SA machine learning startup Invoke Analytics looking to expand to Europe

Barely three years ago Herman Carstens and Dawie Diamond were forced to rely on their spouses’ incomes and left over bursary money from their days in academia to fund their Pretoria-based startup Invoke Analytics.

But things are finally starting to look up for the startup — which has had three approaches from investors since it was founded in 2016 and is now looking to expand to Europe.

So, says, Herman Carstens, the co-founder and chief data scientist of Pretoria-based data science and machine learning consultancy (pictured above, far right with — from left to right — Diamond and the startup’s two developers Takalani Sigama and Keanan Jones).

Carstens and Diamond  established the venture shortly after they finished their respective PhDs.

Invoke Analytics has developed machine learning products for use in the financial services, legal and energy sectors

“Our rates are incredibly competitive in South Africa, and even more so in Europe and the US. Our technical offering is also on par with competitors overseas,” he told Ventureburn.

Invoke Analytics was founded in 2016 by Diamond. Carstens joined in 2017. The startup employs two software engineers.

The startup — which has been part of the University of Pretoria tech business incubator TuksNovation since November — currently builds solutions for clients and uses the profits to fund the development of its own products.

These products include a retirement planning tool (RetireLab), a product that analyses documents (InvokeDocs), a product that automates calculations needed when verifying savings in large energy efficiency projects (Auto Measurement and Verification Tool) and a solution that helps firms to increases revenue from debit-order collections (PayLift).

In addition, to generate revenue the duo worked on projects such as building machine learning models for clients such as retailer Spar and King Price Insurance.

“Since then our turnover has grown to about R2-million per annum, but this figure is very dependent on contracts,” explained Carstens.

The startup has been bootstrapped since launch as the founders have not accepted funding as yet.

“We have turned down three equity offers. That is not to say that we would not accept funding from the right partner,” said Carstens.

He explained that the duo had declined the offers as they were not “favourable” and that he and Diamond preferred to bootstrap rather than take up an offer that they would eventually regret.

He wouldn’t disclose the names of the investors behind the offers, but he said they are SA companies with an international footprint.

Corporates wary of machine learning

But he says corporates, still see machine learning as an “unproven” technology, and that while there has been a lot of hype around it, the vertical is still viewed with a certain bout of scepticism by corporates.

The situation is further compounded by the state of the economy. Carstens said because of the downturn, companies are “sitting on cash” and don’t want to spend it on something they might perceive to be “high risk”.

The general sentiment in the industry at the moment, he says, is not to take on free proof-of-concepts (POCs) or trials.

“Perceived value is co-related with paying for a product or service, and therefore free POCs are not often done by companies in the industry,” he explained.

Startups, he pointed out, now prefer an alternative approach where they build a POC for free and then take five percent of all profits generated from the solution.

The problem there, Carstens said, is that it is difficult to prove how much companies would likely generate from such a trial.

But he admits that things are improving and points out that in the last few months potential clients have approached the startup with some “interesting”, paying projects.

The question now is will European clients bite?

Featured image, left to right: Invoke Analytics CEO Dawie Diamond, Software engineer Takalani Sigama, UX and UI engineer Keanan Jones and COO Herman Carstens (Supplied)