About Olib AI

About Olib AI: the company behind Owl Browser, a browser engine for AI agents

Olib AI LLC is a software company founded in 2023 in Stone Mountain, Georgia. Our product, Owl Browser, is a Chromium-based browser engine for AI agents and large-scale automation: 187 browser tools behind a REST API, Node.js and Python SDKs, a native MCP endpoint for AI agents, and a control panel. We host it for you, or you self-host it on Enterprise.

Founded in Stone Mountain, Georgia
2023
Current Owl Browser release
1.4.2
Browser tools over REST, WebSocket and SDKs
187
Isolated browser contexts per instance
256

The product

What Owl Browser is

One engine, three ways in: REST and the SDKs for code, MCP for agents, and the control panel for people.

Hosted for you, or self-hosted

Hosted for you by Olib AI (shared on Developer Trial and Developer; a dedicated 32 GB or 64 GB AWS EC2 instance in the region you choose on Starter and Business), or self-hosted on Enterprise. Owl ships as one Docker image: Chromium 156, the API server, the control panel and the MCP endpoint. On Enterprise you run that image on your own infrastructure, so page data, cookies and screenshots stay there.

Plans and deployment

Source-level fingerprint engine

Fingerprint controls are compiled into the browser: 27 C++ override modules and 66 Chromium/CEF patches, with 1,400 profiles across Chrome 143 to 156. No browser is undetectable on every site, so we publish our detection tests.

Detection benchmark

Built for agents

Agent rendering returns a page as OwlMark, a compact text view with a handle for every interactive element. A native MCP endpoint (POST /mcp) exposes 21 tools in the default agent profile and up to 82 in the full profile.

How agent rendering works

One API, two SDKs

Every tool is a REST call (POST /execute/{tool}) with an OpenAPI schema. The Node.js and Python SDKs (2.3.0, MIT licensed) are generated from that schema.

API reference

How we build it

Three engineering decisions

  1. 01

    Change the engine, not the page

    JavaScript shims that patch navigator.webdriver or canvas APIs can be found by inspecting prototype chains and Function.prototype.toString. We make those changes in Chromium and Blink C++ instead, so the values come from native code.

  2. 02

    Give agents text they can act on

    A screenshot tells a model what is on the page, not where to click. OwlMark gives it structured text and stable handles. On the Hacker News front page we measured about 1,813 tokens for the render against about 12,415 for the page HTML (Owl Browser 1.4.1, November 2026).

  3. 03

    Be plain about where data goes

    On Enterprise, the self-hosted plan, Owl does not send browsing URLs, page content, cookies or screenshots to Olib AI. On the hosted plans the instance runs on infrastructure Olib AI operates. License activation does contact our license server; the privacy policy lists what it sends.

The engine internals are documented on the technical page; release history is in the changelog and the engineering blog.

Leadership

The people who answer your email

Olib AI combines 13 years of institutional operations experience at JPMorgan Chase with 14 years of hands-on systems engineering. The founders who scope a deployment stay involved through integration and production support.

Tyonna Henry Matthews, Co-Founder & Chief Executive Officer of Olib AI

Tyonna Henry Matthews

Co-Founder & Chief Executive Officer

Former Vice President of Operations at JPMorgan Chase. Oversees global business operations, enterprise SLAs, vendor risk governance, and corporate compliance.

Akram Hasan Sharkar, Founder & Chief Technology Officer of Olib AI

Akram Hasan Sharkar

Founder & Chief Technology Officer

Former Lead Software Developer at Aloha. Principal architect of Owl Browser's Chromium C++ engine, source-level fingerprint virtualization, OwlMark, and MCP server.

Company details

Olib AI LLC · 4602 Birch Ridge Trail, Stone Mountain, GA 30083, United States
Email: hello@owlbrowser.net · Sales: sales@olib.ai