Insight Partners had a banner year in 2021, with over $50 billion in capital commitments in more than 200 investments. As we invest in founders across a wide range of businesses, this series focuses on presenting our theses on four vertical that we are particularly passionate about in 2022: artificial intelligence (AI), fintech, cybersecurity and health technologies.

Artificial intelligence has remained a very popular category for good reason: it has the potential to transform nearly every industry and business.

At Insight, we have long been optimistic about the many use cases of AI. Over the past year we have invested in image recognition software from the Dutch company ScreenPoint Medicalwhich improves early detection of breast cancer, Covera Health, which provides a quality analysis platform to reduce medical errors in radiology, MAP, that helps businesses use and understand spatial analytics, and laminara cloud data security platform to continuously monitor and protect against data leaks – among other game-changing companies.

In total, Insight invested in 49 different companies across a wide range of artificial intelligence and machine learning use cases in 2021, representing a 172% increase over the previous year.

As we look to 2022, we expect AI tools to continue to dominate. We see the ecosystem fall into three main categories:

  • Layer 1 – Core Platform Companies: Algorithms; picture frames; infrastructure and workshops for creating ML systems
  • Layer 2 – Interprofessional companies: Turnkey machine learning-based systems that solve specific problems spanning multiple industries (e.g. cybersecurity)
  • Layer 3 – Industry specific companies: Applications powered by prediction or classification systems that target specific niche use cases within a domain

Businesses and investors will find AI/ML software valuable across all three layers. At Insight, we first focused on layers two and three. We have invested in startups creating robust ML systems that solve specific problems, either vertically (like credit underwriting company Zest AI) or horizontally (like cybersecurity company SentinelOne). We thought the economic moat was the hardest to build on level one; in part thanks to robust open source ecosystems and because large public cloud providers offer many of these tools at low prices.

Insight Partners’ Three-Layer Framework for Machine Learning Enterprises

Throughout this year, however, we realized that we were missing a piece of the puzzle. While there is always market demand for cross-industry capabilities and industry-specific applications, 2021 has taught us that as companies attempt to produce AI for the first time, platforms will also become increasingly important.

After all, more companies than ever are embarking on AI projects in an attempt to guide business decisions and save time and money. According to a recent survey, 90% of companies are actively pursuing AI projects or planning to do so within the next 12 months.

But here’s the catch: most these projects will fail.

There are many reasons for this, including a lack of sufficient data or black box models that spit out indecipherable results. Whatever the reason, high failure rates can harm the ecosystem if they make companies reluctant to move forward with future AI projects.

As a result, Insight is particularly excited about a subsection of AI tools that can significantly increase the likelihood that companies will successfully achieve their project goals: Machine Gain Operations or MLOps.

MLOps tools can improve a machine learning pipeline from start to finish by helping to collect, manage, and label data, experiment and test model selection, deploy multiple models to production at once, and protect against drift and model and data attacks.

Overall, the goal is to improve communication and collaboration between data scientists, data engineers, and business analysts throughout the machine learning lifecycle, the same way DevOps tools help improve communication in the software development life cycle. Machine learning is an iterative process with constant feedback loops and the need for continuous monitoring, which makes MLOps tools to manage this process even more critical.

Overview of some of the key areas of an MLops pipeline

While every business deploying AI projects can benefit from MLOps, the type of tools they need will depend on their needs, resources, and strategies.

Businesses need to ask themselves: how sophisticated is our data science team? How mission-critical do we want our models to be? Are our data sources structured or unstructured? Do we want open source, commercial or local tools?

The answers to these questions will direct companies to specific MLops tools best suited to their specific challenges.

For example, some platforms, such as Dataiku or DataRobot, are end-to-end products with extended functionality, which are generally easier to use and thus encourage model building and experimentation by workers of all functions, regardless of their technical knowledge. These so-called “citizen data scientists” can boost analytical workflows for companies with fewer data science experts and create significant value, but there is a risk that the models deployed are not as well understood. or controlled. These types of tools are best suited for companies that want to use machine learning to generate business insights and analytics rather than as part of their primary business function.

However, if a company has a sophisticated data science team, they can look to specific point solutions, or what Insight calls best-in-class MLOps solutions. While these products require greater technical knowledge to maintain and deploy, they allow for much more control, depth, and sophistication in a system. One startup offered the apt analogy that using platforms versus best-in-class solutions is like driving an automatic car versus a stick shift.

There are multiple MLOps tools for every part of the pipeline, and we see a world where every tool has enough market space to help it grow into a big business.

At Insight, we’ve spent 2021 learning and investing in many of what we consider to be best-in-class MLOps tools, as determined by feedback from highly satisfied customers, clear market dynamics, and well-respected teams and well informed. This prospect has led to our investments in Explorium, Rasgo, Weights & Biases, Deci, Run AI, Fiddler, Tonic, Dataiku and Databricks, along with several others that have yet to be announced.

Insight Partners MLOps Market Map

While Insight expects MLOps to continue to be a key driver of AI success in the new year, we also expect to see several other trends emerging in 2022:

  • First, we will see the emergence of two machine learning pipelines, one for structured data and one for unstructured data. Unstructured data sources, such as images, video, and audio, require specific data warehouses, data management, pipelines, and model management tools that maximize productivity and accuracy. Given the nuances and complexities between the two data types, we expect to see more MLOps tools move towards in-class ownership for one or the other.
  • As companies move from dozens of models to hundreds or even thousands of models, we expect increased industrialization of AI. This will mean increasing layers of orchestration that rely on machine learning pipelines to help manage all the different tools. These layers will integrate into multiple tools in the ML pipeline, whether open source, commercial, or in-house developed, and provide an integrated environment (or so-called “single window”) to better track and manage ML pipelines. Orchestration layers will help companies better control their pipelines while supporting best-in-class tools.
  • Insight believes we are at an inflection point in moving from a model-centric to a data-centric world. In a model-centric approach, you ask how you can modify the code of a model to improve system performance, whereas, in a data-centric model, you ask how you can modify the data to improve it. [system performance]. We expect the most powerful tools of 2022 to focus on supporting the new tasks, workflows, and jobs spawned by the data-centric movement.
  • We will also see a new class of MLOps tools focused on addressing the growing AI skills shortage and gap. through businesses. These tools will focus on increasing the efficiency of the model production process and lower the bar for ML development.
  • Finally, we expect a significant increase in the overall adoption of ML by enterprises, with a shift from experimentation to deployment of models in full production. Last year, we saw many companies forming AI research teams and hiring data scientists to experiment with machine learning. In 2022, we believe these companies will begin to realize the full potential of their AI projects and generate significant business value. This increase in enterprise adoption will also be driven by the explosion of peripheral devices that will enable new use cases of ML adoption.

As active investors in machine learning and artificial intelligence, we are excited to continue monitoring the market and supporting new entrants in what is sure to be another dynamic and exciting year.


George Mathew, Lonne Jaffe and Sophie Beshar co-wrote this piece.