Can AI be used to intelligently rank expertise? Serial entrepreneur Peter Relan thinks it can, and has today launched what he’s billing as a “knowledge-as-service” (KaaS) web platform to connect (human) experts with professionals seeking the answer to specific tech tool-related problems.
In the first instance the target users are professionals wrangling Excel and Google Sheets, with the platform aiming to link them (via their spreadsheet-related problem) to a relevant expert — with the help session taking place via the medium of web-based text chat.
But the wider ambition is to be able to use its expertise-ranking algorithms to intelligently assess all sorts of techie knowledge to be able to connect relevant experts with different professional “knowledge based problems”. Relan likens the concept to how Google’s pagerank dynamically orders online data.
“A page’s rank can drop or get better depending on other pages who may be pointing to the page and their ranks. We use that idea by enabling any expert in a topic to audit the work of other experts in that topic and using their ranks to influence the score of the audit. The difference is it doesn’t take months to “crawl and re-index”. Expert rank is always working to update based on the last audit,” he says.
“We’ve recently received a patent application for this approach. It’s really the first algorithm that combines the link analysis structure of page rank with gig economy measures of performance. If an expert does not work for a long time they may well have to take the onboarding test again in their topic and start with a seedrank.”
The newly launched KaaS platform, which is called Got It Pro, expands on the team’s first app (previously called Got It, now renamed Got It Study). That earlier app specifically targeted students stuck on homework, offering 10-minute chat sessions to help them solve a math or chemistry problem, with help sessions priced at between $0.99 to $3.99.
Got It Pro doubles the chat help session time to 20 minutes to offer “on demand help” for professionals — albeit, only for spreadsheet-related queries in the first instance. Pricing is between $3.99 to $5.99 per session. The intent is to add more subjects and experts to expand the range of professional problems it can cater to.
Relan incubated Got It in one of his own accelerators, bootstrapping it with $120k to build the first app, before raising a $2.5M seed from Kinzon Capital in 2015 to start work on the now launched KaaS platform. The team has also previously taken in $6.4M from Capricorn Investment Group — bringing their total raised to date to just over $10M.
“We are initially focused on problems, not general advice. For example a question like “should I use the new histogram chart in Excel?” is a general advice question. But the question “I’m using countif to do a histogram in this sheet. Here take a look at it. Why is it not generating the right histogram?” is a pressing problem someone is working on,” he tells TechCrunch.
“We are focused on the second type of problem related knowledge. The first type are better for Quora or Stackoverflow. They have thousands of questions posted everyday. Some of the first type and some of the second type. So we will coexist with them but address those problems which are well-defined in any tools or technology or apps related field important to one billion knowledge workers over time.”
There are some 25,000 problem solvers on its platform with an “expertrank” at this point (which is its own algorithmic benchmark based on AI-assisted human analysis of each expert’s problem solving skills), though far more have applied — only around 10 per cent make it through the vetting process, according to Relan.
Some 2.5M 10 minute help sessions have been completed via the earlier Got It app, which has a user-based of 1.5M on iOS and 300,000 on Android — asking an average of two questions per week.
“[Experts] are professionals and teachers and retirees and students and analysts and product managers. We support STEM subjects in Got It Study and Excel and Google Sheets in Got It Pro today. New subjects are being prepared but we will share these as they launch,” he says.
“The main selling point [for users] is that we get you to an expert within seconds so that’s as fast as selecting a link on google and scanning an article. But in our case we also offer a guarantee. That in a 20 minute chat session your problem will be solved. Q&A included. Google articles can’t provide that. It’s a maybe. Same with forums.”
Of course, Googling answers may be variable but its main cost is time — so it remains to be seen how many people are in such haste to fix their spreadsheet nightmare they’re prepared to stump up cash monies for an ‘instant’ 20-minute fix.
As well as its proprietary expertrank algorithm, the other piece of key tech is the platform’s real-time marketplace management of its experts — so every time a user submits a problem it goes through a real-time global auction (taking about 30 seconds, according to Relan) to match them with a relevant expert who can answer their problem.
“First experts are notified of a new problem and they bid and then their ranks and bids are dynamically evaluated to find the most relevant expert at an affordable price,” he says. “Since we are not a physical logistics service we are not constrained by location. We could have hundreds of candidate experts bidding for a problem. So the marketplace management at scale for a relevant match in real time is the other core tech piece. Like running the NYSE but for knowledge problems albeit just a tad slower :)”
The team is planning to launch a freemium layer in future, staffed by chatbots (rather than human experts) — albeit it’ll be training these on its “problem solution corpus”.
“That will be free but a bit generic,” he adds. “So if you try that but still want to dive deep with a human expert with specific Q&A for a very specialized problem that’s when you go for a paid session.”