Bitesize Impact 25: Cognotekt
July 13, 2018 Chris Sandilands
Cognotekt is an artificial intelligence company which helps businesses automate and optimise claims processing and customer relationship management (CRM).
The company is forecasting between €4m and €5m revenue in 2018, following over 50% revenue growth from 2016 to 2017. Somewhat unusually for InsurTechs, Cognotekt has received no private investment to date, although Jobst Landgrebe, Founder and CEO, says he is considering taking on funding to accelerate growth now. The company, which was founded in 2013, is live with three insurance clients in Germany and several non-insurance clients.
Cognotekt has three products:
- Konsequent: claims validation software which reviews, approves and flags inconsistencies in claims documents
- Eloquent: A tool for business process automation, including customer service processes
- Argument: software which processes long text, e.g. to compute a medical bill from a physician’s report
Cognotekt is rolling out Konsequent for motor repairs relating to glass (e.g. lights, wing mirrors) next month after a successful pilot in 2017. The tool creates a detailed assessment for each claim, eliminating human error and improving efficiency; Jobst notes a measured reduction in processing costs of up to 50% based on an automation rate of 90% and a reduction in indemnity spend of between 5% and 10%.
One of its new clients is German public health insurer, which has over 10m customers and c.1 m dental repair plans. Cognotekt software has a created a deep neural network to validate these repair plans. This has resulted in an automation rate of over 93% and a reduction in the number of reviewed cases from 35% to less than 10% – worth, according to Jobst, at least €2.5m in combined efficiency and indemnity cost savings per year.
Cognotekt’s current focus is on Eloquent. The tool can plug into clients’ own CRM systems and initiate actions. It is going into production with clients in the energy and debt collection sectors in the fourth quarter. Discussions with insurers are ongoing.
Jobst provided some interesting insights on the use of artificial intelligence in insurance in a guest post on the Oxbow Partners blog. In short, his hypothesis is that there are insufficient datapoints in insurance for a data-driven approach to machine learning, requiring instead a focus on natural language processing.
The Oxbow Partners view
When we spoke to Jobst around a year ago, he described how the first phase of any project was for his team to ‘listen’ to existing client processes so that they could be digitised. During this phase, his software developers would sit behind client staff and observe what they did.
An interesting development at Cognotekt is that more projects can now be executed using off-the-shelf software – a signal that InsurTech solutions are increasingly products rather than services.
But that’s an interesting point to explore: to what extent is InsurTech a product or a service?
InsurTechs clearly want to be seen as product businesses; having scalable software is the route to high valuations. But InsurTechs are still – to varying degrees – trying to find their product-market fit, which means that they must work with insurers to identify high-value problems to solve and collaborate with insurers to develop those solutions.
So perhaps insurers should view InsurTechs as service companies in the short term. The quality and fit of their existing product is arguably less important than the team’s momentum, capabilities and vision, not to mention their willingness to work in genuine partnership with the insurer to develop solutions.
And that raises a separate question around ownership of IP that is created in those collaborative processes. If an insurer and InsurTech jointly create a solution, who should benefit in what proportion from the commercialisation of that solution? We discussed a possible solution in a post last summer.
