Privacy Policy
Last updated: 9 November 2026
At SureReplies we respect your privacy and are committed to protecting your personal information. This Privacy Policy explains how we collect, use, store, process, share and safeguard information when you visit our website at surereplies.com, and when you or your organisation use SureReplies, our hosted AI documentation and support-chatbot platform at app.surereplies.com.
SureReplies is an AI product. A significant part of what the service does is send text to third-party artificial-intelligence providers so that answers can be generated. We have written the section 'Artificial intelligence and large language models' below to describe that in plain language, because it is the part of this policy most people need to read.
This is a single policy covering two different surfaces, and the obligations are not the same on each. Each section states clearly whether it applies to the website, to the platform, or to both.
Who we are
SureReplies is a Software-as-a-Service platform operated by Brainstorm Force US LLC, which is the controller responsible for the personal data described in this policy, except where we state that we act as a processor on a customer's behalf.
Brainstorm Force US LLC, 2093 Philadelphia Pike #3090, Claymont, Delaware 19703, United States.
Scope of this policy
The marketing website - surereplies.com
- Our public pages, our documentation pages, and our forms
- The cookies, storage and third-party requests those pages create
The platform - app.surereplies.com and api.surereplies.com
- The SureReplies dashboard, where account holders sign in, create bots and attach knowledge sources
- The APIs and backend services behind it
- The embeddable chat widget our customers place on their own websites, and the chat iframe it opens
- The AI pipeline that answers questions, including every third-party AI provider it sends text to
- Our integrations with third-party systems that customers choose to connect, such as GitHub, Google Drive, WordPress, FreeScout, Zendesk, Help Scout, DocsBot, SureContact, Cal.com and calendar-booking providers
- The Model Context Protocol (MCP) interface, which lets an account holder connect an external AI client to their own SureReplies data
What this policy does not cover
- Websites operated by our customers. If you met SureReplies as a chat bubble on somebody else's website, that website's operator decides what is collected and why. See their privacy notice.
- Any third-party service you reach by following a link from ours.
Our role in relation to your data
We act in two distinct roles. Which one applies determines who you should contact about your data, and who decides how long it is kept.
We are a controller
We are the controller for the personal data we collect for our own purposes: the accounts our customers' staff hold with us, the security and audit records we keep about those accounts, our own support correspondence, our billing records, and the limited processing that happens on our marketing website.
That includes, for account holders: name, email address, hashed password, Google account identifier where sign-in with Google is used, avatar URL, language preference, two-factor authentication secrets and recovery codes, and the IP addresses and browser user-agent strings recorded in our session, activity-log and login-attempt records.
We are a processor
We are a processor for the personal data that our customers' own end users generate through the chat widget, and for the content our customers load into the platform. Our customer, the organisation that created the bot and embedded the widget, is the controller of that data. They decide what is collected, why, on which pages, whether to ask visitors for a name and email, which AI model is used, how long the data is kept, and what the lawful basis is.
Data for which we act as a processor includes: verbatim chat questions typed by a customer's website visitors and the answers our AI returns; AI-generated summaries of those chat transcripts; leads captured in chat (a visitor's name and email address); visitor email addresses submitted through the widget's support-ticket feature; attendee names and email addresses submitted through the widget's meeting-booking feature; support tickets our customers import from FreeScout, Zendesk or Help Scout, which routinely contain their own end users' real names, email addresses and case histories; and documentation, website content, uploaded files, GitHub repository content, Google Drive files and WordPress content that a customer attaches as a knowledge source.
If you are a website visitor and you want to access, correct or delete something you typed into a SureReplies chat widget, please contact the operator of the website where you used it. They control that chat and can delete the conversation from their SureReplies dashboard. To help them find yours, you can share the visitor identifier that SureReplies saves in your browser's local storage for that site. If you contact us instead, we will identify the operator, pass your request to them, and tell you that we have done so.
Where we get your data from
- Directly from you, when you create an account, sign in, contact support, complete a form on our website, or type into a chat widget.
- Automatically from your device and browser, when you visit our website or use the platform.
- From our customers, when they attach knowledge sources, import support tickets, invite colleagues, or deploy the widget on their site.
- From services you connect, such as Google (sign-in and Drive), GitHub, FreeScout, Zendesk, Help Scout, DocsBot and WordPress, where a customer authorises the connection.
- From an affiliate referral link, if you arrive at our website through an affiliate link, we record the referring affiliate's identifier.
We do not buy personal data, and we do not enrich our records from data brokers.
Information we collect
A. Account-holder information - we are the controller
- Name, email address, email-verification status, avatar URL and language preference.
- Your password, stored as a one-way hash, which we cannot read. If you sign in with Google, we store no password at all, only your Google account identifier, email, name and avatar URL.
- Two-factor authentication secrets and recovery codes, encrypted at rest.
- Password-reset and email-verification tokens, which expire.
- Organisation and workspace records: name, description, logo, brand colour, timezone, locale, retention setting, and your role and capabilities within them.
- Invitations you send, including the invitee's email address and any message you write.
- API keys you create, stored as hashes with a short non-secret prefix, so we cannot recover the key itself.
- Credentials and connected-account details for integrations you authorise, including the name and email address of the connected account at FreeScout, Zendesk, Help Scout or Google Drive, encrypted at rest.
- Your own AI provider API key, if you supply one, encrypted at rest.
- Your reCAPTCHA site and secret keys, if you enable the widget spam gate.
- Notifications we generate for you, and content you author in the product such as curated answers, feedback on sources and conversation review actions.
Security and audit records. We record: web sessions with your IP address and user-agent string; an activity log of actions taken in the product with your IP address and user-agent string; and login attempts with the email address used and the IP address it came from, including attempts against email addresses that do not have an account, because that is how we detect credential-stuffing.
Product usage and internal reporting. We compile daily counters about each organisation (number of bots, conversations, messages, sources, plan and trial status) and send them to our internal analytics service at analytics-api.brainstormforce.com. That transmission includes the organisation owner's email address, along with organisation and bot identifiers and names.
Payment information
Billing and subscriptions are handled by SureCart, which uses Stripe and PayPal as its payment processors. We record your plan, trial and subscription status in our own database; SureCart manages checkout and the subscription, and your card or payment details are entered directly with Stripe or PayPal. Subscriptions are billed yearly. We do not hold card numbers or card security codes in the platform.
B. End-user chat information - we are a processor
When a member of the public uses a SureReplies chat widget on one of our customers' websites, the following is processed on that customer's behalf:
- The question. Before it is stored, every question is checked for passwords, login links and access tokens, which are removed. If the organisation has turned on PII redaction, further personal data is replaced with placeholders. The remaining text is stored in full, because chat questions frequently contain personal information.
- The answer the AI produced, together with which knowledge chunks were cited, confidence and validation scores, and any thumbs-up or thumbs-down feedback.
- An AI-generated summary of the conversation, and a machine-assigned resolution outcome.
- A visitor identifier, a random value generated in the visitor's browser and sent with every chat request, used to group that browser's conversations together and to give the bot memory of earlier conversations.
- A copy of the question text in two further internal records: an optional query-metrics table used for search-quality analysis, and a spam-fingerprint record holding the first 500 characters of the question, used to recognise repeat abuse.
- A lead, if the widget's lead form is used: name and email address, encrypted at rest and linked to the visitor identifier.
- A support-ticket request, if the widget's ticket feature is used: the visitor's email address, the ticket tier, and the identifier of the ticket created on the customer's own helpdesk. The chat transcript is sent onward to that helpdesk.
- A meeting booking, if the widget's booking feature is used: attendee name, email address, timezone and the chosen slot.
- Deflection and interaction telemetry: event type and response latency.
Visitor IP addresses are not stored. Our abuse and rate-limiting controls key on a one-way hash of IP address and user-agent held in a temporary cache and then discarded. However, IP addresses are necessarily visible to our CDN, and to Google where a customer has enabled reCAPTCHA on their widget.
Questions can also arrive through channels you connect, such as Slack, your FreeScout helpdesk, the WordPress plugin, our API and the MCP interface. The same processing, redaction and retention rules described in this policy apply to every channel.
C. Customer content and knowledge sources
Account holders attach knowledge sources: website pages and sitemaps, uploaded documents, GitHub repositories, Google Drive files, WordPress content, and imported helpdesk tickets. We fetch that content, split it into chunks, convert each chunk into a vector embedding using a third-party AI provider, and store both the text and the embedding so the bot can search it. Where a customer attaches content that contains other people's personal data, and imported helpdesk tickets almost always do, that content is stored by us and sent to an AI provider for embedding and later as retrieval context. The customer is the controller of it.
D. Automatically collected technical data
Both the website and the platform automatically receive limited technical data: browser and platform information, referring URLs, request timestamps, error diagnostics and feature-interaction events. We use it for security, reliability and product improvement.
Which information is required
An email address and either a password or a Google sign-in are necessary to open and keep an account; without them we cannot authenticate you. Two-factor authentication, an avatar, a language preference, an organisation logo and brand colour, integration connections and your own AI provider key are all optional, and leaving them out does not restrict any feature you would otherwise have.
Sensitive and special-category data
We do not seek or intentionally collect special-category data such as health information, racial or ethnic origin, religious or political beliefs, trade union membership, genetic or biometric data, or sexual orientation. Please do not type such information into a chat widget or a support request. Because chat input is free text that we store verbatim and send to an AI provider, we have no technical means of preventing you from doing so. The only data we treat as sensitive on our own account is authentication material: passwords, two-factor secrets and recovery codes, API keys and connected-service credentials.
Artificial intelligence and large language models
This section applies to the platform. Read it before you deploy a SureReplies widget, and before you attach a knowledge source. It is written for the person who has to sign off on the risk.
How SureReplies uses AI with your data
What is sent. To answer a visitor's question, SureReplies sends the question, the most relevant passages from your own knowledge sources, your bot's instructions and recent messages in the conversation to an AI provider. We send it only to generate the answer and to run checks on it (spam detection, answer quality, conversation summaries). When you add a knowledge source, its text is also sent to OpenAI once to create the search index (embeddings).
Who receives it. OpenAI receives all embedding requests and answer requests for OpenAI models. OpenRouter receives answer requests for the Anthropic Claude model you can pick and passes them to Anthropic. The model you choose for your bot decides which vendor sees your visitors' questions. The full list is in the sub-processor table below.
Training. For models reached through OpenRouter (currently Anthropic Claude), we send an instruction on every request telling OpenRouter to route only to endpoints that do not store or train on the content we send. OpenAI, which generates all embeddings and the answers for OpenAI models, is used under its standard commercial API terms. We do not use your content to train our own models.
What we remove first. Passwords, login links and access tokens are always removed from questions before they are stored or sent to any provider. If you turn on PII redaction in your organisation settings, email addresses, phone numbers, card numbers, Social Security numbers, IP addresses and licence keys are also replaced with placeholders such as [EMAIL_1]. Names and addresses are not detected, so please ask visitors not to share sensitive information in chat.
Your own AI key. On paid plans you can connect your own OpenAI or OpenRouter key. Answer generation and indexing then run under your provider account, billing and terms. A few small internal checks (spam and intent detection, translations of canned replies, conversation summaries) still use our account. Your key is stored encrypted and only your organisation owner can add or remove it.
What we keep. We store questions and answers with your bot so you can review them. We do not store visitors' IP addresses. Conversations are kept until you set a retention period in your organisation settings; leads are deleted 730 days after capture; AI usage records hold no question or answer text.
Who is responsible. You decide what your bot knows and how it behaves, so you are the controller of your visitors' data and we process it on your behalf and on your instructions. AI answers can be wrong; please tell your visitors they are chatting with an AI assistant.
Which AI providers receive data
| Provider | How it is reached | What it is used for |
|---|---|---|
| OpenAI | Directly, api.openai.com | Answer generation, question classification, spam and support-intent classification, answer validation, conversation summarisation, and all vector embeddings |
| OpenRouter | Directly, openrouter.ai | A routing broker. Every model whose name contains a slash is sent through OpenRouter rather than to the model's own vendor |
| Anthropic | As a downstream provider of OpenRouter | Answer generation, where a customer selects the Claude model |
| Hugging Face | Directly, if enabled | Optional re-ranking of search results. Disabled by default and not currently enabled |
Two points follow from this table. First, the customer's choice of model determines which company receives their end users' chat data, because OpenRouter is a broker and not a destination. Second, vector embeddings are always generated by OpenAI, regardless of which chat model is selected, because our vector index is built for one specific OpenAI embedding model. If a transient failure occurs, our retry logic may re-send the same request to a different provider than the one originally selected, so the same content can reach two providers for a single question.
Exactly what is sent to an AI provider
For a single chat question, the request sent to the AI provider can contain the end user's question verbatim; retrieved chunks of the customer's documentation, including the chunk text, source URL, page title and relevance scores (where the customer has imported helpdesk tickets, these chunks may be the text of real support tickets); the bot owner's custom instructions and welcome-card context; the host page context (the domain the widget is embedded on, and the URL and title of the page); and conversation history. Where cross-conversation visitor memory is enabled, up to 100 previous messages drawn from every conversation that browser has ever had with that bot may be sent.
Additional AI calls send further content: conversation summarisation sends the entire transcript, up to roughly 12,000 characters; spam and support-intent classification send the raw message as typed; answer validation and outcome classification send the answer text and excerpts of the cited chunks; helpdesk resolution classification sends the full text of an imported support-ticket conversation; and embedding sends every chunk of every knowledge source a customer attaches, every question asked, and the text of curated answers. The MCP interface exposes conversation search, retrieval, message traces, member lists and data export to an external AI client that an account holder connects; anything that client pulls goes to that client's own AI provider, chosen by the account holder.
What we remove before data is sent
Passwords and login secrets, always. Every incoming question, from any channel, is checked for login links, access tokens, passwords and labelled credential fields. These are replaced with a placeholder before the question is saved, logged or sent to an AI provider. The same check runs on every imported helpdesk ticket.
Personal data, when you turn it on. Organisation owners can switch on PII redaction in organisation settings. When it is on, email addresses, phone numbers, card numbers, US Social Security numbers, IP addresses and licence keys in visitor questions are replaced with placeholders before the question is stored or sent anywhere. We keep no copy of the original values. It is off by default, and it does not detect names or postal addresses, does not change the content of your knowledge sources, and does not change conversations stored before you turned it on.
Retention and training at the AI providers
We use each provider's standard commercial API. For models reached through OpenRouter (currently Anthropic Claude), we send a data-collection 'deny' instruction on every request, which tells OpenRouter to route only to endpoints that do not store or train on the prompts we send. This is enforced per request in our configuration, not left to an account-level setting. For OpenAI, which receives all embeddings and the answers for OpenAI models directly, we use the standard commercial API. We do not use your content to train our own models, and we do not make any additional claim about a provider's own retention beyond its published terms, which are linked below.
- OpenAI business and API data usage and retention
- Anthropic commercial terms (no training on customer content)
- OpenRouter privacy, logging and data-policy controls
Bringing your own AI provider key
An organisation may supply its own OpenAI or OpenRouter API key so that answer generation runs under its own provider account and its own provider terms. The key is stored encrypted at rest, and only the organisation owner can add or revoke it. With a customer-supplied OpenAI key, the embeddings generated when indexing your knowledge sources also run on your key. Some internal calls always run under our provider account whichever key you supply: spam-intent and support-intent classification, canned-reply translation, and conversation summarisation (which sends the full transcript). If a request made on your key fails, it may be retried on our platform key so the visitor still receives an answer.
What we store of the AI's output
Answers are stored with the conversation, together with cited-source identifiers, confidence and validation scores. Conversation summaries are stored on the conversation record. Vector embeddings of customer content are stored in our own PostgreSQL database using the pgvector extension; we do not use any external vector database. Our AI cost ledger records organisation, bot and message identifiers, the call type, the model, token counts and cost. It contains no prompt or completion text.
Accuracy and automated decision-making
Answers are generated by a language model from the customer's own content. They can be wrong, incomplete or out of date, and they are not advice. Nothing in the product makes an automated decision that has a legal or similarly significant effect on any individual, and we do not use chat content to build profiles of individuals.
Why we process your data, and our legal basis
Where the EU GDPR or UK GDPR applies, we rely on the following legal bases.
| Purpose | Applies to | Legal basis |
|---|---|---|
| Creating and operating an account, authenticating you, managing organisations, workspaces, roles and invitations, delivering the platform and providing support | Platform | Performance of a contract, Art. 6(1)(b) |
| Sending question text, retrieved content and conversation history to an AI provider to generate an answer, and generating embeddings of attached content | Platform | Performance of a contract, Art. 6(1)(b). For end-user data we act as a processor and the customer determines the basis |
| Processing end-user chat content, leads, widget support tickets, bookings and imported helpdesk tickets on a customer's behalf | Platform | We act as a processor; the customer determines the purpose and the legal basis |
| Transactional email: verification, password reset, invitations, and the conversation digests a customer subscribes to | Platform | Performance of a contract, Art. 6(1)(b) |
| Security and abuse prevention: session records, activity logs, login-attempt records, rate limiting, spam detection, the optional widget reCAPTCHA gate, and protecting our infrastructure | Both | Legitimate interests, Art. 6(1)(f) |
| Reliability, fault diagnosis and service monitoring | Both | Legitimate interests, Art. 6(1)(f) |
| Internal product reporting, using per-organisation counters and the organisation owner's email address | Platform | Legitimate interests, Art. 6(1)(f) |
| Billing, invoicing, and tax and accounting records | Platform | Legal obligation, Art. 6(1)(c), and performance of a contract, Art. 6(1)(b). Billing runs through SureCart, with Stripe and PayPal as the payment processors |
| Marketing emails and product announcements | Both | Consent, Art. 6(1)(a), or legitimate interests where permitted for existing customers, with an opt-out in every message |
| Responding to lawful requests from courts, regulators and law enforcement | Both | Legal obligation, Art. 6(1)(c) |
Where we rely on consent, you may withdraw it at any time. Where we rely on legitimate interests, you may object, and we will stop unless we have compelling grounds to continue.
Cookies and similar technologies
The authenticated application at app.surereplies.com sets only the cookies it needs to run: a session cookie to keep you signed in, a CSRF token cookie to protect form submissions, and a locale cookie to remember your language. These are strictly necessary and are not used for advertising, analytics, retargeting, session replay or tracking of any kind. The platform runs no advertising or session-replay technology.
Our marketing website at surereplies.com will have its own cookie notice and consent controls, and where it uses non-essential cookies they will load only after you accept the relevant category. This policy will be updated to describe them when that site is live.
The chat widget and your visitors' browsers
This section applies to the platform, and it is written for our customers as much as for the people who use their websites. If you embed a SureReplies widget, you need to know what it writes to your visitors' devices, because the consent banner on your site is yours and not ours. The widget loader itself sets no cookies. The chat window stores the following in your visitors' browsers:
| Key | Storage | Contents | Lifetime |
|---|---|---|---|
| Visitor identifier | localStorage | A random value, per bot, per browser | No expiry. It persists until the visitor clears site data |
| Conversation pointer | localStorage | The current conversation ID | 24 hours |
| Lead record | localStorage | The visitor's name and email address, in readable form, once submitted through the lead form | 90 days |
| Lead-form skip marker | localStorage | A timestamp recording that the visitor declined the lead form | 90 days |
| Transcript, feedback and support-ticket state | sessionStorage | The current chat, for continuity while the tab is open | Until the tab is closed |
| Widget UI state | localStorage | Interface preferences only, no personal data | Persistent |
The visitor identifier does not expire. It is deliberately durable so a returning visitor's conversations can be linked and the bot can remember them. It is first-party to the widget and scoped to a single bot, so it is not a cross-site tracker. The lead record holds a name and an email address in readable form on the visitor's own device for 90 days, so the visitor is not asked twice. Under EU and UK ePrivacy rules, writing these entries generally requires the same consent basis as a cookie, so our customers must cover the widget in their own cookie notice and consent mechanism.
Consent, Global Privacy Control and Do Not Track
The platform is an authenticated application and sets only the session and preference cookies needed to run it. We do not sell personal information. Where browsers send a Global Privacy Control signal, we honour it. There is no agreed standard for how a website should interpret the browser Do Not Track signal, so we do not respond to it, and we offer the controls described in this policy instead.
Service providers and sub-processors
We use third parties to deliver the service. Which of them receive personal data depends on what a customer configures; several are only ever contacted because a customer connected them.
| Function | Provider | What they receive | Role |
|---|---|---|---|
| AI answer generation, classification and all embeddings | OpenAI | End-user questions, retrieved customer content chunks, conversation history and visitor memory, bot answers, full transcripts for summarisation, full imported ticket text, every chunk of every knowledge source | Sub-processor for end-user data |
| AI model routing | OpenRouter | The same chat payloads, for any model routed through it, plus our product name and application URL | Sub-processor and onward-transfer broker |
| AI answer generation, via OpenRouter | Anthropic | The same chat payloads, where a customer selects the Claude model | Downstream sub-processor of OpenRouter |
| Dashboard sign-in, and the Drive knowledge-source connector | Google (OAuth, Drive API, Picker) | Google account ID, email, name, avatar; selected Drive file metadata and contents under a per-file consent scope | Service provider |
| Chat widget spam check, where a customer enables reCAPTCHA on their widget | Google reCAPTCHA | The widget visitor's IP address and browser signals. Google's reCAPTCHA script loads in the visitor's browser inside the chat widget | Service provider |
| Repository documentation ingestion | GitHub | Repository, branch and file requests. Pull only | Service provider |
| Helpdesk ticket import, and widget ticket creation | Zendesk, Help Scout, FreeScout (customer-hosted) | Inbound: agent identity, tickets and comments. Outbound: ticket subject, the chat transcript, and the visitor's name and email address | Sub-processor, configured by the customer |
| One-off migration import from a competitor | DocsBot | The customer's DocsBot credentials and the imported content | Sub-processor, configured by the customer |
| CRM lead delivery | SureContact | Widget visitor email address, first and last name, bot name and tags | Sub-processor, configured by the customer |
| In-chat meeting booking and embedded schedulers | Cal.com, Calendly, or a custom scheduler such as TidyCal or Google Calendar | Cal.com in-chat booking: attendee name, email address, timezone and slot, sent by us through its API. Cal.com and Calendly embeds: the visitor's name and email address, where they gave them in the lead form, are pre-filled into the scheduler page, and the visitor completes the booking directly with that provider. Custom schedulers: the page opens unchanged and we pass no attendee data | Sub-processor, configured by the customer |
| Billing and subscriptions | SureCart | Account and billing identifiers and subscription status; checkout and the subscription are handled by SureCart and we store no card data | Service provider |
| Payment processing, behind SureCart | Stripe, PayPal | Card and payment details entered at checkout, and payment and subscription identifiers. Entered directly with the processor; we do not receive or store card numbers | Sub-processor, behind SureCart |
| CDN, reverse proxy, security and edge measurement | Cloudflare | All request traffic to our website and platform, including IP addresses | Service provider |
| Bot protection on the sign-up and forgot-password pages | Cloudflare Turnstile | The IP address and browser signals of the person completing the page. Cloudflare's Turnstile script loads in the browser on those two pages | Service provider |
| Transactional and digest email | Amazon SES | Recipient email address. Conversation digests carry chat content | Service provider |
| Outbound event notifications | Customer-configured webhook URLs | The customer's own endpoint receives the verbatim end-user question and the first 500 characters of the answer. No visitor email, name or IP | Destination chosen and controlled by the customer |
| Internal product reporting | analytics-api.brainstormforce.com | Per-organisation counters plus the organisation owner's email address, and organisation and bot names | Internal Brainstorm Force service |
Sub-processors and agreements. Each provider in the table above is bound by a written agreement that limits it to processing personal data on our instructions and requires appropriate security measures. Where we act as a processor for a customer, we will notify that customer of material changes to this list before a new sub-processor begins processing their data.
Where your data is processed, and international transfers
Our platform runs on servers we manage ourselves, fronted by Cloudflare. Our database, cache, uploaded files and generated exports sit on that infrastructure. Beyond our own infrastructure, personal data is processed in the United States, where several of our providers (including our AI providers) are established; in India, where much of our engineering, product and support work is carried out; and wherever else our providers operate infrastructure. Because Cloudflare serves traffic from points of presence worldwide, requests transit Cloudflare infrastructure in the region nearest to the visitor.
Where we transfer personal data out of the European Economic Area, the United Kingdom or Switzerland, we rely on an adequacy decision by the European Commission or the relevant authority; the European Commission's Standard Contractual Clauses, together with the UK International Data Transfer Addendum where applicable; or another lawful transfer mechanism permitted by the applicable law. We do not claim certification under the EU-U.S. Data Privacy Framework, the UK Extension or the Swiss-U.S. Data Privacy Framework. You can ask us about the safeguards applying to a specific transfer, including a specific AI provider, by writing to the privacy address above.
How long we keep data
We keep personal data only as long as we need it for the purposes described in this policy, then delete it.
| Data | Retention |
|---|---|
| Account and organisation records | For the life of the account. Deleting an organisation removes it from view immediately and permanently deletes it 30 days later |
| End-user chat conversations and messages | There is no automatic deletion by default. Chat content is kept until the customer who owns the bot sets a retention window for their organisation, after which older conversations are purged nightly |
| A purged conversation | Message content is permanently deleted and the AI summary is erased. The conversation shell (its identifier, the visitor identifier, the timestamp and aggregate counts) is retained |
| Leads | Deleted automatically 730 days after capture. You can delete individual leads sooner at any time |
| Lead-form skip records | 180 days |
| Activity log, including account-holder IP address and browser | 90 days |
| Login-attempt records, including email address and IP address | 24 hours, pruned whenever a new attempt is logged |
| Pending invitations, including the invitee's email address | 7 days |
| Spam fingerprints, holding the first 500 characters of a question | 90 days from the last time the same question was seen |
| Search query metrics and deflection events | 90 days |
| Uploaded files attached as knowledge sources | Removed when you delete the source. When a bot or organisation is deleted, its records are purged 30 days later |
| Deleted bots, organisations, workspaces and conversations | 30 days in trash, then permanently deleted |
| AI cost records | Individual call records about 30 days, then rolled into monthly counters kept indefinitely, containing no prompt or answer text |
Where a customer uses the widget to collect chat, leads, tickets or bookings from their own visitors, that customer controls how long it is kept and can ask us to delete it at any time.
Data security
We use industry-standard security measures, including:
- TLS encryption for data in transit.
- One-way hashing of account passwords, so we cannot read them.
- Encryption at rest for two-factor secrets and recovery codes, integration credentials, customer-supplied AI provider keys, and lead names and email addresses.
- API keys stored only as hashes, so a key cannot be recovered from our database.
- Role-based access control scoped to an organisation, and internal access limited on a need-to-know basis.
- Deliberate avoidance of storing end-user IP addresses; abuse controls use short-lived one-way hashes instead.
- Request-time IP pinning on outbound fetches (webhooks, content crawling and scheduling) to prevent server-side request forgery.
- Narrow, per-file consent scope for the Google Drive connector rather than access to a whole Drive.
- Authenticated, organisation-scoped export downloads with no public or signed URLs, and export files sanitised against spreadsheet formula injection.
- Regular security reviews.
No system is completely secure, but we take reasonable steps to protect all data we hold.
Security incidents and breach notification
If a personal data breach occurs, we will:
- notify the competent supervisory authority within 72 hours of becoming aware of it, where Article 33 of the GDPR or an equivalent law requires it;
- inform affected individuals without undue delay where the breach is likely to result in a high risk to their rights and freedoms, in line with Article 34;
- notify the affected customer without undue delay where we act as a processor, so they can meet their own obligations;
- and record the facts of the breach, its effects and the remedial action taken.
Your privacy rights
Depending on where you live, you may have the right to:
- Access the personal data we hold about you, and be told how and why we process it.
- Correct inaccurate or incomplete data.
- Erase your data, the right to be forgotten.
- Restrict our processing while a dispute about accuracy or lawfulness is resolved.
- Object to processing based on our legitimate interests, and to direct marketing at any time.
- Data portability: receive your data in a structured, commonly used, machine-readable format.
- Withdraw consent at any time, where our processing is based on consent.
- Opt out of marketing communications, using the unsubscribe link in any message or by writing to us.
- Not be subject to solely automated decision-making that has a legal or similarly significant effect on you. We do not carry out such decision-making.
- Lodge a complaint with your local data protection authority.
Exercising any of these rights is free of charge, and it will never result in worse treatment, a reduced service or a different price.
How to exercise your rights
Email [email protected] and tell us which right you want to exercise. We may ask for information to verify your identity before we act. Requests under the GDPR and UK GDPR are answered within one month, extendable by up to two further months for complex or numerous requests, with notice. Requests under California law are answered within 45 days, extendable once by a further 45 days with notice. Requests under India's DPDP Act follow the grievance timelines in that section below.
Erasure requests
Organisation owners can delete an organisation from settings; it is removed from view immediately and permanently deleted after 30 days. To delete your personal user account, email [email protected] from the account's email address. We will complete the deletion within 30 days and confirm what was deleted and anything we must keep (for example invoices required by tax law).
If your request concerns a chat on somebody else's website
We are a processor for that data and cannot act on it ourselves. Tell us the website and roughly when, and we will identify our customer, forward your request to them as the controller, and confirm to you that we have done so.
Your California privacy rights (CCPA/CPRA)
If you are a California resident, the California Consumer Privacy Act as amended by the California Privacy Rights Act gives you the right to know, access, correct, delete, opt out of the sale or sharing of your personal information, limit the use of sensitive personal information, and to non-discrimination for exercising these rights. You may exercise them yourself or through an authorised agent.
We do not sell personal information, and we do not share personal information for cross-context behavioural advertising. Where we disclose personal information to a service provider or sub-processor (our AI providers, our CDN, our email relay, and the integrations a customer connects) it is a disclosure for a business purpose under a contract that limits them to processing on our instructions. It is not a sale and it is not sharing for advertising.
| Category | Collect? | Disclosed for a business purpose to | Sold or shared for advertising? |
|---|---|---|---|
| Identifiers: name, email, account and org IDs, Google account ID, IP in security records, widget visitor identifier | Yes | Hosting and CDN, email relay, AI providers (where the identifier appears in content), customer-connected integrations | No |
| Commercial information: plan, trial, subscription status | Yes | Our own systems; SureCart | No |
| Internet or network activity: pages viewed, the page URL a widget was used on, chat and deflection events | Yes | Hosting and CDN, AI providers where the page URL forms part of a prompt | No |
| Contents of chat messages and support requests | Yes | AI providers, and the helpdesk or webhook destination a customer configured | No |
| Geolocation: country-level, derived from IP by our CDN; not otherwise stored | CDN only | CDN provider | No |
| Sensitive personal information: login credentials, 2FA secrets, API keys, connected-service credentials | Yes | No third party; used only to authenticate and secure the account | No |
| Professional or employment: organisation name, role, job title where supplied | Yes, optional | Email relay and support | No |
| Biometric, precise geolocation, government IDs, health, racial or ethnic origin, religious or political beliefs, union membership, audio or visual recordings, profiling inferences | No | Not applicable | No |
Minors. We do not knowingly sell or share the personal information of consumers under 16. We respond to verifiable requests within 45 days, extendable once by a further 45 days where the request is complex, with notice.
Your rights under India's Digital Personal Data Protection Act, 2023
If you are a Data Principal in India, you may request a summary of the personal data we process; request correction, completion, updating or erasure; nominate another person to exercise your rights in the event of your death or incapacity; and access our grievance redressal process.
Grievance Officer
Pratik Chaskar, Chief Technology Officer, Brainstorm Force, [email protected]. This grievance address is the Brainstorm Force group grievance channel, shared across Brainstorm Force products. It is the correct contact for SureReplies even though our other contact addresses use the surereplies.com domain.
We acknowledge grievances within 72 hours of receipt, aim to resolve them within 30 days, and in all cases within 90 days. If your grievance is not resolved to your satisfaction, you may escalate it to the Data Protection Board of India.
Children and minors
SureReplies is intended for business use and is not directed to children. We do not knowingly collect personal data from anyone below the age of digital consent in their country (16 in much of the EU and EEA, and 13 under the US Children's Online Privacy Protection Act). Customers who deploy the chat widget are responsible for not deploying it on a service directed to children without their own lawful basis. If you believe a child has provided us with personal data, contact [email protected] and we will delete it promptly.
Third-party links
Our website, our documentation and the answers our bots produce may include links to third-party services, including links drawn from a customer's own documentation. We are not responsible for their content or their privacy practices. We encourage you to read the privacy policy of any service you visit from ours.
Changes to this policy
We update this policy when our practices change. The Last updated date at the top always reflects the current version, and we will give notice of material changes as the law requires. Our cookie table and our sub-processor table are reviewed whenever a technology or provider is added or removed. A change to the list of AI providers is always treated as material, and customers acting as controllers will be told before a new AI sub-processor begins processing their data.
Contact us
For questions, concerns or privacy requests, contact us at the addresses below.
Mailing address: Brainstorm Force US LLC, 2093 Philadelphia Pike #3090, Claymont, Delaware 19703, United States.