PromptLoop
PromptLoop is a no-code AI research platform from Kiter Inc. Its agents crawl websites and search the open web to turn company lists into structured GTM data, syncing straight into HubSpot, Salesforce, Excel and Google Sheets.
What is PromptLoop?
PromptLoop is a web-research platform built on no-code AI agents, aimed squarely at go-to-market and B2B data work. It is published by Kiter Inc., a two-person San Francisco company from the Y Combinator Winter 2022 batch, and its homepage promise is blunt: build high-quality GTM datasets in minutes, not weeks.
The product rests on three blocks. Tasks are the agents: you describe what you want in plain language and PromptLoop writes the task in ten to twenty seconds, or you copy one from the template library. A Website task crawls a URL you supply; a Search task looks the company up first, then crawls. Crawl depth is your choice, from Single Page to Smart Crawl by default and Deep Research Smart Crawl for enterprise customers, and the answer can be forced into one of ten formats, from plain text and numbers to booleans, single or multiple categories, lists, image links and scripts. An optional Source column returns the exact page an answer came from, confidence scores are documented, and task versions are kept. Chains let one agent feed the next, so a site crawl can hand off to a LinkedIn enrichment step.
Datasets are where the volume lives. Upload a CSV or XLSX, or, on Enterprise plans, generate the list on the platform through a Geographic Crawl by region and keyword or a Global Crawl by business filters. Datasets run to hundreds of thousands of rows, and jobs of 20,000 to 50,000 rows usually finish inside ninety minutes. Everything is shared across the organization by default.
Integrations close the loop: two-way sync with HubSpot and Salesforce, plus Excel and Google Sheets add-ins exposing Label, Browsing and Inference functions cell by cell, the add-on being listed on the Google Workspace Marketplace.
Under the hood, Kiter runs its own models trained for web navigation alongside access to dozens of providers, refreshed weekly. Marketing claims are loud: 85%+ cost saved, 10x faster than prompting ChatGPT or Claude by hand, 100x faster than Deep Research, 10,000+ sales and marketing teams. Customers named on the site include Arketa, Flume AI and CenterOak. What it is not is a pre-built contact database: every answer is researched on demand, and the features structured teams want most, meaning CRM sync, the Web API and contact enrichment, start at the Team plan.
What it does
- Search the open web for companies and websites matching an ICP
- Crawl any website and extract structured columns of data from it
- Enrich CSV or XLSX files running to thousands of rows
- Sync enriched results back into HubSpot or Salesforce
- Find contact email addresses and phone numbers
- Categorize and label text data at scale
- Run AI functions cell by cell inside Excel and Google Sheets
When to use PromptLoop / When not to
A quick filter to help you decide if PromptLoop is the right fit.
When to use PromptLoop
- Sales and GTM teams building prospect lists and qualifying accounts against an ICP
- Market researchers and strategists mapping an entire sector from thousands of company websites
- RevOps and CRM owners who need bulk enrichment written straight back into HubSpot or Salesforce
- Spreadsheet-first analysts who want AI functions inside Excel or Google Sheets
- Founders and small teams who want web research at scale without hiring a scraping engineer or a data agency
When not to use PromptLoop
- Buyers who want an off-the-shelf contact database, since PromptLoop researches on demand rather than serving pre-built records
- Teams that must keep data inside the EU, as processing takes place exclusively in the United States
- Procurement-led organizations that need a signed DPA or a SOC 2 or ISO 27001 certificate before they can sign
- Free and Individual plan users who need CRM integrations, the Web API or email and phone enrichment
- Anyone sourcing phone numbers in volume on a tight budget, where 50 credits per number adds up fast
How to use PromptLoop
A typical end-to-end flow, from setup to results.
- Create an account and claim the 500 free monthly credits, no credit card required
- Build a dataset by uploading a .csv or .xlsx file, or generate one directly on the platform
- Let PromptLoop detect the columns, then tick the one that feeds the agent, usually a website URL or a search term
- Create a Task by describing your goal in plain language, or copy a ready-made template from the library
- Test the task on a sample row and iterate on the instructions and the output format, running the three to five representative examples the documentation recommends
- Hit Launch Job, map your dataset columns to the task inputs and start the run
- Watch progress in real time while the job runs in the background
- Filter, search and review the results, then export to CSV or XLSX or chain another task onto them, each run writing a new dataset version
- For CRM work, connect HubSpot or Salesforce over OAuth, import a list, preview the matching and sync results back to the properties you choose, or set an auto-run that fires on every new company added
- For spreadsheet work, install the add-in through Get Add-Ons in Excel or Google Sheets, then enable it in each sheet
Pros & Cons
Pros
- Genuinely usable free plan: 500 credits every month with no credit card, plus a 14-day free trial on Individual
- Low entry price for the category at $18 per month, with volume pricing announced down to $0.015 per credit
- No technical skill required, since a working task is written from a single sentence in plain language
- Building and editing tasks costs nothing, and sample rows can be tested as often as you like before a run
- Traceable output through a Source column, confidence scores and retained task and dataset versions
- Credits are not charged when the model fails technically, nor for an email or phone number it does not find
- Fits existing workflows and moves fast: two-way CRM sync with a preview before writing, Excel and Google Sheets add-ins, and 20,000 to 50,000 row jobs finishing in under ninety minutes
Cons
- Paid pricing is rendered in JavaScript, so the amounts are invisible on the pricing page without running scripts
- Credits are consumed even when the model returns the wrong answer, provided the run itself did not fail
- Phone enrichment is expensive at 50 credits per number, and list outputs bill one credit per row produced
- CRM integrations, the Web API and contact enrichment are locked behind the Team plan, and dataset generation behind Enterprise
- No DPA published, no security certification displayed, and no explicit GDPR compliance claim, Article 27 representative or DPO
- Data is transferred to and processed in the United States only, with no EU hosting option and no documented way to opt out of model training
- No postal address published, support limited to email and booked calls, no mobile app, no list of interface languages, and parts of the documentation flagged as Legacy
Pricing & Plans
PromptLoop operates on a freemium, credit-based model. A permanent free plan provides 500 credits per month without a credit card, and a 14-day free trial is offered on the first paid tier. The lowest paid entry point is the Individual plan at USD 18.00 per month, which includes 1,000 monthly credits, with additional credits billed at USD 0.035 each. One credit covers a complete AI research run on a single input, meaning one row or website for up to twelve returned items or columns, while list outputs consume one credit per row produced, an email costs 5 credits and a phone number 50 credits, the latter two being charged only when a value is returned. The site estimates 2 to 3 credits per website or contact enriched. Creating and editing tasks is free, and no credit is charged when a run fails on the model side. Volume pricing is announced down to USD 0.015 per credit, with the Team plan starting at USD 400.00 per month; custom pricing applies above 35,000 credits and the sales team must be contacted beyond 50,000. Subscriptions may be cancelled at any time, upgraded or downgraded as research volume changes, although subscription fees are stated to be non-refundable except where the law requires otherwise.
- 500 credits
- 1 seat
- 25 rows per run
- unlimited AI agents
- website data extraction
- custom and advanced data formats
- 25 AI company search results and HubSpot AI enrichment. Excludes CRM integrations
- the Web API
- 1
- 000 credits (roughly 200 to 300 companies)
- 1 seat
- 1
- 000 rows per batch
- 500 AI company search results
- HubSpot AI enrichment
- a 14-day free trial
- up to 50
- 000 credits
- unlimited seats
- 50
- 000+ rows per batch
- everything in Individual plus CRM integrations
- the PromptLoop Web API
- email and phone enrichment
- above 35
- 000 credits the price becomes custom and the sales team must be contacted
- and beyond 50
- 000 credits there is no self-serve option at all
- bespoke plans adding dataset generation through Geographic Crawl and Global Crawl
- Deep Research Smart Crawl
- detailed onboarding and infrastructure that can be customized for stricter security requirements
Data, GDPR & hosting
A consolidated view of how PromptLoop handles your data.
GDPR overview
Partial, and never claimed outright. The privacy policy, effective 20 August 2023, carries a section for EU and EEA residents that opens the standard GDPR rights: access, update, erasure, rectification, objection, restriction, portability and withdrawal of consent, with a link to the official text and an explicit reminder of the right to complain to an EEA supervisory authority. Requests go to mail@kiter.app, and identity may be verified first. Beyond that, Kiter Inc. nowhere states that it is GDPR compliant: no certification is displayed (no SOC 2, no ISO 27001), no Article 27 EU representative is designated, no DPO is named, and no data processing agreement is published or offered. Data from users outside the United States is transferred there and processed there, on the basis of consent. CalOPPA and CCPA sections sit alongside, under California law.
Who owns the data?
Under a privacy policy effective 20 August 2023, Kiter, Inc. declares itself the Data Controller of the personal data it collects. The Terms of Service, last updated 10 January 2022, leave your material with you: you retain any and all rights to content you submit, post or display through the Service, while granting Kiter a license to use, modify, reproduce and distribute it on and through the Service. The policy adds that you are the only owner of your personal data, may request disclosure or deletion at any time, and that Kiter neither sells nor rents it. Results stay in your account indefinitely unless you delete the dataset.
Reuse rights
Nothing in the Terms of Service, last updated 10 January 2022, restricts what you do with the research output you generate: you keep every right to the content you submit and to the datasets you build, so results can be exported to CSV or XLSX, pushed into HubSpot or Salesforce and reused commercially without asking Kiter for permission. On its own side, Kiter limits collection to your email address plus cookies and usage data (IP address, browser, pages viewed, time spent, device identifiers), and uses it to run and maintain the service, notify you of changes, provide support, analyze and improve the product, monitor usage, detect technical incidents, bill subscriptions and send account notices, plus marketing about similar products unless you object. Disclosure to third parties is limited to legal requests and to a merger, acquisition or asset sale. Payments run through PCI-DSS compliant processors and no card details are stored by the publisher. Do Not Track signals are honored, with no tracking, cookies or advertising while the signal is active. The policy says nothing either way about customer data being used to train models.
Data retention & training
Hosting summary
All processing takes place in the United States. The privacy policy is explicit for users located elsewhere: data, including personal data, is transferred to the United States and processed there, with the user's consent as the stated basis. No alternative hosting country or region is named, and no EU data-residency option is offered. Third parties handling data are named in the Service Providers section: Google Analytics, PostHog, GitHub, PayPal or Braintree, and Stripe. As an infrastructure indication rather than a declaration by the publisher, the domain resolves to 76.76.21.21, an anycast node on AS16509 (Amazon.com Inc.) located in the United States. On security, the policy carries the usual disclaimer that no method of transmission over the internet or method of electronic storage is 100% secure, and commits only to commercially reasonable means. No security certification is displayed. The documentation does state that for companies requiring an even higher level of data security than the standard offering, the infrastructure can be customized to the team's requirements, but that is a sales conversation rather than a published guarantee.
Things to keep in mind
Risks and trade-offs to weigh before adopting PromptLoop.
- Credits are debited even when the answer is wrong or incomplete, so a large job launched without testing a subset first can burn a month's allowance on unusable data
- A broken link in your input file still costs a credit, and list-type outputs multiply consumption, since one website returning ten items costs ten credits
- Phone enrichment at 50 credits per number turns a routine bulk job into a serious bill, so check the budget before running one
- CRM sync overwrites the properties you select even when the new value is empty, and two tasks writing the same HubSpot property at once leave whichever finished last
- Never delete the hs_object_id column from a dataset imported from HubSpot, and remember that editing a task does not update an existing auto-run, which has to be recreated
- There is no hosting option outside the United States, a hard stop in some European regulatory contexts, no documented way to keep your data out of model training, and no public documentation for the Web API, whose scope must be confirmed with the team before you commit
- Research this fast makes it tempting to stop reading the sources: the Source column and confidence scores exist for a reason, and handing judgement entirely to an agent is how bad data reaches a CRM at scale
Setup & Integrations
Technical difficulty
Low. PromptLoop is built for users with no prompting or AI-systems knowledge: a task is generated from a plain-language description in ten to twenty seconds, and a dataset is a drag-and-dropped spreadsheet. The site claims onboarding in under fifteen minutes, without an agency. CRM connections use OAuth, need no code and no CSV export, and one team member is enough to set them up; HubSpot auto-enrichment is a five-step, five-minute configuration and importing a list takes one to five minutes. The Excel and Google Sheets add-ins install through Get Add-Ons and must be enabled in each sheet.
Deployment
Integrations
Behind PromptLoop
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement PromptLoop.
Frequently asked questions
What does PromptLoop actually do?
Is there a free plan, and is there a free trial?
How much does the first paid plan cost?
How do credits work?
Which integrations are available?
Which AI models does PromptLoop use?
Is PromptLoop an alternative to Clay?
How big can a job be, and how long does it take?
Where is the data hosted, and is there a minimum age?
How do I get support or exercise my GDPR rights?
Should you pick PromptLoop?
PromptLoop is one of the more convincing entries in a crowded category, and the product side is where it earns that. The documentation is unusually specific, covering crawl depths, output formats, credit arithmetic, error behavior and HubSpot field mapping, and the publisher is out in the open: Kiter Inc., San Francisco, a Y Combinator Winter 2022 company. The free plan is not a token gesture either: 500 credits a month with no credit card is enough to judge output quality on real rows before spending anything, and USD 18.00 per month with a 14-day trial is a low bar for the category.
The reservations are real, and two of them concern transparency rather than capability. The legal and security posture is thin: no DPA, no SOC 2 or ISO 27001, no explicit GDPR compliance claim, no Article 27 representative, no postal address, and data processed exclusively in the United States. For a European buyer with a procurement checklist, that is a short conversation. And the pricing page itself renders in JavaScript, so the paid tiers are invisible in the raw page, an odd choice for a product otherwise so willing to show its workings.
The shape of the offer matters too: CRM sync, the Web API and contact enrichment only unlock on Team from USD 400.00 per month, and dataset generation is Enterprise-only. Individuals get a fast, cheap research engine; teams get the full loop at a very different price. Worth trying for sales, growth and research teams that want web research at scale without hiring a scraping engineer, provided a US-only, self-declared compliance posture is acceptable.
- Choosing a selection results in a full page refresh.
- Opens in a new window.