ai_answer_snapshot
ChatGPTSends up to 3 questions to 2 or 3 AI assistants at once, each with live web search (Perplexity and Gemini by default; Claude and Grok on request), and returns each answer side by side with the sources and domains it cited. With trackBrands it reports, per assistant, whether each brand or website is mentioned, in what order, and whether its site is cited. Use it to see how different AI assistants answer the same question about a market, product or brand.
ai_shopping_results
ChatGPTAsks up to 3 shopping questions (for example "best cordless vacuum for pet hair") to an OpenAI model with live web search, the way an AI shopping assistant answers them, and returns the products, prices, shops and cited sources it recommended in order. With trackBrands it reports whether each brand, product, shop or website appears, at what position, as a shop or as a cited source. Use it to check a brand's visibility in AI shopping answers.
australia_visa_processing_times
ChatGPTOfficial Australian Department of Home Affairs processing times for a visa subclass and stream (how long 25%, 50%, 75% and 90% of applications take), or the list of visas with published times.
best_time_to_post
ChatGPTRanks the weekday and hour slots when an Instagram, TikTok, YouTube, X or Facebook account's posts did best, from post rows the user provides (account, date with time, and likes, comments, shares or views): each post against the account's own median, best days, best hours, the weakest slot and a heatmap, in the time zone asked for. Send 30 to 150 posts per account. Works only on the rows sent. Up to 20 accounts.
calculate_uk_tourist_tax
ChatGPTUK only. Works out the tourist tax (visitor levy) a holiday let, B&B, guest house, hotel, hostel or campsite owes on one booking: Edinburgh 5% (first 5 nights, live since 24 July 2026), Glasgow 5% (from 25 January 2027), Aberdeen 7%, Stirling 3%, West Dunbartonshire 5%, Cardiff £1.30 per person per night (from 1 April 2027). Returns the levy, how it was worked out, and the return deadline. The host owes it even when booked through Airbnb or Booking.com.
check_ai_crawler_access
ChatGPTChecks whether AI crawlers can read up to 3 websites: OpenAI's training, search and user-request crawlers, ClaudeBot (Anthropic), PerplexityBot and Google-Extended by default, Meta, Apple, Common Crawl and more on request. Reads robots.txt for each crawler, requests the page with each crawler's published user agent and with a normal browser to catch firewalls that refuse AI crawlers, and looks for llms.txt. Returns allowed or blocked per crawler, why, what blocking it means, and what to fix. Use it for AI search visibility questions.
check_email_address
ChatGPTChecks one email address and returns valid, risky or invalid with the reasons: address format, whether the domain can receive mail (live MX lookup), known throwaway domains (mailinator and others, invalid), shared inboxes like info@ or sales@ (risky), likely typos such as gmial.com with the fix in didYouMean, and free providers such as gmail.com (flag only). Does not ping the mailbox (no SMTP), so valid is not a delivery guarantee.
check_subcontractor_right_to_work_duty
ChatGPTUK only. Says whether a business must do a right to work check on a subcontractor, gig worker or platform worker under the rule in force from 1 October 2026 (Border Security, Asylum and Immigration Act 2025, s48), and what to do. General information, not legal advice.
china_buy_check
ChatGPTChecks up to 5 product listings from Taobao, Tmall or 1688 that the user provides (title, price, shop name and link, plus any shop ratings or sales they have) and returns a BUY, CHECK or AVOID verdict with reasons: counterfeit brand warning signs, a 0 to 100 seller trust score, an English summary of the Chinese title, a size chart in US, UK and EU sizes for shoes and clothes, an estimated delivered cost to the US, UK or EU (agent fee, shipping, duty and VAT, every assumption listed) and questions to ask the seller. Uses an AI model for the translation and summary. Does not visit the listing pages.
clean_email_list
ChatGPTChecks up to 200 email addresses: valid format, whether the domain can receive mail (MX records), disposable or role addresses (info@, sales@), likely typos (gmial.com) and duplicates. Use it before sending to a list.
clean_rows
ChatGPTCleans up to 150 messy JSON rows: removes duplicates (exact, normalized or fuzzy), flattens nested JSON, trims and fixes emails, phone numbers and URLs, strips HTML, turns number and true/false text into real types, drops empty fields, keeps, removes or renames columns. Returns clean rows plus a report of what changed.
compare_price_per_unit
ChatGPTReads the pack size in each product title (count, loads, weight, volume, "Pack of 6", "6 x 330 ml") from product rows the user provides, for example an Amazon, Walmart or grocery product list. Puts every product in a search or category on one unit, ranks them cheapest per unit, and flags overpriced products and bigger packs that cost more per unit. Works only on the rows sent; no website is contacted. Up to 1,000 products.
compare_two_lists
ChatGPTCompares an old and a new list of JSON rows by a key field and returns what was added, removed and changed, with the old and new value of every changed field. Use it to spot new leads, price changes or deleted items between two scrapes. Up to 500 rows a side.
convert_images
ChatGPTDownloads up to 10 images and converts them to WebP (default), JPEG, PNG or AVIF, optionally resized to a maximum width or height, EXIF stripped and photos turned the right way up. Returns a link to each converted file with its measured width, height and size before and after.
data_breach_deadlines
ChatGPTWorks out every data breach notice a company owes and the deadline for each. Send how many people were affected per US state or country, the data types exposed, whether it was encrypted, the discovery date and flags (SEC registrant, HIPAA, GLBA, NYDFS). Returns each notice owed (people, attorneys general, credit bureaus, HHS, SEC 8-K, FTC, EU and UK regulators, Canada, Australia) with a dated deadline where the law fixes one and the official source. Not legal advice.
detect_tech_stack
ChatGPTDetects the technology behind up to 20 websites: CMS, ecommerce platform, analytics, tag managers, frameworks, hosting and CDN, payment and chat tools, with confidence per finding. Filter for sites using a given technology to qualify leads.
domain_age_and_whois
ChatGPTLooks up up to 15 domains in the official RDAP registries (the modern WHOIS): when each was registered and how old it is, when it expires, the registrar, status codes and name servers. Flags new domains and ones about to expire. Useful for fraud checks, lead qualification and spotting expiring domains.
download_files
ChatGPTDownloads up to 20 files (images, PDFs, documents) from public links and returns a stable download link, file type and size for each, so files from a scrape or a list can be saved in one go.
extract_articles
ChatGPTReturns the full text, headline, author, publish date, publisher, image and tags of up to 20 news or blog articles at once, as plain text or Markdown. Honours robots.txt. Use read_web_page for a single page.
federal_contract_winners
ChatGPTLists companies that signed new US federal contracts recently (USAspending.gov), one row per company scored 0 to 100 as a sales lead: new award value, agency, federal revenue growth, first federal win, small business, veteran, woman owned, 8(a) and HUBZone flags, address and SAM.gov link. Filter by NAICS industry code, keyword, agency or state.
federal_grant_matcher
ChatGPTFinds open and forecast US federal grants on Grants.gov that fit what you do, your organisation type and state, scored 0 to 100 with the reasons: eligibility for your organisation type, deadline and days left, award ceiling and floor, total funding, cost sharing, the agency contact, plus past award sizes and past recipients for the same program from USAspending.gov. Use it when someone asks which grants a nonprofit, school, city, tribe, university or small business could apply for.
ffxiv_market_prices
ChatGPTCurrent Final Fantasy XIV market board listings and recent sale history for up to 10 items on a world, data centre or region, with the cheapest price per world. Item IDs are the numeric game item IDs.
filter_and_transform_rows
ChatGPTFilters, transforms, sorts and dedupes up to 500 JSON rows in one call: conditions (equals, contains, greater than, regex, empty), field renames, trims, case changes, number and date parsing, computed fields, sort, distinct by field, offset and limit. Returns the kept rows plus counts of rows dropped and why.
find_company_logos
ChatGPTFinds the logo, the best app icon and the favicon of up to 4 company websites from the site's own home page, structured data and web app manifest, downloads each candidate to prove it is a real image, and returns its link, format, width and height, plus the site name, theme colour and social share image. Honours robots.txt.
find_long_running_ads
ChatGPTScores ads from ad rows the user provides, for example an export from the Meta (Facebook and Instagram) Ad Library or Google Ads Transparency Center: days running, whether still live, how many copies of the creative run, a winner score 0 to 100, a tier (proven winner, scaling, testing, stopped early) and a reason, plus a summary per advertiser. Works only on the rows sent; no website is contacted. Up to 500 ads.
find_viral_video_outliers
ChatGPTFinds TikTok, Instagram Reels and YouTube videos that beat their creator's normal views, from video rows the user provides (creator, views, and optionally likes, comments, date, link and caption): views against the same creator's median, a level (viral 10x, strong 5x, outlier, rising), the hook and a plain reason, plus hashtags that overperform. Works only on the rows sent; no website is contacted. Up to 500 videos.
find_website_contacts
ChatGPTVisits up to 10 business websites (home and contact pages) and returns the published contact emails (role inboxes such as info@ by default), phone numbers and social profiles for each. Honours robots.txt; nothing is guessed.
google_review_velocity
ChatGPTScores each business from Google Maps review rows the user provides (business name, review date, stars, and the owner's reply if any): reviews in the last 90 days against the 90 before, momentum (growing, slowing, stalled), owner reply rate, unanswered 1 and 2 star reviews, rating trend and a 0 to 100 lead score with a one-line reason. Works only on the rows sent; no website is contacted. Up to 100 businesses.
group_and_pivot_rows
ChatGPTGroups up to 500 JSON rows and works out totals per group: count, sum, average, min, max, median, distinct count, first and last, with date buckets (day, week, month, quarter, year), a pivot table, sorting, top N and a totals row. Returns one row per group plus a summary.
how_to_do_a_right_to_work_check
ChatGPTUK only. Returns the three accepted ways to check someone's right to work, what records to keep and for how long, and the fines.
ireland_company_lookup
ChatGPTLooks up Irish companies in the official CRO register by company number or by name: status, type, registered address, incorporation date and recent filings. Up to 20 results a call.
job_skills_demand
ChatGPTRanks the skills, tools, licences and certificates named in job listing rows the user provides (title, description and pay): share of jobs asking for each per job title or role family, rising or falling over time, median pay with and without each skill, and a learn this next line. Company boilerplate is skipped. Works only on the rows sent. Up to 1,000 jobs.
join_rows
ChatGPTJoins two lists of JSON rows on one or more key fields, like a VLOOKUP or SQL join: inner, left, right, full or anti. Keys match exactly, or normalized (case, spacing and number-as-text differences ignored). Returns joined rows plus match counts and unmatched rows. Up to 500 rows a side.
local_market_saturation
ChatGPTScores how easy it is to compete in each local area from business rows the user provides (name, area or postcode, rating, review count, website): reviews the top 3 have, the share of weak rivals (no website, few reviews, low rating), other trades filling the results, and the easiest rivals to beat, as a 0 to 100 score per area. Works only on the rows sent; no website is contacted. Up to 200 areas.
make_charts
ChatGPTTurns up to 2,000 rows of data into chart images (bar, line, area, pie, doughnut, scatter) and optionally a PDF report with the charts and a table. Returns a link to each chart and the report.
merge_and_dedupe_lists
ChatGPTMerges two or more lists of JSON rows (for example leads from Google Maps, a CRM export and a spreadsheet) into one list and removes the duplicates between them by a key such as phone, email or website, ignoring spacing, case and http/www. Keeps the most complete row, fills its blanks from the duplicates and labels which lists each record was in. Up to 500 rows in all.
new_zealand_visa_processing_times
ChatGPTOfficial Immigration New Zealand processing times for a visa (how long most applications take), or the list of visas with published times.
ocr_images_and_scanned_pdfs
ChatGPTReads the text in images (PNG, JPEG, WebP, TIFF and more) and scanned or born-digital PDFs with OCR, in 100+ languages, and returns the text per file and per page with confidence scores. Up to 10 files a call.
pagespeed_check
ChatGPTRuns Google PageSpeed Insights (Lighthouse) on up to 10 pages and returns performance, accessibility, best-practices and SEO scores, Core Web Vitals (LCP, CLS, INP, TBT) and the top fixes to make each page faster. Mobile, desktop or both.
pay_transparency_check
ChatGPTChecks a job post against pay transparency laws before it is published. Send the locations (and whether it is remote), employer headcount and posting date. Returns each law that applies or may apply, a merged checklist of what the post must show (pay range, benefits), salary history bans, penalties and the official source per law. Covers US states and cities, Canadian provinces and the EU Pay Transparency Directive country by country. Information, not legal advice.
perplexity_search
ChatGPTAsks up to 5 questions to Perplexity (its official Sonar API with live web search) and returns each full answer with its numbered citations (url, title, domain, position) and the list of cited domains. With brands, domains and competitors it also reports whether the brand is named, its position among the names, which competitors are named, and where the brand's own site is cited. Use it for research with sources, or to see how Perplexity talks about a brand.
poland_company_lookup
ChatGPTLooks up Polish companies in the official KRS court register by KRS number: name, legal form, status, NIP and REGON, address, share capital, board members and registration dates. Up to 20 a call.
product_review_trend
ChatGPTReads product review rows the user provides (product, date, stars and text, for example Amazon reviews): stars in the last 90 days against the 90 before, recent stars against the lifetime rating, stars by month, and the words repeated in 1 and 2 star reviews, with rising complaints flagged. Every number has its sample size and confidence. Works only on the rows sent; no website is contacted. Up to 200 products.
read_web_page
ChatGPTFetches one public web page and returns its main article text, cleaned of menus, ads and footers (Mozilla Readability), plus title, author, publish and update dates, site name, language, image and canonical URL from the publisher's own data. Honours robots.txt. Use it to read a link before summarising or quoting it.
reddit_most_recommended
ChatGPTRanks the brands and products people recommend in Reddit comment rows the user provides for a "best X" thread or search (comment text and upvotes): comments mentioning each name, upvote-weighted score, positive and negative mentions, models seen and the top quote with its link. Text rules, no AI. Works only on the rows sent; no website is contacted. Up to 2,000 comments.
reddit_top_questions
ChatGPTPulls out the questions people ask from Reddit post and comment rows the user provides (title or text, upvotes, replies), groups the same question asked in different words, ranks them by repeats, upvotes and replies, and flags unanswered ones. Good for FAQ pages, content and product ideas. Works only on the rows sent; no website is contacted. Up to 3,000 rows.
salary_benchmark
ChatGPTTurns the pay in job listing rows the user provides (title, location, company and salary text or min, max and period) into yearly pay in one currency, then gives the 25th percentile, median and 75th percentile per job title and city, falling back to the title anywhere and the role family, and flags each job low, fair or high against its peers. Works only on the rows sent. Up to 2,000 jobs.
score_property_deals
ChatGPTScores each for-sale home in listing rows the user provides (price, square feet, zip code, and optionally days on market, price cuts, estimated value, tax value and rent estimate): price per square foot against the median for its zip and home type, price cuts, days on market, rent yield, a 0 to 100 deal score, a verdict and a reason. Works only on the rows sent; no website is contacted. Up to 1,000 listings.
seo_audit
ChatGPTOn-page SEO audit of up to 20 websites or pages: a 0-100 SEO score, A to F grade and the issues to fix, worst first (missing title or meta description, H1, noindex, HTTPS, mobile viewport, thin content, image alt text, Open Graph, structured data, sitemap, robots.txt), plus optional broken-link check.
stock_photo_keywords
ChatGPTLooks at up to 5 images (public image links) and writes stock agency metadata for each: a buyer-friendly title (under 70 characters), a description, up to 49 keywords ranked by importance, the Adobe Stock category (name and number) and Shutterstock categories, plus warnings for logos, recognizable people (model release), readable text and images under 4 megapixels. Brand, artist and AI tool names are removed. Use it when someone sells photos or AI images on Adobe Stock, Shutterstock or Freepik.
uk_business_trust_check
ChatGPTChecks up to 10 UK businesses or traders and returns a trust verdict for each: Companies House status, age, accounts and filings, directors, Gazette insolvency notices, whether the website matches the company, and food hygiene rating where relevant, with the reasons. Give a name, website, company number or postcode.
uk_company_lookup
ChatGPTLooks up UK companies in Companies House by company number: name, status, type, registered address, incorporation date, SIC codes, accounts and confirmation statement due dates, and officers. Up to 20 a call.
uk_company_risk_report
ChatGPTBuilds a risk report on up to 5 UK companies by number or name: Companies House profile, overdue accounts, charges, director history and disqualifications, Gazette insolvency notices and a scored list of red flags with the evidence, optionally as Markdown. Use it before a deal, a supplier contract or credit.
uk_tourist_tax_rules
ChatGPTUK only. Returns the tourist tax (visitor levy) rules, start dates, the next return deadlines and the Wales registration duty for one area, or a list of every area.
uk_waste_tips_near_postcode
ChatGPTFinds household waste recycling centres, tips and trade waste sites near a UK postcode, with distance, address and opening details where published. Choose resident or trade and the materials to get rid of.
upscale_images
ChatGPTUpscales up to 3 images by 2x, 3x or 4x with an AI model and repairs blurry faces, returning a link to each sharper image. Use it to fix small, blurry or low-resolution photos.
us_minimum_wage
ChatGPTReturns the legal hourly minimum wage at a US address (or state plus city or county) on a date: the highest of the federal, state, county and city rates, which one sets it, the tipped cash wage, special industry rates, the next scheduled change and the government source per figure. Use it for payroll, job offers or wage checks. Unclear cases give the highest rate that could apply plus what it depends on. Information, not legal advice.
us_sales_tax_nexus
ChatGPTChecks which US states a remote seller must register in to collect sales tax (economic nexus). Send sales and order counts per state for the last 12 months or a calendar year, and whether they include marketplace sales (Amazon, Etsy). Returns states over the threshold (must register), close to it (80 percent or more) and under it, with each state's threshold, measurement period, what counts and the revenue department source. Information, not tax advice.
validate_phone_numbers
ChatGPTChecks up to 100 phone numbers from any country: valid or not, the number in international E.164 format (+441174960012) and national format, country, and line type (mobile, landline, toll free, VoIP) from Google's libphonenumber data. Set defaultRegion (GB, US...) for numbers written without a country code. Does not call the number.
website_screenshots
ChatGPTTakes screenshots of up to 10 websites, desktop and or mobile, with cookie banners hidden, and returns a public image link for each. Use it to see what sites look like, for audits, lead lists or competitor checks.