AI Funding Weekly - August 13th 2023

Fresh AI funding news, delivered weekly

Hello, happy Sunday, and welcome to the next edition of the AI Funding Weekly Substack. I have been experimenting with different days to publish my Substack so this week I’m giving Sunday a try.

I’m also going to shorten my preamble at the beginning of the Substack as my goal really is to make this as digestible as possible. That being said, I’m not doing away with it completely as I do like to highlight one topic I think should be on everyone’s radar from the AI world each week.

This week I wanted to highlight the continuing buzz around how often ChatGPT is wrong. Coverage of the topic started in July after Fortune covered a Stanford study that proved the world’s most popular LLM can get worse over time. Now The Register has jumped in to join the party with a new article highlighting a study from Purdue University.

"Our analysis shows that 52 percent of ChatGPT answers are incorrect and 77 percent are verbose," the team's paper concluded. "Nonetheless, ChatGPT answers are still preferred 39.34 percent of the time due to their comprehensiveness and well-articulated language style." Among the set of preferred ChatGPT answers, 77 percent were wrong. (Source - The Register)

My take? Nobody should be surprised by results like this. For some reason it seems like people expected OpenAI to build an Oracle that somehow is magically always right. The reality is improving prediction accuracy takes two very valuable resources, data, and time.

Ten years from now we’ll have infinitely more data than we have today, but for now everyone needs to realize, LLMs are in their infancy, let’s give them time, and data, to learn rather than expecting them to do everything perfectly out of the gate.

Okay, enough from me, let’s get to the good stuff - here are three awesome startups that just snagged a fresh round of funding.

---

Virtualitics - $37M Series C

Companies are generating more data than ever before but many have a challenging time getting insights from their data. Traditional data tools often require teams of data analysts to extract insights from the data and then visualize those insights in a way that makes sense to the business side of the house.

Since 2016 Virtualitics has been introducing some pioneering innovations to the data visualization space by going beyond simple visualizations and actually helping companies find valuable insights in their data.

After getting some new government contracts under their belt the company saw a 370% increase in annual revenue leading to this most recent round.

---

Wint - $35M Series C

Let’s face it - leaks are a nightmare for everyone and Israeli startup Wint is using AI to help detect and stop leaks. Started twelve years ago Wint has moved into rocket ship mode relatively recently after raising a $15M Series B round last year and revealing an impressive customer list including juggernauts like Dell, Mastercard, and Google to name a few.

“We’re excited to close this round at a time when water scarcity and climate change are becoming some of humanity’s greatest challenges, while the costs of water leak damage in buildings are reaching unacceptable levels for insurers, owners, developers and contractors” (Alon Geva, CEO)

The connection between software and hardware, and way in which AI will become more pervasive in controlling physical systems is an area I think will continue to grow exponentially over the next decade. Rewind ten years ago and most companies would never trust AI making decisions that impact physical systems, now huge Fortune 100 companies are relying on them to do just that - the times, they are a changing.

---

Weights and Biases - $50M

While $50M might sound like a big number (okay, it is a big number), this actually represents a much smaller round that Weights and Biases raised in it’s previous round which clocked in at a whopping $135M.

So why all the dough? Weights and Baises has an all-star founding team who spent years working on tools for machine learning engineers and data scientists. They’re also in the sizzling hot MLOps space.

I look at MLOps a lot like selling pick axes and blue jeans to gold miners back in the 1800’s - there’s a boom going on and this is the space that’s providing critical tooling to the people in the boom. While not every company in the AI space will succeed, they’ll all need some form of MLOps to do what they do and with a client list that includes companies like OpenAI, it’s safe to say Weights and Biases has become a leader in the space.

“Weights & Biases is the leading machine learning platform to help developers build better models faster,” Shukla said. “We build lightweight, interoperable tools to quickly track experiments, version and iterate on datasets, evaluate model performance, reproduce models, visualize results and spot regressions, and share findings with colleagues. This lets machine learning engineers quickly iterate on their machine learning pipelines with the confidence that their datasets and models are tracked and versioned in a reliable system of record.” (Source - TechCrunch)

---

As always, thanks for reading, and see you next week!