Privacy policy¶
WitFoo Limited (New Zealand company number 9413280)
Level 2, 47 Salisbury Street, Christchurch Central City, Christchurch 8013, New Zealand
Effective: 17 November 2026 · Version: 1.0
1. Who we are and what this policy covers¶
WitFoo Limited ("WitFoo", "we", "us") provides WitFoo security operations software and the online services that support it. WitFoo Limited is a subsidiary of WitFoo, Inc., a Delaware company.
This policy covers personal information connected with:
- the WitFoo appliance software: WitFoo Conductor, WitFoo Reporter, WitFoo Analytics, WitFoo Console, and the WitFoo appliance agent;
- WitFoo's online services: the licensing service at licensing.witfoo.com and the intelligence service at intel.witfoo.com;
- the WitFoo connector that lets an AI assistant query a customer's own appliance, including its relay at mcp-relay.witfoo.com and its sign-in page at connect.witfoo.com;
- support we provide to customers and partners;
- our website (witfoo.com), our marketing, and our sales and partner dealings.
This policy covers all WitFoo systems that hold personal information, including our marketing and sales systems. We are bound by the New Zealand Privacy Act 2020.
2. Your security data stays on your appliance¶
WitFoo software runs on an appliance the customer deploys and controls. The security data the customer's systems send to it, and everything the appliance derives from it, is held on the customer's infrastructure. WitFoo receives it only in the ways described in section 4, and the customer can switch each of them off or block it.
Security logs often contain personal information, such as user names, email addresses, host names and network addresses. The customer decides what its systems collect and send to the appliance, and is responsible for its obligations to the people that data is about.
3. Personal information WitFoo collects¶
| What | Whose | Why | Source |
|---|---|---|---|
| Licence records: licence key, organisation identifier, company name, a contact name and email, product and tier, CPU-core count, expiry and internal notes | The customer organisation and its named contact | To issue, verify and renew licences | From you or your reseller |
| Licensing-service audit records: a log of changes to licence records | WitFoo staff and customer contacts who make changes | To secure the service and investigate problems | Generated by the service |
| Free-licence offer claims: organisation name and an identifier derived from it, email address, declared country, network address and the country it maps to, the record of the agreement accepted (which agreement, its version and when), the licence identifier and the first eight characters of the licence key, the claim's status and dates, the reason an automated check held the claim for staff review (if it did), and staff decisions on it (who decided, when, and a note) | The person who claims a free licence, and their organisation | To issue the free licence, to allow one claim per organisation and email address, and to record acceptance of the agreement | From you, when you claim a free licence |
| Appliance registration: appliance identifier, configured IP address, host name (the operating-system host name), role, agent version and the time of the last check-in, sent about every 60 seconds when the appliance checks for work | The customer organisation (a host name can identify a person) | To deliver support and diagnostic jobs, and to run the fleet | Generated by the appliance |
| Support requests: the content of tickets and attachments | The person raising the request, and anyone named in it | To resolve the request | From you |
| AI-connector relay records (only if the relay is switched on; section 5): a registration record (organisation and appliance identifiers, and the name the appliance gave as the person who switched the relay on), and connection logs, including requests to the sign-in page at connect.witfoo.com (which record the appliance code a person enters only when it names an appliance whose relay is switched on) | The customer organisation, the person named, and users of the connector (their network address passes through the relay and is kept in memory only, as a rate-limit key; it is not recorded) | To operate and secure the relay | Generated by the relay, and supplied by the appliance at registration |
| Training records (Absorb; section 7): name, email address, username, department, job title, enrolments, completions, certificates, scores and attempts, the time of last sign-in, and either a hashed password or a single-sign-on identifier; also phone, address, manager and employee number where they are filled in | Customer and partner staff who take WitFoo training | To run the training, record results and issue certificates | From you or your employer |
WitFoo staff sign in to the licensing service with their WitFoo Microsoft accounts.
Website, marketing and sales. We also collect personal information when you visit witfoo.com, contact us, subscribe to our communications, attend an event, or deal with our sales and partner teams: your name, business email address, company, role, phone number and country, what you tell us, records of our communications with you, and technical data about your visit, such as your network address, browser and the pages you view. The website, www.witfoo.com, is hosted by Netlify (United States). When you send us a form on the website (the contact form, or the investor contact form, which also asks for your organisation), Netlify receives the submission, which holds your name, email address, the kind of enquiry and your message, and keeps it until it is deleted; Netlify also sends a notification email to WitFoo staff, through Microsoft 365. The sign-up form for our investor newsletter sends your details straight to Mailchimp (United States), and nothing passes through our own servers. Our sales and partner teams keep a record of their dealings with you in Salesforce (United States, US East 1). Our staff enter their notes by hand, and Salesforce's Einstein Activity Capture also copies in the emails and calendar events between our staff and you; information created by customers' use of our products is not sent to Salesforce. We send marketing emails through Mailchimp (United States), which tells us whether you open an email and which links you click. We use this information to respond to you, to run our sales and marketing, and to keep the website secure and working.
Cookies and analytics. The website shows a cookie banner with the buttons "Cookie settings", "No" and "Yes". It sets a cookie of its own that records your choices (necessary, tracking, analytics and marketing), and it uses Google Analytics, through Google Tag Manager, to understand how the site is used. In 32 countries (the 27 member states of the European Union, Iceland, Liechtenstein, Norway, the United Kingdom and Switzerland), analytics is off until you press "Yes". Everywhere else, including New Zealand, the United States and Australia, analytics starts when the page first loads, before you choose, unless your browser sends a Do Not Track or Global Privacy Control signal. Pressing "No" turns analytics off, deletes the Google Analytics cookies and records your choice, and the site applies your choice on your later visits before analytics starts. The website also embeds a video player from Vimeo that loads with the page, before you press play, so Vimeo receives your network address whether or not you play the video (the player is requested with Vimeo's do-not-track option set).
4. What leaves a customer's network¶
An appliance can send the following to WitFoo, or to the services named below. Most categories go through intel.witfoo.com, authenticated with the customer's licence key. An appliance with no licence key sends nothing in categories 1 to 3. The exceptions are stated below the table.
| # | Category | What it contains | Default |
|---|---|---|---|
| 1 | Operational metrics | CPU, disk and memory; container and service health; throughput; runtime counters. No log content. | Sent only if the customer chooses it when the appliance is set up; the same question as categories 2 and 3. |
| 2 | Operational logs and status | Lines from the appliance's own status journal, including informational, warning and error lines, sent without deduplication. A status report: its creation time; the organisation and appliance identifiers; the appliance's host name, role, software (WFA) version and operating-system version; the status of its services and containers; warning and error log lines; metrics; whether a reboot is required; pending and maintenance package updates, with their count and the auto-update setting (the list also keeps fresh installs and packages from any source other than Ubuntu or WitFoo, so it can include package names from a customer's own repositories); FIPS mode; whether staff-requested jobs are switched off; whether credentials are awaiting rotation; whether the helper version differs; and each of the appliance's Beats agents, with its host name. | Sent only if the customer chooses it when the appliance is set up; one question covers categories 1 to 3. |
| 3 | Unrecognised log lines | Log lines that a Conductor, Reporter or All-in-One Conductor node could not recognise, used to build new parsers. Each line is sent separately and without deduplication, with the organisation and appliance identifiers, the source entry point, and the host name or address of the device that sent it. The host name or address is not redacted. The category most likely to contain personal information from the customer's security data. | Sent only if the customer chooses it when the appliance is set up. Installs without a set-up question (cloud images, scripted installs) send nothing unless it is explicitly enabled. |
| 4 | Automatic problem reports | Appliance and service identifiers, severity, error counts and a fixed vocabulary of error types. No raw log text. A report can be copied into an internal issue in our GitHub account. | On |
| 5 | Support and diagnostic jobs | Support packs, log collection and diagnostics, requested by WitFoo support staff or created and sent by the customer. A support pack can contain one day of the system journal, recent container logs, system log files, the appliance's user-account and group lists, login records, network name-resolution files, and database, pipeline, connector and messaging diagnostics. These can include customer telemetry, user names and network addresses. The least redacted category. | Run when WitFoo support creates a job, or when the customer sends a pack. Customers can opt out of staff-requested jobs on the appliance, which stops every job WitFoo staff request, including reboot and reset jobs. |
| 6 | Threat-intelligence enrichment and sharing | Every external (non-internal) network address the appliance enriches, not only addresses an analyst looks up. Indicator sharing only where the customer enables it. Lookups, and directory searches (category 10), are recorded in the intelligence service's access log: the appliance's network address, the licence key with its middle masked (the first 4 and last 4 characters are kept), the request (which includes the address looked up, or the search terms and filters), the software identifier the appliance sends, the result and the time. Addresses not already cached are looked up with MaxMind's web service for geolocation: WitFoo's intelligence service sends MaxMind the address, one lookup for each address not already in its cache, and keeps the answer (country, city, latitude and longitude, and the network that owns the address) for 7 days. Website visitors' addresses are not sent to MaxMind. | Enrichment on, sharing off |
| 7 | Licence verification | The licence key, to licensing.witfoo.com. | On |
| 8 | AI-connector relay | See section 5. | Off |
| 9 | Reverse-DNS lookups | External network addresses seen in customer telemetry, looked up from the appliance directly with Google Public DNS (8.8.8.8), unless the customer sets another DNS server. This does not pass through WitFoo. | On |
| 10 | CyberGrid directory searches | The search terms and filters a user enters, or an AI assistant sends, when browsing the CyberGrid directory of published feeds. | Sent when a user searches the directory |
Appliances set up earlier. Appliances set up under an earlier version, which sent operational metrics and status logs to WitFoo by default, keep that setting after an upgrade. On Conductor, Reporter and All-in-One Conductor nodes the same setting also sends unrecognised log lines (category 3), although the prompt those appliances showed speaks of performance metrics. The release notes and the product tell customers how to turn it off.
Where it goes. Categories 1 to 3 go to our monitoring provider, Grafana Labs (Grafana Cloud). It receives and stores the data on Amazon Web Services in the United States (US East, Ohio). It keeps the log lines for 30 days and the metrics for 13 months: the log lines are those in categories 2 and 3, and the metrics are those in category 1 and in the status report. Normally the appliance sends these through intel.witfoo.com. Appliance agents older than version 2.4.5, and appliances an operator has set to send directly, send them straight to Grafana Cloud. The appliance's diagnostic command also contacts Grafana Cloud directly, so Grafana sees the appliance's outbound network address. Package downloads come from GitHub, and container images from intel.witfoo.com; both see the downloading appliance's network address.
The off switch. The customer's network policy controls all of this. To stop categories 1 to 6 and 10, block outbound connections to intel.witfoo.com and *.grafana.net. Licence verification uses licensing.witfoo.com. To stop category 9, set the appliance to use your own DNS server. The platform keeps running, with reduced enrichment.
Redaction. Some data passes through a redaction filter before it leaves the appliance: the free text of categories 2 and 3 (journal lines, the status report's warning and error lines, and the raw text of unrecognised log lines), and the host-name label on metric and log streams, which the same setting hashes (standard mode) or drops (strict mode). It does not apply to the identifiers sent with each record: the organisation and appliance identifiers and, for unrecognised log lines, the host name or address of the sending device. The appliance's own host name still travels in clear in every status report and as a label on each metric inside it, so strict mode does not remove it. In its default ("standard") mode the filter replaces detected values with consistent tokens, so problems can still be diagnosed. It covers IPv4 and IPv6 addresses, email addresses, URLs, file paths under home directories, payment card numbers, US Social Security numbers, and credentials such as access keys, tokens and private keys. It does not redact host names (apart from the stream label described above), domain names outside URLs, user names (unless in an email address or home-directory path), personal names, MAC addresses or serial numbers. Standard mode is pseudonymisation: it reduces exposure, but values that are easy to guess can be recovered, so it is not anonymisation. A stricter mode replaces detected values with fixed labels, and also masks phone numbers and bank account numbers (IBANs). Appliance agents 2.2.7 to 2.2.9 send unrecognised log lines without redaction; update to a later agent. Redaction is pattern-based and best-effort, and it is not a data-loss-prevention control. A customer who needs to prevent disclosure should block the traffic.
5. The AI-assistant connector¶
From WitFoo release 1.2.0, a customer's administrator can let users connect an AI assistant to the customer's appliance. Each user signs in as themselves, and the connector acts with that user's WitFoo permissions.
- Direct connection. Wherever the assistant can reach the appliance and the user connects it to the appliance's own address, it connects directly, and nothing passes through WitFoo. A connection made through WitFoo's single connector address (below), which is the address any WitFoo listing in an AI assistant's app directory uses, always passes through the relay.
- Optional relay (off by default). Where the assistant cannot reach the appliance, or to let users connect through WitFoo's single connector address, the administrator can switch on a relay, after acknowledging a disclosure. The appliance opens an outbound connection to WitFoo's relay service at mcp-relay.witfoo.com, which is separate from licence verification, so a customer can block it on its own. The relay is off by default for every organisation and every appliance, and switching it on is the customer's own act on the appliance.
- Signing in through WitFoo's single connector address. From WitFoo Analytics 1.2.1, with appliance agent 2.5.5 or later, an administrator who has switched the relay on can let users connect an AI assistant at one address for every customer, https://mcp-relay.witfoo.com/mcp. When a user connects an assistant there, the assistant opens WitFoo's sign-in page, https://connect.witfoo.com, in the user's browser.
- What the page receives. It receives the assistant's sign-in request: the assistant's web address, the address to return to, a one-time value, and a check value. It also receives the appliance code the user types, which is the code in the appliance's relay address.
- What the relay asks the appliance. The relay asks that appliance, through the relay, for the address of its own sign-in page. It does not pass on the user's network address with this question. It then shows a page that names that address, with a link to it.
- No redirect, no password. The page does not redirect by itself. The user signs in on the appliance's own page, which does not pass through WitFoo, so WitFoo never sees the user name or password entered there.
- Nothing stored about the sign-in. The appliance writes its relay code into the sign-in codes it issues, so the relay can pass the assistant's next request to that appliance and to no other. The relay adds the code and a signature to the front of the tokens the appliance issues, so that later requests reach the appliance. It holds them in memory only while it passes them on.
- Operator and hosting. The page is part of WitFoo's relay service, runs in the same software on the same server, and is served through Cloudflare like the relay's address.
- No script, no cookie. The page runs no script, sets no cookie of its own, and its pages are not cached. Cloudflare's features that rewrite pages are switched off for it, and when we checked it through Cloudflare, Cloudflare added no cookie. If Cloudflare's protection against attacks decides to check a visitor's browser, Cloudflare shows its own check page first, which runs Cloudflare's script and can set Cloudflare's cookie.
- What the relay can read. While the relay is on, the AI assistant's requests and the appliance's answers pass through WitFoo's relay service. The relay reads each one into memory to pass it on, so the sign-in tokens, personal keys and content in it can be read by the relay's software while they pass through. This applies to requests sent to the relay's address, including the sign-in requests and appliance codes sent to its sign-in page at connect.witfoo.com; an AI assistant that connects to the appliance's own address does not pass through the relay. The appliance's own calls to the relay (to register, connect, acknowledge, report its status and revoke) also reach the relay readable, and carry the licence key or the relay credential.
- Cloudflare. The relay's address and its sign-in page at connect.witfoo.com are served through Cloudflare, which acts as a proxy in front of WitFoo's relay. Cloudflare ends the encrypted connection at its network edge and passes the traffic on to WitFoo, so Cloudflare can also see what passes through: requests, answers, access and refresh tokens, personal keys, the licence key when the appliance registers, reports its status or revokes, and the relay credential when it connects and acknowledges. The registration and tunnel connections use the same address, so they pass through Cloudflare too. Cloudflare also sees the details of each connection: the network addresses of the caller and of the appliance, the time, the sizes, and the address's handle. On the sign-in page Cloudflare also sees the assistant's sign-in request (including the address to return to), the appliance code the user types, and the page that names the appliance's sign-in address. The sign-in page has Cloudflare's ordinary protection against attacks, so Cloudflare's security records are more likely to include requests to it, with the caller's network address, the requested address including its query, the browser's description of itself, and the country. Cloudflare keeps these security records, for the relay's address and for the sign-in page, for up to 31 days. Cloudflare is set not to cache anything served from the relay's address.
- What the relay does not keep. WitFoo's relay holds the content of requests and answers, and any token or key in them or sent with them, in memory only while it passes them on, and writes none of it to its database or logs. It keeps a one-way hash of the credential the appliance uses to connect for as long as the address exists. The sign-in page stores nothing about a sign-in: no sign-in request, token or network address is written to the relay's database or logs, and the appliance code a person types reaches the logs only as described below. To limit the questions it puts to an appliance, the relay keeps the address of the appliance's sign-in page in memory for up to a minute (ten seconds when the appliance's answer was refused); an expired entry can stay in memory until the relay next stores an entry or restarts, and it is never written to disk.
- What the relay's logs record. For each request, including each request to the sign-in page: a request identifier, the address's handle (on the sign-in page, only when the code entered names an appliance whose relay is switched on), the method, the path, the time, the status, the sizes, the duration and the outcome. They never record the sign-in request's details, the code typed when it is not a registered one, the appliance's sign-in address, or any token. The method is the text the caller sent, kept up to 256 bytes, so a caller that puts anything other than a method there puts it in the log. The logs also record when a tunnel comes up or goes down, with the code that says why it closed, when an address is created, acknowledged or revoked, and an alert when a registering licence looks like a trial, cloud or offer licence. Some lines carry a short one-way fingerprint of a licence key or of the relay credential: the first eight characters of its SHA-256 hash. Logs are kept for 30 days, and are gone about a day after that.
- What the relay keeps about the appliance. Until the address is revoked, the relay keeps a record for each appliance, up to four per organisation: the organisation and appliance identifiers, the name the appliance gave as the person who switched the relay on (a claim the relay cannot verify), the times, and a hash of the relay's credential. Anyone who holds the organisation's licence key can read that name. If the licence lapses, or the appliance is lost, the record stays until the address is revoked. The relay also keeps a second record of the organisation's entitlement (organisation, product, tier, expiry and the time it was last checked). It is rewritten about every 15 minutes while a tunnel is up. It expires 48 hours after it was last written and is removed from the database's storage 3 days after that write, plus the time until the database's next clean-up, so it can outlast a revocation. Event records that name the person who switched the relay on expire after 7 days and are removed from storage 8 days after they were written, plus the time until the database's next clean-up; the name kept in the first record above is not part of them, and stays until the address is revoked. The licensing service keeps nothing about the appliance for the relay: it answers the relay's entitlement question from the licence tables and writes nothing.
- The caller's network address. The network address the relay sees for the machine that connects to it is passed to the customer's appliance with each relayed request, so the customer's own logs may record it, and a support pack made from those logs can reach WitFoo. The relay does not write it to its logs, database or metrics. It keeps it in memory as a rate-limit key. On the sign-in page the relay uses the browser's network address only in memory, as a rate-limit key (the whole address for IPv4, the first half for IPv6), treats it as forgotten after an hour without requests (the entry can stay in memory until the relay next uses that limit or restarts, and is never written to disk), and does not pass it to the appliance.
- Where it runs, and switching it off. The relay runs on Oracle Cloud Infrastructure in Chicago. Cloudflare handles the traffic at its own network edge, which may be outside that region. When the administrator switches the relay off and the relay can be reached, the address is revoked and the relay clears its record linking the address to the appliance from its tables. What remains: a small record that the address existed and was revoked (the address's random identifier and the time it was revoked, kept permanently); event records and log lines already written, which expire on their own schedule; and the entitlement record, which expires 48 hours after it was last written and is removed from storage 3 days after that write, plus the time until the database's next clean-up. A revocation is final once the relay has confirmed it. Switching the relay on again gives a new address. If the appliance can no longer revoke the address, the customer can ask WitFoo support (support@witfoo.com) to revoke it. Support checks the request through account verification before an operator revokes the address.
- What a connection can carry. The assistant's requests and the appliance's answers come from the customer's own WitFoo data, limited to what the connected user's permissions allow and what the user asks for.
- From the customer's security records, an answer can contain personal information about the people those records concern: user and account names, email addresses, IP and other network addresses, host names, and the text of log records.
- About the customer's own WitFoo users, an answer can contain their names, user names and email addresses (for example when a user is looked up or a work unit is assigned), the notes and chat messages they wrote, and records of who changed what.
- In a request, the same kinds of information can travel the other way: a user name or address the user asks about, and, when the user has allowed the assistant to make changes, the text of a note or message it writes.
- Where it goes. WitFoo does not collect this information through the connector. It is read from, and written to, the customer's appliance at the user's request. It passes through WitFoo's relay and Cloudflare only when the connection uses the relay, as described above, and goes to the AI assistant the user connected.
- Two exceptions. A search of the CyberGrid directory is sent from the appliance to WitFoo's intelligence service at intel.witfoo.com. A message sent in a chat room shared with another organisation is sent by the customer's appliance directly to that organisation's appliance, at the address its CyberGrid subscription records.
- The AI provider. What the connector retrieves goes to the AI assistant the user connected, which handles it, and keeps it for as long as its own terms say, under those terms. WitFoo does not keep it. The customer controls the connection:
- Who can connect. Connecting is off until an administrator turns it on, and only users given the permission to connect can use it.
- Which assistants. The customer chooses which assistants to allow.
- Read-only by default. An assistant can only read unless the organisation allows changes and the user also ticks "Also let this app make changes for you".
- Revoking. A user or an administrator can revoke an assistant's access at any time.
- The relay. Switching the relay off ends every connection made through it.
6. How we use personal information¶
We use personal information to issue and verify licences and trials; to provide, maintain, secure and improve our software and services; to diagnose and fix problems and answer support requests; to administer customer and partner relationships; and to comply with the law.
We do not train AI models on customer data. We use an AI service, Anthropic's Claude API, for automated analysis:
- up to 10 unrecognised log lines (category 3) for each new parser, with the organisation and appliance identifiers, the entry point and the address or host name of the sending device, to draft the parser, and up to 6 more lines each time the parser is refined;
- short samples of operational logs (category 2), to suggest fixes to faults;
- the content of internal issues, including copied problem reports (category 4), when an investigation is run.
In WitFoo's Anthropic account, data sharing and export are switched off. Anthropic deletes what we send within 30 days. Under its usage policy, Anthropic may keep content it flags for a policy violation for up to 2 years.
Samples in our software. Samples from customers' unrecognised log lines (category 3) can be kept in our software and in our development systems, and nothing deletes them automatically today. When our automation generates a parser, it can keep 3 to 5 sample lines in the parser's test code, and merged parsers are compiled into the parser software that runs on every appliance. When it cannot generate a parser, it opens an internal issue that can hold up to 3 decoded samples and the customer's organisation identifier. We have decided to remove these samples and to stop it happening again, and this paragraph changes when that is done.
We also run security monitoring agents on the servers that host our licensing and intelligence services, to protect those services. They report to WitFoo's own security operations team. The data they collect can include request details, such as network addresses, and log lines from those servers. We keep it for 1 year (section 8).
We do not sell personal information.
7. Who we share it with¶
- Our related companies, WitFoo, Inc. and WitFoo Federal, Inc. (United States), where they help provide the services.
- Service providers acting for us:
- Oracle Cloud Infrastructure, which hosts the servers that run licensing.witfoo.com and intel.witfoo.com and, if the relay is switched on, the AI-connector relay and its sign-in page (section 5), in its US Midwest (Chicago) region;
- Netlify (United States), which hosts the www.witfoo.com website and receives and stores the contact and investor form submissions sent through it (section 3);
- Grafana Labs (monitoring; section 4);
- Cloudflare (network security and delivery for the relay's addresses, mcp-relay.witfoo.com and its sign-in page connect.witfoo.com, and for other WitFoo addresses: the public documentation at docs.witfoo.com, where this policy is posted, the public demonstration, the download address for the public appliance images, WitFoo's investor due-diligence portal and WitFoo's own appliances; it sees relayed AI-connector traffic and sign-ins through the single connector address in transit; section 5);
- Anthropic (automated analysis; section 6);
- GitHub (source code, the automation that processes diagnostic data, internal records of parser failures, and package and image hosting; sections 6 and 8);
- MaxMind (network-address geolocation: our intelligence service sends it the public network addresses that appliances ask about; section 4);
- Microsoft (staff sign-in, and Microsoft 365, which is WitFoo's email, receives the notification emails for website form submissions and stores its data in the United States);
- Slack (internal notifications to WitFoo staff, which can include licence contact details, organisation identifiers, appliance host names and addresses, and support-pack notices);
- Zendesk (support tickets; United States);
- Absorb (our training platform, which holds learner accounts, enrolments and progress for customer staff who take WitFoo training) (United States, US East; learner records are kept for the life of the learner's account, and a learner can ask us at any time to delete their data);
- Salesforce (our record of our dealings with customers, partners and prospects, which our staff enter by hand and which also receives emails and calendar events between our staff and you through Einstein Activity Capture) (United States, US East 1);
- Mailchimp (marketing emails: the names and email addresses of people who subscribe to our communications, including those who sign up through the investor newsletter form on the website, which sends its details straight to Mailchimp, and whether they open an email or click a link) (United States);
- Google (website analytics, through Google Analytics and Google Tag Manager on www.witfoo.com; section 3);
- Vimeo (training videos, and a video player on www.witfoo.com that loads with the page: the browser sends Vimeo its network address and, when a video is played, playback data);
- The customer's reseller or distributor, for licensing and billing.
- The AI provider a user chooses (section 5), at the customer's direction.
- Authorities, where the law requires it.
Outside New Zealand. Most of these recipients are in the United States. Before disclosing personal information overseas, we make sure the recipient is required to protect it with safeguards comparable to the Privacy Act, for example by contract. A contractual safeguard is the basis.
8. How long we keep it¶
We keep diagnostic and operational data no longer than 7 days, apart from the exceptions below. That covers categories 4 to 6 (including support-pack files and support-job records), relay event records (7 days, and removed from the database's storage 8 days after they were written, plus the wait for its next clean-up; not the relay's registration record or its logs, see below), the intelligence service's access log and its cache of address lookups, and copies our staff make while troubleshooting. Categories 1 to 3 are kept by Grafana Labs for longer, and our development systems keep some diagnostic data for longer: the first two exceptions say how long.
Licence and account records (section 3: licence records, free-licence offer claims, the licensing audit file and account contacts) are kept for the life of the licence and for as long as WitFoo needs them afterwards to support the customer, meet its legal obligations and resolve disputes. WitFoo is putting in place automatic deletion of these records 6 months after a licence ends, and of licensing audit entries that are not tied to a licence 6 months after they are written. This page will say when that deletion is in effect.
Exceptions:
- Grafana Labs keeps the log lines in categories 2 and 3 for 30 days and the metrics in category 1 and in the status report for 13 months (section 4).
- Our development systems (GitHub). The automation that reads diagnostic data from our monitoring service every 6 hours keeps its decoded sample log lines as workflow files for 7 days and its other analysis files for 30 days, and keeps the records of its runs for 90 days. Sample log lines can also be kept without an expiry, as section 6 describes, and they stay in the software that carries them: images of a release, and images tagged main, stable, beta or latest, are kept for ever, and the other images of our software are deleted after 30 days, apart from the 3 most recent of each.
- Anthropic deletes what we send within 30 days (section 6).
- Slack notifications are deleted after 90 days.
- Security-monitoring data from the servers that host our services is kept for 1 year (section 6).
- Appliance registrations (section 3) are kept until an administrator deletes them; nothing deletes them automatically.
- Support tickets are kept for as long as the customer's account exists, and attachments are deleted with their ticket.
- Training-platform learner records (Absorb) are kept for the life of the learner's account. A learner can ask us at any time to delete their data, and we ask Absorb to delete it (a learner cannot delete it in Absorb's portal). Copies in Absorb's backup systems are kept until its ordinary deletion cycle.
- Marketing records are kept for 90 days from our last contact with the person; this covers website enquiries, including the copy that Netlify holds, and sales leads who have not become customers. People who subscribe to our emails stay on our list until they unsubscribe or ask us to remove them, and we then delete them.
- Website analytics data (Google Analytics) is kept for 2 months.
- Customer records are kept for as long as there is an active relationship with WitFoo, and for 1 year after it ends.
- The relay's registration record (organisation and appliance identifiers, and the name given when the relay was switched on) is kept until the address is revoked, then cleared, leaving a permanent stub (the address's random identifier and the time it was revoked). The licence-entitlement record the relay keeps expires 48 hours after it was last written and is removed from storage 3 days after that write, plus the time until the database's next clean-up. Event records expire after 7 days and are removed from storage 8 days after they were written, plus the same wait (section 5). The relay's logs, including those of requests to its sign-in page at connect.witfoo.com, are kept for 30 days and are gone about a day after that. The sign-in page writes nothing else to the relay's database or disk.
Backups and recovery. We do not schedule snapshots or backups of our Oracle Cloud storage, except for the AI-connector relay's database: it is backed up every night and within an hour of each change, encrypted before it leaves the relay's server, kept in Oracle Cloud Object Storage in Chicago and deleted after 35 days (section 5). A record deleted from Salesforce stays in its Recycle Bin, where it can be recovered, for 15 days.
9. How we protect it¶
- Appliance connections are authenticated with the customer's licence key.
- Connections appliances make over the internet to licensing.witfoo.com and intel.witfoo.com, and our service's connections to our monitoring provider, are encrypted with TLS 1.2 or higher.
- Our hosting provider encrypts our servers' storage volumes at rest.
- The relay's address, and WitFoo's other addresses that are served through Cloudflare, use TLS 1.2 or higher between the visitor and Cloudflare, and Cloudflare connects to WitFoo's server over an encrypted connection and checks its certificate.
- The sign-in page at connect.witfoo.com runs no script, sets no cookie of its own (Cloudflare adds none, except on the check page it shows a browser its protection decides to check; section 5), cannot be shown inside another site's page, is not cached, and accepts the appliance code only from its own form.
- Some data passes through a redaction filter before it leaves the appliance, with the limits stated in section 4.
If a privacy breach involving personal information we hold is likely to cause serious harm, we notify the Privacy Commissioner and the affected people as soon as practicable.
10. Your rights¶
You can ask whether we hold personal information about you, ask to see it, and ask us to correct it. Contact our privacy officer (section 11). We respond within 20 working days.
If the information is on a customer's appliance rather than held by WitFoo (section 2), the customer controls it, and we will refer you to them.
If you aren't satisfied with our response, you can complain to the Office of the Privacy Commissioner at www.privacy.org.nz.
11. Contact¶
Privacy officer: Charles Herring, Chairman
Email: legal@witfoo.com
Post: Privacy Officer, WitFoo Limited, Level 2, 47 Salisbury Street, Christchurch Central City, Christchurch 8013, New Zealand
12. Changes to this policy¶
We update this policy when our practices change, and show the effective date at the top. Where a change materially affects information we already hold, we tell customers before it takes effect.