At Index Ventures, we view the emergence of vertical SaaS (vSaaS) — cloud-based software tailor-made for specific industries — as a part of a broader trend of end users increasingly demanding superior technology products.
Consumers want solutions-oriented software made specifically to unravel their exact business problems. In an environment where we’re inundated with software, narrow and specific is well-positioned versus broad and generalized.
The concept shouldn’t be recent: Even the biggest horizontal tech corporations verticalize their sales organizations and product features once they have enough scale inside each vertical for that to be a wise approach.
Cloud giants AWS, Azure, and Google Cloud Platform prominently feature vertical industry solutions with dedicated sales teams, as do other large platforms like Salesforce, ServiceNow, Snowflake and Workday.
These tech leaders verticalize their offerings over time since it’s a high-quality experience for patrons and end users when a technology vendor deeply understands the industry, has sales and support reps attending the identical conferences as users, and is rapidly evolving the product to suit customer needs.
The AI category is rapidly evolving, but developing into three layers: foundational models, AI infrastructure, and AI applications.
With the AI platform shift upon us, we imagine that the following logical iteration of vertical SaaS might be vertical AI – vertically-focused AI platforms, bundled alongside workflow SaaS, built on top of models which have been uniquely trained on industry-specific datasets.
Why vertical AI?
The AI category is rapidly evolving, but developing into three layers: foundational models, AI infrastructure, and AI applications.
Examples of AI stack startups. Image Credits: Index Ventures
Foundational models are the bedrock of the AI stack. Leaders on this space include Anthropic, Cohere, and OpenAI. It’s likely there might be a limited variety of vendors within the foundational LLM space given the high capital requirements to construct and train models.
The “picks and shovels” of AI sit on the infrastructure layer, a catch-all which incorporates quite a lot of categories including data enhancement, fine-tuning, databases, and model training tools. For instance, vector databases like Pinecone and Weaviate are gaining significant adoption.
Other corporations like Scale are getting used for data generation, labeling, and training. Hugging Face has emerged as a pacesetter for model discovery and inference. Weights & Biases is well known inside MLOps. LangChain is an open-source development framework used to simplify the creation of latest applications using LLMs. These are a couple of of many corporations that are helping corporations transform models and data into products.
Foundational models and infrastructure are enabling an explosion of AI business applications. These AI-powered applications could possibly be utilized by any end user, in any industry, to perform an array of tasks.
