Ciao 👋 Venture Capital (for now), Hello Timeseer.AI

Get Ready to Roll — My New Startup Is on Its Way

After the TrendMiner (acquired by Software AG) exit, I worked in Venture Capital for 3 years as one of the partners of Smartfin Capital, a Belgian B2B technology fund investing across Europe with nearly 400M€ AUM. During this period, I had the opportunity to investigate literally thousands of interesting companies resulting in some outstanding transactions such as Roamler, Recharge.com, Betty Blocks, and many more. A large chunk of my time though was spent hunting for the ideal target within the machine learning, analytics, and data (MAD) and IoT spaces. But I just couldn’t find that one company that grabbed all my imagination and interest, a company that really stood out. The logical solution then was to establish my own “ideal company” and begin another startup and to do so with some of the finest people I had worked with, in the past. Many thanks to Niels Verheijen, Thomas Dhollander, Stijn Meganck, Jeroen Hoekx, Yorick Bloemen for joining me again and welcome Frederic Hanika as our M&A advisor and Jeroen Van Godtsenhoven (former GM EMEA C3.AI) as our Chairman. And here’s where I am today, with a new startup, with my new baby: Timeseer.AI.

Data Observability — a New Breakout Segment

Back in 2019, I predicted that the hype train about AI/ML would mature. What I mean is that it would calm down resulting in fewer early financing rounds but more aggressive later stage rounds and that a refocus would happen on the infrastructure. This was confirmed by the 2021 MAD landscape of Matt Turck (VC and partner at Firstmark). I was excited to see infrastructure gain momentum, and it was also great to see “data observability” pop-up as a breakout segment. (I hinted that this would happen in my 2020 article on AI Sobering.)

And why was I happy about seeing “data observability” as a breakout segment? Well, because there is a serious need for this. Let me explain. One main observation I made during my Venture Capital days was that many companies (80% and more) wanted to leverage their data and experiment with AI but were stuck at the PoC’s. A common challenge that kept coming up in conversations with Fortune 5000 customers was that the reliability of their data was the main reason their AI projects were failing. Sure, this makes sense because “garbage in equals garbage out”. Moreover, the negative impact of unreliable data has never been as significant as it is today. For me, it’s best to frame and explain this problem at 4 levels:

• Company Level: Every company that is leveraging its data is vulnerable to data downtime especially if companies push employees to become data literate, so they can do more with data. In that scenario, trust in the available data is critical because the impact of data downtime on downstream analytics is significant. If this issue goes undetected, it is extremely difficult to repair.
• Data Level: In itself the hygiene of the data quality today is unquantified, so data can be both an asset and a liability. It’s an asset when it can be used for data-driven decision-making and operational improvement. However, it becomes a liability when the data provides incorrect insights that can lead you to make bad decisions. Unreliable data also causes failing business optimization.
• People Level: Data teams are aware of this problem and the severity of this problem and spend a lot of their time (up to 60%) on cleaning data. This job is a manual, labor-intensive, and repetitive task. It’s expensive too. More dangerous is the fact that most of the data consumers are unaware or ignorant of data integrity issues during their analytics journey. This can lead to wrong insights and thus wrong decisions.
• Process Level: Currently the quality of data is cleaned at the end of the data pipeline (i.e., at the end of the analytics project), but this is a one-off job and as such is not sustainable. A better approach is to be more proactive as close as possible to the data source (edge, fog, cloud) to assess the quality of the data early on.

So, what can be concluded? If companies want to truly shift to data-driven decision making and advanced automation and get AI in production, then as a prerequisite, these companies need the ability to work with reliable trusted data. Given this, the timing of Timeseer.AI makes perfect sense. Since first you needed the big unlock that made it possible to store and process data at scale start building data-driven applications that are mission-critical before as an enterprise you began worrying about observability and the quality of your data.

How We Developed Our Business Model

The four levels of the problem I explained above basically apply to every company working with and analyzing data. And many of the new data quality and data observability players focus on relational columnar data intertwined in a very complex and fragmented data stack. Additionally, these companies sell to IT whereas we sell to OT (operational technology). However, I believe in focus and verticalization to get your product/market and as such we developed our business model accordingly by focusing on one type of data: time-series data.

Now for some interesting facts about time-series data:

· It is also referred to as time-stamped data and is a sequence of data points indexed in time order where the data is collected at different moments in time. These data points typically consist of successive measurements made from the same source and are used to track change over time. Examples of time-series data include sensor data, DevOps monitoring data, Weather data, financial pricing data.

· It is a difficult data type to handle, and causality plays a role here. Many data quality expressions (metrics) are special purpose-built for time-series and are not relevant for relational data. Additionally, vendors in the data quality space have developed tools that are not fit for the purpose of framing time-series data.

· It is a huge market with data coming mainly from IoT sensors, financial pricing data, DevOps monitoring data, and operational analytics and often is not the focus of horizontal IT vendors.

· It is the fastest growing database segment. Made-to-measure or special purpose databases were in high demand, but time-series databases as a category is leading this race.

In addition, data pipelines in OT /IoT are entirely different from the data pipelines we find in the data warehousing domain. The streaming element plays an important role with technologies like KAFKA.

In the end, the cost of bad time-series data quality can be devastating. For example, using unreliable data can result in abnormal situations and unplanned downtime, impact safety (and in some cases lead to explosions), and cause issues with governmental compliance and reporting, all of which is unacceptable and costly.

Timeseer.AI in the Middle of a Perfect Storm

Whenever I’m reviewing a company (with my VC hat on of course), I look for four fundamental elements to evaluate the potential success of a company: timing, team, market size, and product (in that order). And for me, Timeseer.AI ticks all these boxes.

The timing is perfect because organizations that want to stay competitive and succeed in the digital transformation age are shifting to data-driven decision making and hyper automation. And data reliability is crucial to this success. In fact, a lot of prominent investors are making considerable bets in this space. Our Timeseer.AI team has 80 years+ of cumulative experience in this domain due to our TrendMiner roots and most importantly, we are a group of passionate, highly driven, serial entrepreneurs with the know-how and motivation to make this happen. Simply put, we have the best team to pull this off. The time-series data market is huge and will continue to grow. This is proven by the fact that time-series databases are the fastest growing segment. Lastly, we are currently building the best product on the market that will empower data teams with quality augmented data through an AI powered time-series data observability platform.

Shooting to Become #1 for Time-Series Data Quality/Observability

Ultimately, it all boils down to this: we have the ambition to become #1 for time-series data quality/observability. With $6M SEED capital to give us oxygen for growth and more than a dozen Fortune 5000 customers, we’ve already established significant traction in the market.

Hey if you want to be part of this rocket ship 🚀, feel free to drop me a direct message and until then, stay tuned.