AI firm Ozlo debuts knowledge base that contains over 2B data points on a wide variety of topics, available via three APIs that are priced by volume
Kristen Nicole / SiliconANGLE :
Context & Ripple Effects
Ozlo arrived in May 2016 as an AI helper bot backed by $14M from Greylock and others, explicitly built without human intermediaries, and moved out of private beta onto iOS and the web that October ($14M launch, iOS and web launch). The knowledge-base debut marks a pivot: instead of competing as a consumer assistant, Ozlo is now selling the underlying data itself — over 2 billion data points exposed through three volume-priced APIs.
That reframing matters because it turns Ozlo from an app into infrastructure, and within months Facebook bought the company to help build Messenger's personal assistant, shutting down the enterprise API and app (Facebook's acquisition of Ozlo). The API business was brief, but the acquisition validates what the pivot was really pricing: the knowledge base, not the chatbot.
First-order effects
- Developers gain programmatic access to Ozlo's 2-billion-point knowledge base across three APIs, with usage-based pricing replacing any consumer subscription model.
Second-order effects
- Selling the data layer rather than the assistant makes Ozlo an obvious component for platform companies building their own bots — exactly the position that led Facebook to acquire it for Messenger months later.
Third-order effects
- If the pattern holds, AI startups increasingly monetize curated knowledge as metered infrastructure while consumer-facing apps become distribution wrappers — a split later visible at the other extreme when AI2 open-sourced OLMo and its Dolma dataset rather than charging for access.
The trend: AI startups are shifting from shipping end-user assistants to licensing their underlying knowledge as volume-priced APIs, with acquisitions by platform companies absorbing the data layer.