Is There Value in Venture Scanner’s Iffy AI Funding Research?

I was drawn in by Venture Scanner’s pitch — “we are tracking 855 AI companies across 13 categories, with a combined funding amount of $8.75 billion” with “a unique combination of our data, technology, and expert analysts” that “enables corporations to research, identify, and connect with the most innovative technologies and companies” — but judging from a closer look at one of the sectors covered — a category where I have some expertise — their research is polluted with outdated and misclassified information and it is quite narrow.

I’m left with the question: Is there — can there be — value in Venture Scanner’s iffy AI funding research?

What’s iffy about it?

My own expertise is in the Natural Language Processing market — check out my March, 2013 NLP: Everyday, Analytical & Unusual Uses — in research, technology and solution providers, and business adoption. That’s why I took a close look at Venture Scanner’s Artificial Intelligence investment teaser, which splashes a couple of hundred (guessing) company logos across 13 AI categories.

I have no issue with the categories, although you should keep in mind that we’re looking at a categorization of invested-in companies and not of tech developers globally, which include academic, government and industry research labs; publicly traded companies; open source projects; and companies that have not taken venture funding. Some of the most interesting and important AI comes from these sources! So point #1 is that the narrow provider-market survey gives a possibly-distorted, and certainly very incomplete, impression of the overall AI investment scene.

Let’s look closely at one sector, an NLP-Gen (presumably for General). I’ll include a snip, citing fair-use rules for copyrighted materials:

Venture Scanner's sampling of funding recipients in the natural language processing-general category
Venture Scanner’s sampling of funding recipients in the natural language processing-general category

Fourteen logos are displayed, presumably representative of the 130 companies Venture Scanner is currently tracking. Venture Scanner says, “The… companies either build natural language processing technology or utilize it as the core offering in their products.” Note the use of the present tense. Yet of those fourteen:

  1. ClearForest was bought by Thomson Reuters in 2007 and no longer operates independently. Venture funding dates back over a decade, I’m guessing.
  2. Cognition was acquired by Nuance in 2013 and no longer operates independently.
  3. TopicMarks was acquired in December 2011
  4. DigitalTrowel died in 2013/4.
  5. DataRPM has a natural language query interface, but the company isn’t a “general” NLP company and doesn’t sell NLP tools.
  6. SwiftKey likely uses NLP, but it similarly is not an NLP company.

So six of the fourteen companies shown don’t belong. (I don’t recognize two other of the fourteen logos, the robot head and the desk chair.) On the surface, Venture Scanner betrays the shallowest attention to (at least one of) the sectors they are blindly tracking. Extrapolating: If I were to review the full set of 130 companies assigned to the NLP-Gen category, would I find 57 mis-inclusions?

But back to my question, Is There Value in Venture Scanner’s Iffy AI Funding Research?

Despite my criticisms — despite the outdated and misclassified information and the narrowness of the overall exercise — my answer is Yes. For me, the value was in learning about Delver, which provides natural language interface capabilities, a company I hadn’t known about. I am sure I’d learn about quite a few other, newer companies I hadn’t known about. In a dynamic technology scene, it’s impossible to keep up with every development. And I could see where Venture Scanner’s data — the tally of cumulative, combined AI funding over the years, presumably broken out by category — without regard to outcome (Does the funded company still exist or was it acquired or did it fail?) would be of interest to someone trying to understand the venture capital big picture.

But others — for instance if you’re looking for an NLP (or other AI) solution or if you’re a tech company or investor in the market for an acquisition — should realize that there are no competitive research short-cuts, that even “a unique combination of… data, technology, and expert analysts” is not guaranteed to provide accurate, useful market insights.

One thought on “Is There Value in Venture Scanner’s Iffy AI Funding Research?

Leave a Reply