Privacy Policy

1. Who we are

CommunityTracker is a software product operated by Miraa Labs LLC, at 1309 Coffee St, Sheridan, Wyoming 82801, United States. We refer to ourselves below as “we”, “us” or “CommunityTracker”.

Our product helps businesses find relevant public conversations across community platforms and, optionally, reach out to the people involved. That second part means we handle contact details of people who are not our customers, and — if you choose to connect a mailbox — we handle access to your email account. Both deserve plain explanation, and both get one below.

This policy covers our website at communitytracker.ai and our application at app.communitytracker.ai.

2. When we decide, and when you decide

Two different relationships run through this product, and your rights differ between them.

• Your account information is ours to look after. We decide what to collect and why — your name, email, billing details, and how you use the product. In data protection terms we are the controller, and this policy is our accounting to you.
• The prospect data you work with is yours to direct. When you monitor a keyword, build a contact list, or send outreach, you decide who is contacted and what is said. We act on your instructions. You are responsible for having a lawful basis to contact those people, and for honouring their objections. One exception is the shared copy behind lead monitoring (section 3.3). It serves every customer who monitors the same person, so we decide how it is kept and for how long.

Where we act on your instructions, we still apply our own limits — see section 8. We will not send messages to an address that has asked us to stop, whoever instructed it. Nor will we monitor a person who has asked us to stop, under any address we know is theirs. If they ask, we also delete what monitoring found about them, apart from what section 3.3 says can stay.

3. What we collect

3.1 Account and billing information

Your name, email address and password (stored only as a cryptographic hash — we cannot read it). If you subscribe, our payment processor collects your card details directly and we receive only the subscription status, the plan, and the last four digits. Card numbers never reach our servers.

3.2 Public posts you ask us to monitor

This is the core of the product. When you monitor a keyword, we collect posts and comments that are publicly visible on the platforms you select — currently Reddit, X, LinkedIn, Hacker News, Indie Hackers, Dev.to, Stack Overflow, GitHub and Slack communities. For each post we store its text, its public author handle, its link, its timestamp and its public engagement counts.

We collect this from public sources. We do not access private groups, direct messages, or anything behind a login that you have not connected yourself.

Where a LinkedIn author is based. For some customer accounts we also find out where the author of a highly rated LinkedIn post is based, so the customer can filter their posts by country. Apify collects the author’s public LinkedIn profile for us, with the HarvestAPI scraper described under “Leads you monitor” in section 3.3. From it we store only the location the profile shows and the country it names. We keep that answer so we do not look the same person up again for another customer. Apify keeps the profile it collected for about 31 days, then deletes it.

3.3 Business contact details and professional profiles we find for you

If you use the contact-finding features, we look up business contact details — typically a work email address, job title, employer and public professional profile — for people whose posts you have surfaced. We obtain these from third-party data providers listed in our sub-processor list, and from public web sources.

Where a person posted publicly on LinkedIn, that record includes the public address of their LinkedIn profile and the professional details shown on it, such as their name, headline, job title and employer. If our customer runs outreach on LinkedIn, we use that profile address to send them a connection request or a message on the customer’s behalf — see section 3.5 for how that works and section 8 if you are the person who received one.

These people did not sign up for our product. We treat that seriously: they can ask to be removed at any time using the address in section 15, and we keep a permanent suppression record so the removal actually holds.

Lead lists you upload. If you upload a list of people to contact, we store it in your account. Only you, your team and, when they need to, our support staff can see it (section 11). We send the details needed for each lookup you run to the enrichment provider named in our sub-processor list. You can delete a list at any time; leads you have already added to a campaign stay with that campaign until you delete it or your account.

Leads you monitor. You can turn on monitoring for leads in your lists. For each one, Apify collects the person’s public LinkedIn profile and recent posts for us, and recent posts from their company’s LinkedIn page. Apify does this with scrapers that HarvestAPI publishes. These scrapers send the addresses we ask about to HarvestAPI’s own service, which collects the data and sends it back. A company page’s posts can include posts it shared from other people. We treat those like the company’s own posts: we store and judge them, and your results can quote them. From the profile we store their headline, location, industry and current job, and the previous version when their job changes. We buy data about their company from Prospeo: its size, industry, country, funding and open job postings.

Apify keeps everything it finds in our Apify account for about 31 days, then deletes it. That includes parts of the profile we do not store ourselves, such as their name, About section, full work history, education and skills. It also includes details of other people that LinkedIn shows with a profile or a post. For example, a profile can come with the names, job titles and profile addresses of other members, and recommendations that others wrote.

An AI model from TypeSafe compares the person’s job and company with the ideal customer profile in your account. It also labels each post, for example whether it asks for a recommendation, complains about a tool or describes a problem you solve. You see these labels on the lead, but they do not change its rank. For each judgment the model receives your ideal customer profile and the person’s job title and company. It also receives either their headline and industry or the text of one post. Each request may only go to an AI provider that keeps none of it. We use an answer only if TypeSafe itself gave it (section 7). Our own code then ranks the lead. It uses the model’s answer on the job, the size of their company, and facts such as a new job, recent funding or open roles. You also see the model’s answer on the company, but it does not change the rank.

While monitoring is on, we check again about every:

• 3 days for their posts;
• 7 days for their company’s posts;
• 14 days for their profile;
• 30 days for their company data.

We keep one shared copy of this data for all customers who monitor the same person or company. We delete it on this schedule (section 9 lists how long we keep each part):

• posts, 120 days after we last collected them;
• profiles and company data, 180 days after any customer last used them;
• the AI model’s answers, 180 days after it gave them.

Only you, your team and, when they need to, our support staff can see your results (section 11). Your results stay with the lead and are deleted with it. They include short quotes from the posts, and the details of the lead that monitoring fills in or updates, such as its LinkedIn profile address. If the person asks us to delete what we hold about them, we delete your results about them too (see below). That includes any thumbs up or down your team gave the lead, and any note with it.

Monitoring pauses when your account can no longer use it, for example when the beta ends or your plan lapses. It starts again if your account can use it again. To stop it, turn it off. The results you already have stay with the lead. A lead whose email address or LinkedIn profile is on your do-not-contact list cannot be monitored.

If a customer monitors you. To make it stop, email us at the address in section 15. Include the address of your LinkedIn profile, and any older address of it that you know. Once we have recorded your request, no customer can monitor you under those addresses. That includes the common spellings of each address, such as one with its accented letters encoded. A customer’s list may give another address of your profile, such as an older one, that we have not linked to yours. Monitoring under that address stops as soon as a check of that profile links the two.

You can also ask us to delete what we hold about you (section 10). We then delete it under every address of yours that we know of at that time. That covers our shared copy of your profile and posts, and the AI model’s answers about them. In each customer’s account, it covers what monitoring added to their lead: quotes, labels, facts and the details it filled in. It also covers our records of checking you and of who monitored you. The customer keeps the lead they uploaded, including the address of your LinkedIn profile. If monitoring changed that address, we put back the one they uploaded. Where we cannot, the changed address stays on the lead. If the customer added the lead to a campaign, the campaign keeps the details it copied from the lead. That includes any that monitoring filled in or updated. We keep a one-way hash of each address, and any link between them, so that your request keeps working. We also keep the email address you wrote from, and a record of your request (section 8). Apify deletes its copies within about 31 days. HarvestAPI’s logs of our requests can hold your address. HarvestAPI says it keeps them for 30 days, and we cannot delete them.

Two things can stay in customers’ results until those customers delete the leads. First, a company page may share one of your posts. We may hold it as the company’s post, and then neither request reaches it. The AI model may still judge it, and customers who monitor people at that company can see a link to it and quotes from it. Our shared copy of it is deleted on the schedule above.

Second, a customer’s list may hold an address of your profile that we had not linked to yours at that time. What monitoring found under it can stay with that customer’s lead until they delete it. If you ask us again later, we delete it too, if a check has linked that address by then.

3.4 A mailbox you connect

Optional, and off unless you turn it on. If you connect a Gmail or Outlook mailbox so that outreach sends as you, we receive access to that mailbox. Replies to the conversations our product started then appear in your CommunityTracker Inbox, where you can read them, see a label, and answer. Because Google account data carries specific obligations, it has its own section — see section 5.

3.5 A LinkedIn account you connect

Optional, and off unless you turn it on. If you connect your LinkedIn account so that outreach goes out as you, we hold an authorisation that lets us act as you on LinkedIn. We use it to send the connection requests and messages your campaign defines, to notice what comes back, to show you those replies in your CommunityTracker Inbox and label them (see section 7), and to send the replies you write there.

We store the messages we send, whether an invitation was accepted, whether the person replied, and the public profile details of the people you contact. We also store your own display name, profile address and connection count, which we use to show you the account is healthy and to pace sending sensibly. Of each reply we keep a short preview — at most 140 characters of what the person newly wrote — so your Inbox can list it, and the label it was given. We do not keep the full text: when you open a conversation we load it from LinkedIn, show it to you, and discard it. The one exception is the same as for email: when a reply may be asking us to stop, but we cannot be sure, we keep its first 500 characters so a person can review it (see section 5.2).

We do not read your LinkedIn inbox. We see replies to conversations our own product started, because that is how we know to stop a sequence, and we show you those conversations. Messages from before our first message in a conversation are never loaded. We do not collect your other conversations, your connection list, your feed, or anything you were doing on LinkedIn before you connected the account.

This is separate from anything Google. A connected LinkedIn account involves no Google service and no Google account data, and nothing in section 5 is affected by it.

You can disconnect at any time from the application’s settings, or change your LinkedIn password, which ends our access immediately. Note that automated messaging is against LinkedIn’s own User Agreement, and the account you connect is yours — we apply conservative sending limits and randomised timing to reduce the risk, but we cannot eliminate it, and we tell you this before you connect rather than afterwards.

3.6 Other services you connect

If you connect Slack, we receive the messages in the channels you select, so we can surface relevant conversations. If you use our API or MCP integration, we log the queries made against your account.

3.7 Usage and diagnostic information

When you use the application we automatically collect your IP address, browser type, the pages you visit and the actions you take, along with error diagnostics when something breaks.

We use Microsoft Clarity, which records anonymised session replays — a reconstruction of how you moved through the interface — so we can find where the product confuses people. Message content, recipient names and email addresses are masked before leaving your browser, so they do not appear in a replay. We also use ipapi.co to look up the approximate country of your IP address, purely so that prices display in a sensible currency.

4. How we use information, and on what basis

Where we rely on legitimate interest, you can object — see section 10, and we will stop unless we have a compelling reason not to.

5. Google user data

This section applies only if you connect a Google account. It describes exactly what access we ask for, what we do with it, what we keep, and who else can see it. If you never connect a mailbox, none of it applies to you.

5.1 What we ask for, and why

Connecting a Google account grants us two capabilities:

• Permission to send email as you. This is the point of the feature: your outreach arrives from your own address, in your own name, and replies come back to your own inbox where you can continue the conversation normally.
• Read access to your mail. We use it for four purposes, all about the conversations our product started: to notice when someone replies, to notice when a message bounces, to show you those replies in your CommunityTracker Inbox, and to label them (section 5.4). Other mail is never shown, labelled or stored.

Read access deserves an explanation, because it sounds broader than what we do with it. Detecting a reply reliably means reading the reply, not just its headers. An automatic out-of-office is not a real reply and must not stop your sequence; a person writing “please take me off your list” is an opt-out and must stop it immediately and permanently. Neither can be distinguished from message headers alone. Bounces are the same problem: the difference between a mailbox that is full today and an address that does not exist lives in the text of the delivery notice, and treating one as the other means either discarding a valid contact or repeatedly mailing a dead address.

One thing on the permission screen is worth explaining before you meet it. The connection is made through Unipile, the provider that connects your mailbox for us (see section 5.3). So Google’s permission screen shows Unipile’s name and logo, not ours. That is expected.

The permissions requested are exactly the two described above — sending email as you, and reading your mail — and nothing else. Sending also covers the replies you write in your Inbox; they go out in the same thread, from your own address. You can see and withdraw whatever you have granted, at any time, at myaccount.google.com/permissions.

5.2 What we actually store

Less than you might expect, and this is the part worth reading carefully.

What we do keep from a connected mailbox:

• The mailbox address and display name, so you can see which account is connected and we can show it in the interface.
• Messages you sent through us — the subject and body of your own outreach, which we composed and you approved. This is your record of what was said.
• Message and thread identifiers, which let us keep a follow-up in the same conversation rather than starting a new one, and let us match a reply to the sequence it belongs to.
• A short preview of each reply — at most 140 characters of what the person newly wrote (text quoted from earlier messages is removed first), so your Inbox can list the conversation. It is never longer, and it is deleted with the rest.
• The label of each reply — for example interested, out of office, or a request to stop — and whether a person on your team or our AI model set it. The conclusion, not the text it came from. That a message bounced is kept the same way.
• The replies you write in CommunityTracker — your own text, who on your team wrote it, and when it was sent.
• Unclear requests to stop — when a reply may be asking us to stop contacting someone, but the wording leaves room for doubt, we keep the first 500 characters of what the person newly wrote (text quoted from earlier messages is removed first). A person on your team then reviews it, so the request is honoured rather than missed. A clear request to stop is acted on at once, and its text is not kept.

All of it is deleted 180 days after a sequence finishes, or after the last message in the conversation if that is later, by an automated job that runs every night. Disconnecting a mailbox deletes its conversations, previews and labels at once. Retention is covered in full in section 9.

5.3 Who else can see it

A short list, named here rather than linked away, because you should not have to click to find out who can read your mailbox data:

• Supabase — Database, authentication and server-side functions. Hosted on AWS us-east-2 (Ohio, United States).
• Unipile — Brokers the connection to your Gmail or Outlook mailbox and holds the authorisation token. Sends the messages you approve and relays reply notifications back to us.
• OpenRouter — Routes each request to a model provider and enforces our requirement that the provider does not retain it. Which provider serves a given request is decided at the time of the request. For reply labels in your Inbox, it receives the text of a reply, the last message you sent that person and the start of the public post you found them through, each with addresses, phone numbers and links removed; it keeps nothing and does not train on it.
• Amazon Web Services, Google Cloud and Microsoft Azure — Operate the zero-retention endpoints that actually serve those models to us, including the one that labels the replies to your campaigns. They keep nothing and do not train on it.

That is the complete list. No one else receives data obtained from a connected Google account, and we do not sell it, rent it, or use it for advertising.

5.4 Limited Use, and no AI training

One consequence of that policy is worth stating on its own, because it is a question people increasingly ask and the answer is unambiguous:

Labelling is a feature you see: the label sits on each conversation in your Inbox, and you can change it. A label never unsubscribes anyone by itself — if the model thinks a reply asks us to stop, a person on your team confirms it first. Our other AI features draft outreach from the prospect’s public post, the settings you configured and, for a follow-up, our own copy of the campaign’s earlier messages to that person; nothing from your mailbox goes into them. Your mailbox address and your recipients’ addresses are never sent to a model provider.

5.5 Taking access away

There are two different things you might want here, and it matters which one you use, so we would rather be precise than reassuring.

1. Stop us using the mailbox. In the application, open Settings and choose Disconnect. Sending stops immediately, any sequence relying on that mailbox is paused, and we stop reading it. To be clear about what this does not do: on its own it does not withdraw the permission you granted at Google. That authorisation continues to exist until it is removed by one of the routes below.
2. Withdraw the permission itself. In your Google account, visit myaccount.google.com/permissions and remove our access. This revokes the authorisation at source, takes effect at once, and works even if you can no longer sign in to our product. It is the definitive route, and we would rather send you there than let you believe our own button is equivalent.

Deleting your account also withdraws the permission, and does so without you having to visit Google: we hand the authorisation back at the moment you request deletion, without waiting for the 30-day window to pass. If that call fails we retry it every hour and say so on the screen.

Whichever route you take, data we already hold is deleted on the schedule in section 9, or under section 10 if you asked us to delete the whole account — which takes effect after the 30-day window described there, not instantly.

If you delete your account, we hand the authorisation back to Google at the moment you request it, without waiting for the 30 days to pass. If that call fails we retry it every hour and tell you so on the screen; you can always remove it yourself at myaccount.google.com/permissions rather than wait for us.

6. Who we share information with

We share information with the service providers that make the product work — hosting, payments, email delivery, data sources and support. Each one receives only what it needs for its specific job.

We publish the complete list, including what each provider does and whether it can see data from a connected Google account, at communitytracker.ai/sub-processors.

Beyond those providers, we share information only when:

• you ask us to, or connect an integration that requires it;
• the law requires it — a valid court order or legal process. We will tell you unless we are legally prohibited from doing so;
• we need to establish or defend a legal claim, or to protect someone’s safety;
• the business is sold or merged, in which case the acquirer is bound by this policy until you are told otherwise and given a chance to object.

7. How AI is used in the product

We use large language models, supplied by the providers named in our sub-processor list. Their jobs include scoring whether a public post is relevant to what you monitor, drafting outreach copy, labelling the replies to your campaigns in your Inbox, suggesting how the columns of a lead list you upload match our fields, and judging the leads you monitor.

For scoring and drafting, what we send is the public post, the prospect’s public professional details, and the settings you configured; a follow-up draft also gets the campaign’s earlier messages to that person. For labelling, we send the text of one reply, the last message you sent that person and the start of the public post you found them through, each with email addresses, phone numbers and links removed, and nothing else — the rules for mail from a connected Google account are in section 5.4. For column matching, we send the column names and up to three example rows, with email addresses and long numbers masked. What we send to judge a lead you monitor is in section 3.3.

Every AI request we make carries an explicit restriction that it may only be sent to a model endpoint which does not retain it. Our requests are routed by OpenRouter, which selects a provider at the moment of the request from that restricted set, so the company that actually processes a request is not always the company that made the model — our sub-processor list therefore names both the model creators and the clouds that serve them.

One thing you should know about how sending works. Outreach can run in a mode where each message waits for you to approve it, or in an automatic mode where messages drafted by AI are sent without individual review. The automatic mode is a choice you make. If you turn it on, messages composed by a model will be sent from your mailbox, in your name, without a human reading them first. We think that is a reasonable option to offer, and we think you should be told about it in plain words rather than discovering it later.

8. If you received an email or a LinkedIn message from someone using CommunityTracker

This section is for you, not for our customer. You can act on any of it without an account and without proving anything.

Our customers use our product to send business email, and to send connection requests and messages on LinkedIn. We supplied the tooling; the customer chose to contact you and wrote or approved what was said. On LinkedIn the message comes from that person’s own account, not from us. Your details were most likely obtained from a public post you made, or from a business contact data provider — the providers we use are listed on our sub-processor page.

What you can do:

• Unsubscribe, by email. Every email we send carries a working unsubscribe link. Using it stops that sender’s messages to you immediately, including any already queued, and records a permanent suppression so it does not restart.
• On LinkedIn, there is no unsubscribe link — the platform has no such mechanism, and we would rather say so than imply one exists. Any of the following works instead: reply asking to be removed, ignore or withdraw the connection request, or use LinkedIn’s own block and report tools. Telling us by any route also suppresses your email address, and telling us by email also suppresses your LinkedIn profile where we know it — an objection on one channel stops both.
• Just reply and say so. If you reply asking to be removed, our system recognises that and stops the sequence. This works the same way for a LinkedIn reply as for an email one.
• Ask us directly. Email support@communitytracker.ai and we will suppress your address across every customer on our platform, and tell you what we held about you. If a customer monitors you as a lead, include the address of your LinkedIn profile and we will stop that too (section 3.3).

We keep suppression records indefinitely, and deliberately so. A suppression list only works if it outlives the account that triggered it — deleting it would mean you could be contacted again by the next customer who finds your address. We keep the minimum needed to recognise your address and block it. For a LinkedIn profile, that can include a link between two addresses of the same profile (section 9). We keep that link as plain text, so that a request to stop covers both addresses. If you email us to stop or to delete, we keep the address you wrote from on our do-not-contact list, in plain text. Parts of our product can only check a plain address. Some requests are about lead monitoring (section 3.3). For each of those we act on, we also keep a record, so that we can show what we did. It holds when we acted and which of our staff acted. It also holds whether we stopped or deleted, one-way hashes of your profile’s addresses, and counts of what we held and what we changed. We delete our email thread about such a request 90 days after our last reply.

9. How long we keep things

Each of these windows is enforced by an automated job, not by intention, unless its row says otherwise.

10. Your rights, and how to use them

You can ask us to:

• tell you what we hold about you, and give you a copy;
• correct anything that is wrong;
• delete what we hold;
• stop a particular use, or object to it;
• export your data in a portable format;
• withdraw a consent you previously gave.

One address, one procedure. Email support@communitytracker.ai from the address you want us to act on, and tell us what you want. There is no form, no subject-line format to get right, and no account required. We respond within 30 days and will tell you sooner if a request will take longer.

Deletion you can do yourself. The account owner can delete the account from Settings in the application, without asking us first. You type your account email to confirm, and then:

• everything stops immediately — no outreach is sent, no monitoring runs, and the account is frozen so it cannot be used while the deletion is pending;
• the authorisation for any connected mailbox is handed back to Unipile, the provider that holds it, so nothing can read or send from your mailbox from that moment. If that call fails we retry it hourly and say so on the screen, and you can remove our access yourself at myaccount.google.com/permissions;
• we cancel any active subscription so it does not renew. You keep the period you have already paid for and there is no refund for it. If the payment provider is unreachable we retry hourly and tell you on the screen, so you can watch for a charge;
• 30 days later the data is permanently deleted, by an automated job that runs every night.

Frozen means frozen. While a deletion is pending you can still sign in and read your data — you need to, to decide whether to go through with it — but you cannot change anything. That is enforced by the database itself, not by the interface, so it holds even for a stolen session.

You have 30 days to change your mind. Sign back in and cancel it, or use the link in the email we send you the moment you request deletion — that link works without signing in, which matters if someone else scheduled the deletion from your session. Cancelling puts back exactly what the request stopped: the monitoring it paused, the sequences it stopped, the queued messages it cancelled and the API keys it revoked. Anything you had paused yourself stays paused.

Two things do not come back on their own. Your mailbox has to be reconnected, because we handed the authorisation back to Google at the moment you asked, rather than sitting on it for a month. And access tokens for the Claude connector stay revoked, so you would authorise it once more; your API keys are restored.

If you are a team member on someone else’s workspace, your membership is deliberately left in place during the 30 days — so that cancelling can restore it rather than leaving you to be re-invited. You can still see that workspace and still cannot change anything. Both your membership and your ability to sign in end when the deletion completes. Their workspace and its data are untouched throughout.

If you want a copy, take it before you delete. You can export your leads and your action items to CSV yourself from inside the application. There is no one-click export of the whole account, so for anything else — posts, keywords, sequences, settings — email support@communitytracker.ai before you delete and we will send it to you. That address is also how you exercise every other right above, and it remains the route to deletion if you cannot sign in.

Deletion removes your account data and everything associated with it, including the public posts we collected for your keywords and the contact details we found. Seven things deliberately survive, and this is the complete list:

• Suppression records, for the reason given in section 8 — reduced to a one-way hash of the address, enough to recognise it and keep blocking it, not enough to read it or write to it. Deleting these would let someone be emailed again after they told us to stop, which is the harm the right to erasure exists to prevent. Two parts are kept as plain text (section 8). One is a link between two addresses of the same LinkedIn profile. The other is the email address a person wrote to us from, to ask us to stop or to delete.
• Unsubscribe links in mail already delivered, for 30 days, so a recipient can still opt out after the sender is gone. Held only as hashes.
• Billing and tax records, with your name, email, phone and address removed — see section 9.
• A record that a deletion was carried out, identified by a one-way hash rather than by you, so we can show it happened without keeping whose it was.
• The shared copy behind lead monitoring: the public LinkedIn and company data collected for the leads you monitored, the AI model’s answers about it, our records of those checks and Apify’s copy. Some of it is about people you did not monitor, such as posts a company page shared from them (section 3.3). The data and answers are not stored under your account, so they are deleted on the schedule in section 9 instead. Our records of those checks keep the internal ID of your account for up to 90 days after each check. They also record which people and companies you monitored. A check that has not finished, such as one still waiting to run, can keep them longer.
• Cost records of our data collection, for your keywords and for the leads you monitored. They keep the internal ID of your account with no set end date (section 9). Those for your keywords also hold the keywords. Some of those for monitoring also name companies you monitored.
• HarvestAPI’s logs of our requests, which hold your LinkedIn search keywords and the addresses of the leads you monitored and their company pages. We cannot delete them. HarvestAPI says it keeps them for 30 days (section 9).

If you think we have handled your information badly, please tell us first — we would like the chance to fix it. You can also complain to your local data protection authority.

11. How we protect information

• Everything is encrypted in transit (TLS) and at rest.
• Access to customer data is restricted to the small number of people who need it to operate and support the product.
• Database access is enforced per-tenant at the database layer, so one customer’s queries cannot reach another customer’s rows.
• Mailbox authorisation tokens are held by our mailbox provider, not in our application code.
• Passwords are stored only as salted cryptographic hashes.
• Two-factor authentication is available on your account, and we recommend it.

No system is perfectly secure, and we will not pretend otherwise. If a breach affects your information we will tell you and the relevant regulator without undue delay.

12. Where your information is processed

We operate across three locations, and it is worth being specific about which is which:

• Our database — including everything described in section 5 — runs on Amazon Web Services, us-east-2 (Ohio, United States), via Supabase.
• Our website and application are served by Vercel (global edge network).
• Our community-monitoring automation runs on A dedicated server in Mumbai, India (Hostinger). This system handles public posts only. It has no access to connected mailboxes and never receives Google account data.

Our other service providers operate in their own jurisdictions, principally the United States and the European Union. Where information moves between countries, we rely on the transfer mechanisms in each provider’s data processing terms, including the European Commission’s Standard Contractual Clauses where they apply. HarvestAPI is the exception. We use it only through Apify, and we have no data processing terms with it yet. We do not know which country it runs its service in.

13. Children

This is a business product and it is not intended for anyone under 18. We do not knowingly collect information from children. If you believe a child has given us information, email us and we will delete it.

14. Changes to this policy

We update this policy when the product changes. The date at the top always reflects the current version. If a change materially affects how we handle your information — particularly anything in section 5 — we will tell you by email before it takes effect, rather than quietly changing the page.

15. Contact us

For anything in this policy, including any request under section 10, write to support@communitytracker.ai. It reaches a monitored mailbox and a person will answer.

Miraa Labs LLC
1309 Coffee St, Sheridan, Wyoming 82801, United States