--- title: "Browse AI Integration" description: "Run no-code web scraping with Browse AI. Agents discover trained robots, run them against target URLs one at a time or in bulk, poll tasks for extracted lists and texts, and set up monitors that watch pages for changes." url: https://flowrunner.ai/integrations/browse-ai date_modified: 2026-08-01T02:40:32-07:00 --- # Browse AI [Developer Tools](https://flowrunner.ai/integrations/category/developer-infrastructure) Run no-code web scraping with Browse AI. Agents discover trained robots, run them against target URLs one at a time or in bulk, poll tasks for extracted lists and texts, and set up monitors that watch pages for changes. [Verified](https://flowrunner.ai/integrations/verified "What does verified mean?") · 10 actions · API key · available [Browse AI website](https://browse.ai/) · [Platform Documentation](https://www.browse.ai/docs/api/v2) · Capability data verified 2026-08-12 1. The Monday competitive sweep starts on schedule 2. Get Robot confirms the pricing robot's expected input parameters before anything runs 3. Run Bulk Tasks launches the robot across every tracked competitor product URL in one call 4. Get Bulk Run polls until the tasks finish, then Get Task collects the captured price rows per page 5. The agent diffs this week's captured prices against last week's table and flags moves beyond the threshold 6. Significant moves post to the pricing channel with old and new values side by side 7. The pricing owner decides whether to reprice, hold, or dig into why the competitor moved ## What This Integration Enables The web is the largest database in your market, and none of it has an API. Competitor prices, distributor listings, job boards, availability pages: the information is public, current, and locked inside HTML. Browse AI's answer is the trained robot, a scraper you teach once by example and then run forever. This connector puts those robots under FlowRunner's control, which changes what they are: not a scraping tool someone runs when they remember, but a standing data supply line with an agent on the receiving end deciding what the data means. - Discover robots and inspect the input parameters and capture schemas they expect - Run a robot against one URL or hundreds in a single bulk run, then poll tasks for captured lists and texts - Create scheduled monitors that re-run robots and surface changes in captured data - Review past runs to audit what was captured and when The scraping runs at machine speed. What FlowRunner adds is the other half of the loop: diffing, routing, and the [human-in-the-loop](https://flowrunner.ai/concepts/human-in-the-loop) step where market data becomes a market decision. ### Without FlowRunner **Competitor checks by hand**: Someone eyeballs rival pages when there is time, and the spreadsheet is stale by Wednesday **Data trapped in pages**: The market's prices, listings, and availability are visible to everyone and structured for no one **Changes noticed late**: A rival's price cut runs for weeks before anyone happens to look ### With FlowRunner **Sweeps on a schedule**: Robots run in bulk and prices land as structured rows, every week without fail **Pages become tables**: Captured lists and texts flow straight into sheets, bases, and downstream flows **Changes trigger judgment**: Monitors detect the move, the agent quantifies it, and a person decides the response ## Use Case Scenarios ### The pricing desk's standing brief A weekly flow runs the price-capture robot across the competitor URL list with Run Bulk Tasks, collects the finished tasks through Get Bulk Run, and appends every captured row to [Google Sheets](https://flowrunner.ai/integrations/google-sheets). The agent computes deltas against the prior week and posts a digest to [Slack](https://flowrunner.ai/integrations/slack): who moved, by how much, on which SKUs. Repricing decisions stay with the pricing owner; what changed is that they are made on Monday's data instead of last month's. ### Directory listings into a working pipeline A robot trained on an industry directory extracts firm names, locations, and contact pages. The agent runs it across the directory's category URLs, dedupes the captured rows against existing records, and files new prospects into [Airtable](https://flowrunner.ai/integrations/airtable) with the source URL preserved. The sales team wakes up to a pipeline of firms that published their own details publicly, structured and deduplicated, without anyone copying and pasting a single row. ### The back-in-stock tripwire Create Monitor puts a robot on a supplier's product page at a sensible frequency. When a run's captured availability text flips from out of stock to available, the flow reacts within the monitor's cycle: a purchase request drafts itself, the category manager gets pinged, and the order goes out hours after restock instead of days. The same pattern watches a competitor's discontinued line, where the interesting event is the opposite one. ## Human-in-Loop Highlight The spend in this connector hides in two places, and both get a gate. Run Bulk Tasks converts a URL list into credits consumed the moment it fires: a bulk run across a few thousand pages is a real bill, so the agent presents the run size and the account's remaining credit position before launching anything beyond the routine sweeps. Create Monitor is subtler and more dangerous to the budget, because a monitor is not a run, it is a commitment: a robot re-running on schedule, consuming credits every cycle until someone remembers it exists. So no FlowRunner agent creates or upgrades a monitor's frequency on its own judgment. It proposes: "Watching these 40 product pages every 15 minutes costs roughly 12x the current weekly sweep. The three pages that actually change daily could be monitored alone at a fraction of that. Which do you want?" The person who owns the budget picks, because a standing spend deserves a standing decision. Agent processes routinely Detects exception requiring judgment Clear match Continues automatically Ambiguous Routes to human via preferred channel Human decides Agent resumes with decision ## Agent Capabilities 10 actions ### Robots 2 - **List Robots** Lists all robots on the account with pagination. Each robot is a trained scraper or monitor with its ID, name, configured input parameters, and captured lists. The discovery step for every other operation. - **Get Robot** Retrieves a robot's full definition, including the input parameters it expects (such as originUrl) and the schema of the lists and texts it captures. Run before supplying parameters to Run Robot, Create Monitor, or Run Bulk Tasks. ### Tasks 3 - **Run Robot** Runs a robot by creating a new task against a target page URL with any additional input parameters. The task runs asynchronously and returns in a queued or in-progress state. - **List Tasks** Lists a robot's tasks with pagination, optionally filtered by status and sorted by creation time. Each finished task carries its captured data. Used to audit past runs or locate a task ID. - **Get Task** Retrieves a single task by ID with its current status and, once successful, the extracted capturedLists (repeating rows such as products) and capturedTexts (single values). The polling step after every run. ### Monitors 3 - **List Monitors** Lists the monitors configured on a robot, each with its schedule, input parameters, and enabled state. - **Create Monitor** Creates a scheduled monitor so a robot re-runs automatically at a chosen frequency and can alert when captured data changes. A recurring spend commitment, which is why it sits behind this page's human gate. - **Get Monitor** Retrieves a single monitor by ID, including its schedule, input parameters, enabled state, and last-run information. ### Bulk Runs 2 - **Run Bulk Tasks** Runs a robot against many input sets in a single bulk run, creating one task per input. The workhorse for sweeps across URL lists. - **Get Bulk Run** Retrieves a bulk run by ID with its overall status, task counts, and the created tasks. Used to track progress and collect task IDs for fetching captured data. ## Frequently Asked Questions ### What can FlowRunner do with Browse AI? FlowRunner agents can run List Robots, Get Robot, and Run Robot in Browse AI, plus 7 more actions. ### Does connecting Browse AI to FlowRunner require OAuth? No. Browse AI connects to FlowRunner with an API key, no OAuth flow required. ### Can Browse AI trigger a FlowRunner workflow automatically? Browse AI doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger. **Work at Browse AI?** This integration exposes Browse AI to AI agents on every FlowRunner plan, including through MCP, at no cost to you. [See what FlowRunner offers integration partners](https://flowrunner.ai/integrations/partners), including how to keep this page current. --- Markdown version of https://flowrunner.ai/integrations/browse-ai. Site index: https://flowrunner.ai/llms.txt