AI Subroutines

What it does

AI Subroutines is a tool for recording and replaying browser tasks without LLM token consumption. Users record a task once in their browser, and the tool converts it into a reusable script. When replayed, the script runs deterministically without calling an LLM, incurring token costs, or adding inference delays.

The script itself consists of discovered network calls and page interactions like clicks, typing, and element finding. It executes directly inside the webpage rather than through a proxy, headless worker, or separate process. This in-browser execution means requests dispatch from the tab's own context, preserving authentication, CSRF tokens, TLS sessions, and other browser state.

Who it is for

This is built for developers and teams automating repetitive browser workflows. The tool suits scenarios where tasks are deterministic and don't require dynamic reasoning. Anyone managing multiple identical web interactions—form filling, data extraction, navigation flows—could benefit from eliminating repeated LLM calls.

Pricing

The site does not show prices.

How it stands out

The core differentiation is architectural. By executing scripts inside the browser tab itself rather than externally, the tool preserves the full context of a user's session. This eliminates common friction points: re-authenticating after recording, dealing with CSRF token mismatches, or rebuilding TLS state. It also sidesteps the token consumption model entirely—a significant cost advantage when running the same task dozens or hundreds of times.

The approach trades LLM flexibility for determinism and cost. Once a script is recorded and validated, replaying it becomes lightweight and predictable.

What a founder should check

A competitor should investigate three concrete areas:

First, how sticky is the browser-context execution model? If competitors can match the auth-preservation and cost benefits through other means (better headless automation, browser context management), the architectural advantage narrows.

Second, what are the actual switching costs? If users have only recorded a handful of simple tasks, migration friction is low. If they have hundreds of complex, interdependent subroutines deeply integrated into their workflow, retention is higher. Check whether the tool incentivizes moving more work into the platform.

Third, how much pricing leverage does zero token consumption create? The value proposition depends on end users actually facing token bills today. If LLM providers lower inference costs or introduce cheaper models, the primary benefit erodes. Monitor whether competitors can compete on both cost and flexibility by improving their headless automation capabilities.