leading AI firms can deconstruct data and build truly scalable machine learning software.

“The challenge is how do you capitalise on your internal aspects that you do have and further develop their skills, how do you simplify access to external expert advice that you otherwise would have to go through all kinds of procurement processes to do, and how do you continuously engage and upscale your workforce, which is a challenge for all organisations,” says Muirhead.

“The way we do that is the news service we offer is free. You can set up as many of your users as you want to get access to that, and we’re adding personalisation to it too. And when you click on an article, if you’re an enterprise subscriber it will say this is how many experts internally or externally could answer a question about the topic of this news article.

“If you click on it you are then directly connected to that expert… They might already be doing something in that space, or they might not, or they might have looked into it but stopped. It’s how we create a much tighter loop between people being inspired about how they can apply technology to solve problems and actually finding the internal or external expert to give them advice on how they do that.”

AI future

Muirhead and Goldstaub still attend multiple tech meetups every week, particularly in Goldstaub’s government work as the chair and spokesperson of the AI Council and AI Business Champion.

She is also the director of CogX, CognitionX’s AI festival, which attracted more than 6,500 attendees in June 2018 of the AI community to discuss the impact of AI, and led the CognitionX team commissioned by the Mayor of London’s office to write a report on the capital’s AI scene.

Their research led her to conclude that “London is the AI capital of Europe,” with 758 companies – more than Paris and Berlin combined – and investment that passed £200 million in 2017, a rise of more than 50 percent on the previous year.

London also has more diversity than the rest of the tech industry, with 45 percent of AI startup founders born outside the UK, while 25 percent of them have a female founded, a figure that drops to 17 percent among all global startups.

The report attributed the capital’s success to growing funding, access to talent, connected community and a research ecosystem that helps Europe publish more AI research than both the US and China.

Muirhead expects that in future AI will no longer be considered a differentiator for the city’s startups, as the technology will become pervasive in their products.

“You won’t necessarily announce yourself saying you’re an AI startup, you’re just a startup that solves this problem,” he says. “And of course you’re using machine learning, because if you aren’t, what planet are you living on?”