Making AI easier to understand and develop

Why we invested in Tenyks

A thirty-minute meeting turned into a funding decision

For years we have been listening to “black box” models in machine learning. This has been preventing the adoption of different algorithms in many real-life use-cases, where you need to observe how AI thinks and derive conclusions. Moreover, the explanation of how decisions are taken in a neural network might turn into a regulatory requirement in specific industries, like healthcare or autonomous vehicles.

With all this in mind we were thrilled to enter a meeting with Botty Dimanov — CEO and co-founder of Tenyks. Thirty minutes into the conversation, which was scheduled as a “get to know each other”, we had committed to invest in the company.

Tenyks approach to data explainability and AI model improvement

In order to understand what Tenyks does, we must look at them through the lens of their customers. In one of the early implementations, the software has identified that the majority of the cases with algorithm failures are resulting from wrongly labeled data — due to its origination from one singular place with a specific way to annotate images. In other words, Tenyks helps computer vision developers to improve the success of their models by helping them discover how data impacts the algorithm output. As Botty says, “Engineers get frustrated wandering in the dark, trying to figure out if the change they made will improve performance, and not having the insights needed to know what to do next. We remove this stress, making the process a breeze.”

If you look at trends, Tenyks could be described as Data-Centric AI meets Explainable AI 🤨 🧐. In broader classification it is a ML Ops company aiming to turn the most boring part of a machine learning engineer’s job — manually sifting through data to improve the success rate of their AI — into the pinnacle of their day. This allows for building more reliable software, faster. Dmitry Kazhdan a co-founder and CTO explains the benefits, “Gradually, in the same way coding is now a required subject in schools, AI programming will be a commonplace skill empowered by Tenyks products.”

University of Cambridge comes to YC 👨‍🎓+ 💻

It is a challenging problem with significant benefits, but no better team to solve it. The three cofounders Botty, Dmitry and Maleahki graduated the University of Cambridge. During his PhD there, Botty has come up with some patent pending research in explainable AI, which is the foundation of Tenyks. I remember discussing over dinner, the different life-paths he was considering — being a researcher vs. being a founder, and I’m extremely happy he has taken the second one. Dmitry and Maleahki have built upon the research and allowed the product to come closer to customers.

With such a stealer team it is no surprise that the Cambridge University is a co-investor in the round. Another great moment into the company’s short life was going through YCombinator in 2021. This allowed the team not only to excel in domain knowledge, but to sharpen their skills in company building and fundraising. The success in the latter came shortly afterwards — with $3.4 million in seed funding from great co-investors as Speedinvest and First Minute Capital, Tenyks is very well equipped to further strengthen the team, double the engineering power and get more customers.

Tenyks is here to protect the World from the Terminator

Of course, the short-term goal is to get the solution in the hands of more machine vision companies and help developers to easily interact with data, enhancing the models’ full potential. One of the early adopters is the digital health industry — more specifically companies detecting diseases with machine vision. In reality the product could be applied to any company with a complex neural network powered by numbers of images. In the long run, Tenyks aims to develop an ecosystem of products designed to make the job of a machine learning engineer easier, while reducing the learning curve necessary for developing, understanding, monitoring, and auditing AI.

Gaining more visibility on how computer algorithms “think”, is a way to push forward in an ecosystem, where we benefit from machine capabilities, and make sure that we get to a very peaceful AI singularity. This is the core of the company mission.

Shameless plug

* An open call for all great engineers, which wants to work on challenging tasks in a great environment and help the way humanity interact with AI.

** All my friends in the machine vision industry would love to hear your thoughts and experience with the problem.