Privacy Policy
Last updated: 17 November 2026 · Version 2026-11-17
This Privacy Policy explains how LLMEKNOW (Pty) Ltd ("llmeknow", "we", "us") collects, uses, shares, and protects personal information when you use llmeknow (the "Service"). Our contact details and Information Officer are listed in section 15.
We are the responsible party under POPIA, and the controller under the GDPR and UK GDPR, for account, billing, website, security, and agent request data. For the personal information in content you submit to studies and markets, we act on your behalf as an operator (processor) under our Data Processing Addendum.
Card payments are processed by Paystack. Murmur Intelligence (Pty) Ltd is the merchant of record for those payments: its name appears on your card statement, and it receives the payment details it needs to take payment and keep tax records.
1. Where we operate and where data is processed
We operate from South Africa. Our application servers are hosted by Hetzner in Germany. Our sub-processors process data in South Africa, the European Union, the United States, and other regions where they host their services. Study questions and segment context go to the AI model providers selected for each study.
We transfer personal information outside South Africa only where section 72 of POPIA allows it: where the recipient is bound by law, binding corporate rules, or a binding agreement that gives protection substantially similar to POPIA, or where the transfer is needed to perform our contract with you or to do what you asked us to do. For transfers out of the EEA or UK we rely on adequacy decisions or on the standard contractual clauses (and, for the UK, the international data transfer addendum) offered by our sub-processors.
2. Information we collect
- Account information (name, email address, organisation membership)
- Authentication and session data (strictly necessary cookies)
- Study and market content you submit (questions, segments, brand lists)
- LLM responses and derived analytics from your studies
- Billing and payment metadata (we do not store full card numbers)
- Technical data (browser, device, IP address, application logs)
- Messages you send us, such as support requests and demo requests
- If you use the free brand check: the brand, what it sells, its website, the country and your email address. We use the email address to confirm it is yours and send you the results, and keep it for up to 12 months. We only send you marketing mail if you tick the box that asks for it.
- Conversations with Lemme, our in-app assistant: your messages, its replies, files you attach, the study drafts it helps you build, and the notes it keeps with your approval
- Your email preferences, and records of the service and product emails we send you
- If you connect a Google Analytics 4 property to a study: the property ID, encrypted access tokens, the member who connected it, and the reports we read from it
- If you connect Google Search Console to a study: the property you choose and the search query reports we read from it (queries, impressions and clicks)
- API keys and AI connections: the name you give a key, which AI application you connected, the member who created or approved it, when it was created and last used, any spend cap you set, and who revoked it or changed its cap. Keys, access tokens, and authorisation codes are stored only as one-way hashes, plus the first 12 characters of each key so you can recognise it in Settings.
- Agent requests: for each request an API key or connected AI application makes, the tool it called, the time, the outcome, the amount reserved or charged against its spend cap, the wave it started (if any), a one-way hash of the full request, and the first 2,000 characters of the request it sent
You must give us your email address to create an account, and billing details to pay for a subscription or prepaid funds. Without them we cannot provide those parts of the Service. Everything else is optional.
3. How we use information
- To provide, secure, and improve the Service
- To run studies, extraction, and analytics you request
- To diagnose and improve Lemme: we review its conversations and the traces of how it handled them to fix errors and improve its answers
- To authenticate API keys and connected AI applications, apply spend caps and rate limits, and keep a record of what they did on your organisation's behalf
- To process subscriptions and prepaid funds
- To support customers and respond to enquiries
- To send service emails, such as sign-in codes, receipts, and trial reminders, and to send account holders product updates and tips about using the Service. You can stop product emails at any time with the unsubscribe link in each one, or under Settings
- To comply with law and enforce our terms
4. Legal basis
We process personal information to perform our contract with you (providing the Service), for our legitimate interests (security, fraud prevention, product improvement, measuring our marketing, and telling account holders about the Service), to meet legal obligations (tax and billing records), and with your consent where the law requires it (for example analytics and advertising cookies where we show a consent banner). Where we rely on consent, you can withdraw it at any time. Confirming that you have read this policy at registration is an acknowledgement, not the legal basis for processing. South Africa's POPIA applies; where you are in the EEA or UK, the GDPR or UK GDPR may also apply.
5. Cookies and similar technologies
Essential cookies. Supabase authentication cookies are required to keep you signed in. They cannot be switched off while using the app.
Visitors in the EEA and the UK. We show a consent banner (CookieYes), and advertising and Google Analytics tags wait for your choice. Google tags use Consent Mode v2 (denied by default until you accept the analytics or advertising categories). The LinkedIn Insight Tag loads only after advertising consent.
Visitors elsewhere, including South Africa. We do not show a consent banner. Google Analytics, Google Ads, and the LinkedIn Insight Tag load when you visit the site. You can block or delete these cookies in your browser settings, or use the opt-out tools Google and LinkedIn provide.
Product analytics (PostHog, EU-hosted). We use PostHog to measure how the marketing site and product are used (for example page views and which buttons are clicked) and to record errors. It also keeps session replays: recordings of layout, scrolling and clicks in which all page text and form input is hidden, and images, charts, links to other sites and labelled buttons are blanked out. Click records name the element clicked, not its text, and keep links without their query parameters. In the EEA and the UK, replays and click records start only after you accept the analytics category, and until then PostHog keeps its identifier in page memory only, discarded when you leave; after you accept, it stores a durable identifier (in its cookie and local storage) so we can recognise return visits. Elsewhere it does all of this when you visit. Where we show the banner, you can change cookie preferences via the CookieYes controls on the site. For a summary of the cookies we use and how to manage them, see our Cookie Notice.
6. Sub-processors
We use the following sub-processors to run the Service. Where we link them, you can read each provider's privacy policy and its data processing agreement (DPA):
- Supabase (database, authentication): Privacy, DPA
- Cloudflare (CDN, Turnstile bot protection, email sending): Privacy, DPA
- Paystack (payments, PCI DSS Level 1): Privacy
- OpenAI (LLM inference via our proxy): Privacy, DPA
- Anthropic (LLM inference): Privacy, Commercial terms (incl. DPA)
- Google (Gemini / Vertex AI inference): Privacy, Cloud DPA
- xAI (Grok inference): Privacy
- DeepSeek (LLM inference, only for studies that select a DeepSeek model)
- Moonshot AI (Kimi LLM inference, only for studies that select a Kimi model)
- OpenRouter (image generation for segment portraits, from segment descriptions)
- Exa, Firecrawl, and Bright Data (web search, page fetching, and search result collection for Search Intelligence and source checks)
- Logo.dev (brand logos, looked up by brand name or domain)
- Hetzner (application server hosting, Germany)
- Linear (issue tracking for bug reports and feedback you send from the app, including your email address and any screenshot you attach)
- Pydantic Logfire (observability: error and performance traces of the Service, with content tracing disabled for Lemme and connected Google report analysis; this does not remove historical traces): Privacy
- PostHog (product analytics, EU cloud): Privacy, DPA
- CookieYes (cookie consent banner): Privacy
- Google Analytics / Google Ads (marketing measurement, consent-gated): Privacy
- LinkedIn Insight Tag (advertising measurement, consent-gated): Privacy
Prompts sent to LLM providers contain study questions and segment context you define. Do not include personal data about identifiable individuals unless you have a lawful basis to do so.
7. Data retention
We retain data only as long as needed for the purposes below, then delete or anonymise it:
- Study run outputs (raw LLM text and response metadata): 12 months from creation, then cleared while aggregated analytics may remain.
- Billing and payment records: 7 years (statutory and tax requirements).
- API usage / cost logs: 2 years.
- Agent request logs (requests made with API keys and AI connections): 2 years.
- Observability traces (held by Pydantic Logfire; historical traces may include Lemme conversation content): 30 days.
- API key and AI connection records, including who created, revoked, or re-capped each one: until the organisation is deleted, so a revoked key or connection stays visible in its history.
- AI connection tokens: access tokens expire after one hour and refresh tokens after 30 days. Only their hashes are stored.
- Deleted accounts: when you delete your account we delete your sign-in account and remove your email address from billing and invitation records straight away. Any remaining personal identifiers on your profile are anonymised within 30 days. Organisation data you created may be retained where other members still use the workspace, unless the organisation is deleted.
Automated retention jobs run daily. You may request earlier deletion where the law allows.
8. AI applications and API keys you connect
You can give an AI application, such as Claude, ChatGPT or Cursor, access to your organisation in llmeknow, or create an API key for your own software. When you do, that application sends us requests and we send it the results it asks for. Results can include your study and market content, AI responses, synthetic audience members, analytics, your organisation's name and plan, its wallet balance, and its billing history (amounts and dates). A connected application can also create markets, segments, studies, and audience members, and start paid work such as waves, within the limits in our Terms of Service.
The AI application you connect is not our sub-processor. It acts for you, under your agreement with its provider, and its provider's terms and privacy policy govern what happens to the results once they reach it. We do not receive your conversation with the AI application, only the requests it sends us. Those requests can contain text the application wrote from your conversation, such as a question you asked it to add to a study, and we keep the first 2,000 characters of each one as described in section 2.
Only an owner or admin can connect an AI application or create an API key. Each one acts with the current permissions of the member who created or approved it, and stops working if that member leaves the organisation or deletes their account. An owner or admin can revoke a key or connection at any time under Settings, API keys. Revoking stops further requests; it does not recall results the application has already received, and work it already started, such as a running wave, continues until it finishes or you pause it in the app.
9. Google data you connect
When you connect Google Analytics or Google Search Console to a study, the connection is read-only. We use the reports to provide study analysis and AI assistance you request. Relevant report content may be sent to the AI model providers described in section 6. We do not sell connected Google data or use it for advertising or to train our own AI models.
Disconnecting one connector stops that connector in the study. If another Google connector remains enabled in the same study, we keep the shared Google connection for it. When the last connector is disconnected, we delete the study’s stored Google tokens and ask Google to revoke access. If Google does not confirm that request, the app shows a warning and a link to remove access in your Google Account. Removing access there can affect other studies connected with the same Google account.
Disconnecting does not delete reports or conversations already saved in the study, or historical records held by our service providers. Retention and deletion requests are described in sections 7 and 11.
llmeknow's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.
10. Security
We use technical and organisational measures including encryption in transit, access controls, row-level security in our database, and secret management. No method of transmission or storage is completely secure.
11. Your rights
Depending on applicable law, you may ask us whether we hold personal information about you and for a copy of it, and you may request correction, deletion, restriction, or portability, or object to processing, including processing for direct marketing. Where we rely on consent you may withdraw it at any time; this does not affect processing that already happened. Contact us using the details in section 15. You may lodge a complaint with the Information Regulator (South Africa) or your local supervisory authority.
If we believe that personal information has been accessed or acquired by an unauthorised person, we will notify the Information Regulator and affected people as POPIA and other applicable law require.
12. Data processing agreements
Enterprise customers may request a countersigned DPA covering our processing of personal data on your behalf. Contact [email protected]. Our standard terms are published in our Data Processing Addendum.
13. Children
The Service is intended for business use and is not directed at children. You must be at least 18 years old to use it. If you believe we have collected personal information from a child, contact us and we will delete it as the law requires.
14. Changes to this policy
We may update this policy from time to time. When we make material changes we will update the "Last updated" date and version above and ask you to re-accept the policy at your next checkout. Continued use of the Service after a change takes effect means the updated policy applies to your use of the Service.
15. Contact
Responsible party: LLMEKNOW (Pty) Ltd, registration number 2026/498002/07, WeWork, 80 Strand Street, Cape Town, Western Cape, 8001, South Africa.
Information Officer: the head of LLMEKNOW (Pty) Ltd (the Information Officer under POPIA), reachable at [email protected]
Privacy enquiries: [email protected]
General support: [email protected]
Information Regulator (South Africa): inforegulator.org.za, complaints to [email protected]. In the EEA or UK you may also contact your local data protection authority, such as the UK Information Commissioner's Office at ico.org.uk.