“AI startup” can mean anything from a new way to search the web to software that drafts legal documents. These six young companies have taken different routes, building products for specific tasks rather than trying to be useful for everything.
Startups to know
- Sierra (2023): Sierra builds AI customer service agents that can respond to customers and carry out tasks for a business, such as helping resolve an account issue. Its focus is on completing a service interaction, not just suggesting a reply for a human agent.
- Cognition (2023): Cognition introduced Devin, an AI software engineering agent designed to handle coding tasks across a development workflow. The company’s work points to a bigger question for software teams: which parts of engineering can an AI agent reliably take on?
- Mistral AI (2023): The French company develops language models, including open-weight language models that developers can adapt and run in different settings. That makes it relevant to organizations weighing control and customization against the convenience of hosted AI services.
- Harvey (2022): Harvey is building AI tools for legal professionals, including support for research and document-heavy work. It is an example of a startup applying general-purpose AI to a field where accuracy, confidentiality, and careful review matter.
- Perplexity (2022): Perplexity offers a search and answer product that responds to questions with sources. Its approach puts citation-backed search answers at the center of the experience, rather than asking users to sift through a page of links first.
- ElevenLabs (2022): ElevenLabs develops voice-generation technology and products for audio, including synthetic voice and dubbing tools. The company’s work serves creators and businesses looking to produce spoken content in different voices or languages.
These companies are not interchangeable, and their products are changing quickly. What they share is a narrow starting point: customer support, software development, legal work, search, language models, or audio. For anyone following young AI startups, those specific use cases are often more revealing than the broad promises attached to the technology.
