Turn website data into structured tables with an AI-assisted browser scraper. Choose fields, extract information and export it to spreadsheet and database destinations.
Used this tool? Rate it
Used this tool? Rate it
Thunderbit is an AI web scraper positioned for sales and operations teams that need website information in structured tables. Its browser extension uses AI to suggest fields, while the company describes a web app and other entry points for managing extraction work. The Chrome and Edge listings identify browser extensions; the claims about what they extract and how well they work are the publisher's descriptions, not an independent test.
Page content and extraction instructions are used to produce structured data for export to spreadsheet and database tools. Web data extraction can reduce repeated copying across lists and detail pages. AI interpretation can vary, however, and the company says user review is important.
The extension listing says AI suggests columns and structures the output. Instead of specifying a selector for every field, a user can describe the desired information in ordinary language. The listing also describes extraction from websites, PDFs and images; this is a publisher claim and does not establish that every file or site will work.
The terms describe AI Suggest Fields for creating scraper templates and AI Improve Fields for refining them. They also describe per-column prompts and transformations such as summarization, categorization, translation, formatting and calculations. These are separate steps from selecting the page: the desired column and the transformation applied to its values determine what the resulting table is meant to contain.
The listing says subpage scraping visits linked pages and enriches a table with additional details. A list may contain a product name while its linked detail page contains further information; the described feature is intended to bring those details into the same table. The company's comparison says this feature is not included on Free, and describes two credits per subpage-enriched row. Availability and credit use are therefore distinct from the presence of a link on the page.
The extension listing also describes pagination and infinite scroll, while the company's comparison limits Free pagination to three pages and excludes infinite scroll on that tier. Those statements describe a capability and a plan boundary, not a promise that any given page layout will work. The company's comparison says that neither tool it discusses guarantees success on every JavaScript-heavy site because anti-bot protections, render timing and site-specific behavior vary.
Scheduled scraping is described for monitoring prices and inventory over time. Terms say scheduled operations consume credits when an execution produces metered activity. A schedule therefore changes the number of runs, not the need for an allowance to cover the work performed on each run.
A company comparison dated August 13, 2026 lists one scheduled scraper at daily frequency and five saved scrapers on Free.
The terms list Excel, Google Sheets, Airtable, Notion, CSV and JSON among export options. They say export does not consume credits, although compatibility and availability of third-party services affect export functionality. Formats such as CSV and destinations such as an online database serve different handoff paths; a data export and the extraction that created its rows are separately described for credit purposes.
The company describes the following Chrome extension workflow on a compatible, authorized page:
Scheduled scraping can support price monitoring and inventory monitoring, according to the extension publisher. These use cases depend on collecting a series of observations rather than only producing a one-time list. Its listing also describes collecting details from linked product pages and company profiles, where the additional fields are on a page reached from the original list. The suitability of either workflow depends on the available plan and on what the target pages actually expose.
The company's examples also cover real estate listings and e-commerce product research. In a listing context, fields such as the displayed price, status and property details may be relevant; a product catalog workflow instead starts with fields such as product names and prices. The company's own descriptions frame these as examples of data that can be requested, not as evidence of accuracy on a particular site.
Thunderbit positions the scraper for sales and operations teams. Its publisher presents lead lists, price tracking and database building as use cases. These tasks have different outputs, but share a need to organize visible information into columns.
The distinction between the browser and cloud modes matters for anyone working with pages behind a login. The terms describe browser-based execution using the user's own browser session, including authenticated sites. They separately say cloud execution does not use that session and is available only for public content unless the user provides credentials. The terms also require valid authorized access to authenticated content; a logged-in session is not itself a grant of rights to use the data.
The published terms require users to be at least 18 years old or the age of majority in their jurisdiction, whichever is greater. Certain features require an account, and developer access through the API or SDK requires registration and additional terms. The company also says an organization administrator is responsible for managing user access and permissions when an organization manages multiple users under one account. These account conditions concern access to the service; they do not override conditions imposed by the websites being accessed.
The terms allow only one account unless the company permits more in writing, and prohibit sharing account credentials or allowing a third party to access the account. They separately describe organization accounts with multiple authorized users without explaining how that access restriction applies to them.
Thunderbit has a Chrome extension distributed in the Chrome Web Store and an Edge extension distributed in Edge Add-ons. The Edge listing states that the extension can open in the browser sidebar. Browser-based execution is described in the terms as operating through the Chrome extension using the user's session; the terms describe cloud execution separately. These statements identify entry points and modes rather than proving identical behavior in each browser.
The web app is described in the terms as a cloud-hosted dashboard for managing accounts, workflows and sessions. The site also advertises a web scraper API for applications and internal workflows. API access is a different entry point from opening a page in an extension, and the terms describe additional account requirements for developers accessing the service through the API or SDK.
For the Chrome extension, the terms list access to web page content, active tab interaction, clipboard and storage among the permissions needed to provide the service.
Free includes six pages per month, with a maximum of 30 credits per page. Paid plans use credit allowances that vary by billing period.
| Plan | Captured monthly billing | Allowance |
|---|---|---|
| Free | $0/month | 6 pages/month; maximum 30 credits per page |
| Starter | $15/month | 500 credits/month |
| Pro | $38/month | 3,000 credits/month |
Annual Starter is listed at $9 per month billed yearly, with 5,000 credits provided upfront for the year. An earlier company comparison lists a Pro 1 tier at approximately $24 per month when billed annually. The pricing page displays a time-limited offer for Pro.
The Free allowance uses both a page limit and a per-page credit ceiling. The company's comparison approximates one standard output row as one credit, and says a subpage-enriched row costs two credits. The terms describe scraping as metered per output row and distinguish it from non-metered export. These units are not interchangeable: the number of pages visited does not by itself state how many output rows a task will produce.
The terms say unused credits normally expire at the end of each billing cycle unless the plan states otherwise. Metered operations pause when the balance reaches zero, until credits become available through renewal or upgrade. Paid plans renew automatically unless cancelled before renewal; cancellation through account settings takes effect at the end of the current billing period. Fees are generally non-refundable, with exceptions including a pro-rated refund if the company terminates an account without cause.
The terms distinguish a plan's recurring fee from the consumption of its included credits. Different metered activities can have different rates: scraping is described per output row, agent activity per action or session as specified on the site, and workflows per execution. The terms say that certain agent sessions and complex workflows can use a variable number of credits depending on their actions. A schedule with repeated executions or a task with subpage enrichment can thus have a different consumption pattern from a single list extraction, even if the same plan is used.
For a paid subscription, the terms require at least 30 days' notice of a fee increase, with the new fee applying at the next billing cycle after the notice period. They separately require at least 30 days' notice before increasing the credit consumption rate for an existing activity type.
The company says Free excludes infinite scroll and subpage scraping, and its comparison describes pagination on Free as limited to three pages. Its broader extension listing describes all three functions, so the plan boundary matters when reading that feature list. The company also says AI interpretation can vary and that user review is important; a suggested column name is not a guarantee that each extracted value matches the intended field.
The terms restrict browser scraping of authenticated content to cases where the user has valid authorized access. They also make the user responsible for respecting third-party technical restrictions and for determining the right to use extracted data. Those are the vendor's conditions of service, not a legal conclusion that any particular extraction or reuse is permitted.
The terms also describe an AI Agent that can interact with page elements, fill forms and navigate between pages under user authorization. The terms warn that the agent may make errors. Some actions, such as a submitted order or sent email, may not be reversible. Purchases, orders and financial transactions require the user’s express, informed authorization for each transaction or category of transactions under those terms.
For personal data collected through scraping, the terms require a lawful basis and assign the user the role and responsibilities of a data controller.
For Chrome extension extraction, the privacy policy says the selected page’s HTML is sent to Thunderbit’s servers for AI processing and is not retained after processing. Extracted data is stored in the account; page URLs may be logged for functionality, troubleshooting and analytics.
Extracted data is retained until deletion or account termination under the privacy policy. The plan matrix lists data retention of 14 days for Free and 60 days for both Starter and Pro; it does not specify in that row which data the limits cover. A company comparison also lists 14 days for Free. The terms separately say User Data is retained for 30 days after account termination to allow export, after which the company may delete it from active systems, with legally required retention possible. These statements use different scopes and do not explain how the periods fit together.
The privacy policy identifies Google Gemini as the current AI service provider and says Thunderbit does not use extracted data, page content or API calls to develop or train generative AI or machine learning models. It also says its agreements require AI providers not to use the data for their own purposes, including training. These are policy statements by Thunderbit; they are not independent verification of the providers' practices.
The terms describe a wider set of possible providers: Google, OpenAI, Anthropic, DeepSeek and others, which the company may change. The policy's current-provider statement and the terms' list of possible providers have different scopes; the latter does not establish that every listed model processes every extraction.
The privacy policy says Google sign-in provides the company with a name, email address and profile picture. It also says integrations can provide authentication tokens needed to export data to connected services. When a user chooses an export, the policy says the extracted data is transferred on that user's instruction, and that the destination's own privacy policy and terms apply after transfer. The existence of an export destination does not mean that the destination follows the same retention rules as Thunderbit.
The policy describes form auto-fill data separately from extracted website data: it says form field data is stored locally in the extension and is not transmitted to Thunderbit's servers. It also says site and service use can produce automatically collected information such as IP address, approximate location, browser and device details, pages viewed and actions taken. Its stated retention for usage and analytics data in aggregated or anonymized form is up to 24 months; that is a different category from extracted data and page HTML.
Thunderbit's policy says users may request access, correction, deletion and data portability, with legal exceptions applying to deletion, and may opt out of promotional communications. The policy says administrative messages may still be sent after a marketing opt-out. It states a 30-day response period for privacy requests. The policy also states that no method of internet transmission or electronic storage is completely secure.
Agent session data is a separate category: the terms include screenshots, action logs and interaction records, and say the company may retain session logs for a reasonable period for service, troubleshooting and quality improvement.
The terms grant a limited license to use the service for lawful purposes, including commercial use, subject to the subscription plan. They also say extracted data is user data and that Thunderbit does not claim ownership of it. The terms also give the company a license to host, store, process, transmit, display and distribute user data for providing and improving the service, with the license ending on deletion of the data or account subject to stated archival exceptions. The same terms place responsibility for determining ownership and rights to use third-party data on the user. The service license should therefore be read separately from any rights in material collected from another site.
The terms do not obligate the company to back up User Data and assign responsibility for maintaining backup copies to the user. They also say AI Output may not qualify for intellectual property protection in every jurisdiction and that similar or identical output may be generated for other users.
