Switchfrog
Switchfrog is the human-or-agent layer for logged-in software. It helps software companies identify whether logged-in sessions are human- or AI-operated, govern protected actions, understand agent activity, and meter or monetize agent work.
At a Glance
- B2B SaaS companies
- Software companies with logged-in customer applications
- Products with paid seats and concerns about AI-agent usage
- Companies that need authorization and compliance controls for browser-based automation
- +1 more
AI Tools by Switchfrog
(1)Switchfrog
AI Browser Agent Detection
Discussions
No discussions yet
Be the first to start a discussion about Switchfrog
Latest News
Switchfrog launched/presented WebMCP workflow tooling that turns product workflows into agent-ready actions.
Switchfrog's documentation published a quickstart for verifying protected actions using browser session tokens and human/agent/unknown verdicts.
LinkedIn company data identified Switchfrog as a San Francisco AI company founded in 2026 and backed by a16z Speedrun.
Switchfrog was publicly listed as a SaaS side project, describing its one-line integration and agent-detection workflow.
Products & Services
A browser SDK and service that detects whether a signed-in browser session is human-driven, AI-agent-driven, or unknown; it exposes verdicts via API/webhook so a product can allow, review, block, or otherwise respond.
A workflow-to-tool product that discovers repeated workflows from product usage and turns selected workflows into WebMCP actions that agents can execute within a customer's existing account and permissions.
Integration tooling including a browser SDK, session tokens, identity attachment, and a POST /v1/verify backend endpoint for applying action-specific authorization policies.
A no-install demonstration that classifies live browser activity and shows a session profiler for human versus automated interaction.
Market Position
Switchfrog positions itself between traditional bot detection/security controls and agent-enablement or workflow automation tooling. Its differentiator is detecting agent takeover inside a real, logged-in customer browser session and combining that signal with action-level authorization, usage analytics, agent metering/billing, and automatically generated WebMCP tools rather than relying only on traffic-level bot signatures or requiring a separately hosted MCP server.
Leadership
Founders
Thomas (Tommy) McCormick
Founder and CEO. According to a16z Speedrun, he began working on security and cryptography at Microsoft at age 16, later built social-commerce software embedded by brands including Nike and Target, and led data infrastructure at Faire when ChatGPT launched.
Executive Team
Thomas (Tommy) McCormick
Founder & CEO
Previously worked on security and cryptography at Microsoft, built social-commerce software used by brands such as Nike and Target, and led data infrastructure at Faire.
Founding Story
Tommy McCormick says Switchfrog grew from his experience building technology that helped internet businesses capture new demand channels: while leading data infrastructure at Faire when ChatGPT launched, he saw that AI agents would become a new way people interact with software. Switchfrog was created to help businesses see, govern, and monetize that agent activity while making their products easier for agents to use.
Business Model
Revenue Model
B2B software/SaaS model. Switchfrog provides SDK/API and workflow tooling to software companies, with agent activity intended to be metered and billed like API usage or an agent seat. The company says it is working out agent-seat and WebMCP usage pricing with initial beta customers and currently offers early access.
Target Markets
- B2B SaaS companies
- Software companies with logged-in customer applications
- Products with paid seats and concerns about AI-agent usage
- Companies that need authorization and compliance controls for browser-based automation
- Teams seeking agent-ready workflows and WebMCP support
- Protect transfers, invoices, exports, and other sensitive actions from unauthorized automation
- Distinguish AI-agent activity from human activity inside SaaS products
- Measure and meter agent work performed through paid customer seats
- Bill or price agent usage differently from human seat usage
- Create direct agent interfaces for multi-step SaaS workflows
- Understand where browser agents get stuck and which workflows they use