Privacy

Who we are

Insight Lab is built by S2AI Agency Unipessoal Lda, registered in Portugal, and that company is the data controller for the website, for your account and for everything else on this page that we decide on our own.EN101, Avenida Barros e Soares, Nº 423 4715-214 Braga PortugalIf you have a question about any of it, or want something corrected or deleted, write to hello@s2agency.ai and a person will answer.

When your coach decides

In the app, a coach keeps records about their clients: sessions, notes, tasks, files and what the AI writes from them. For those records the coach is the controller, because the coach decides what is written about a client and why, and we process them on the coach’s behalf and on the coach’s instructions. If you are a client and want to see, correct or delete what your coach keeps about you, ask your coach first. If you write to us instead, we will pass the request on and help your coach answer it. A coach who needs data processing terms for their own practice can write to us about them at hello@s2agency.ai.

The waiting list

The website’s waiting list form asks for your email address, and keeps it with the language you were reading the site in and the date you signed up. We use it to write to you once, when Insight Lab opens, and for nothing else. We do not ask for your name and we do not build a profile of you from it. The list cannot be read from the browser: only our server can write to it or read it.

Your account

Coaches and clients both have an account in the app. You sign up with an email address and a password, or with your Google account, and sign-in runs on Supabase Auth. If you use Google, Google tells us your name, your email address and a link to your profile picture, and we never see your Google password. Supabase Auth keeps what Google sent with your sign-in record. Your profile in the app takes your name and email from it, and does not use the Google picture. Your profile holds what you type into it: your name, email, phone, date of birth, country, location, occupation, a short description of yourself, links to your website or social profiles, your language and your timezone. A coach also has a professional profile: a title, a bio, the languages they coach in, how they prefer to work, their years of experience and the address of their public page. The public page can be read by anyone, which is what it is for, and the coach can unpublish it at any time. Someone who asks for a session from a public page creates an account to do it, and the request carries the time they asked for and an optional note to the coach. There is no billing in the app yet, so we hold no card or bank details.

What coaches keep in the app

A coach adds their clients to the app, often before the client has an account: name, email, phone, date of birth, country, location, occupation and links. Once a client joins, the client’s own profile is shown instead of what the coach typed. Around each client the coach keeps the work itself: coaching processes, sessions, tasks, notes, documents, uploaded files, coaching tools such as assessments and wheels, and the agreement the client accepted. From these the app builds a record of what was observed and a picture of the person over time, which the app calls the Person Map. Some of this is shared with the client, and some is for the coach only, such as the coach’s private notes and the AI’s reading of a file. Uploaded files are kept in private storage that only the coach can open, and the client too when the coach shares a file with them. Insight Lab is for coaching. It is not for therapy or health care, and coaches should not store clinical records, diagnoses or therapy notes in it. Notes about a person’s life can still say something about their health, so the agreement the app provides for coaches to send asks the client for explicit consent to their notes being kept and read by the AI on the platform. A coach can edit that agreement. If a coach removes that consent from their agreement, the coach, as the controller of those records, needs another legal basis for keeping them.

How the AI uses your data

The app uses AI models to read uploaded files, summarise sessions and notes, write documents, answer the coach in the assistant and help the coach during a call. It works only on the records of the coach and the client in question. A coach can turn off the AI reading of any single file. Every request to a model goes through the Vercel AI Gateway, and every request asks for zero data retention and no training on the content. If no provider of the model meets both terms, the request is not sent. The models in use are GPT 5.6 Luna from OpenAI for reading and writing, Jev from TypeSafe AI for ranking suggestions during a call, and Grok speech-to-text from xAI, which the gateway lists as SpaceXAI, for transcribing calls. Each request goes first to the model’s own provider. If that provider is unavailable, the gateway can send the same request to another provider of the same model that meets both terms. Today that is Microsoft Azure for GPT 5.6 Luna. Jev is served only by DigitalOcean and Grok speech-to-text only by xAI, so those two have no other provider to fall back to.

What we keep about each AI request

For every AI request we keep two records. The first is a usage record: which coach made it, for which task, with which model and provider, how many tokens it used, what it cost and whether it worked. It holds no content of any kind. It stays when a client is deleted, with the link to that client removed, because it is how we count cost. The second is a redacted copy of the request and the answer, which we use to check whether the AI did its job and to improve how we ask. Before it is written, everything a client, a file or the coach contributed is replaced with a placeholder that only says what kind of text it was and how long it was, so what remains is our own instructions. The answer is cleaned with the same placeholders. That catches names and quoted passages but can miss a fact the model put in its own words, so we treat these copies as personal data. Each copy carries a deletion date twelve months after it was made, and is deleted earlier if its client is deleted. The job that deletes copies on that date is not running yet. Until it is, a copy stays until its client or the coach’s account is deleted, or until you ask us to delete it. The raw request, the raw answer, file contents and text extracted from files are never stored in either record. During the alpha, coaches agree to this recording in the alpha agreement.

Video calls and transcription

Sessions can take place as video calls inside the app. The call runs on LiveKit Cloud, which carries the audio and video between the two people. No audio or video is recorded. We keep when each person joined and left, to count the coach’s call time. LiveKit chooses which of its servers carry a call, and we have not tied calls to a region, so a call can pass through servers outside the European Union. A coach can ask to transcribe a call, and transcription starts only after the client accepts on their own screen. Either person can stop it at any time. While it runs, a transcriber on our server at Hetzner in Nuremberg, Germany, listens to each person’s audio, cuts it into spoken phrases and sends each phrase to Grok speech-to-text through the AI Gateway, with zero data retention. Silence is never sent, and the audio is held in memory only until the text comes back. The text is kept as the call’s transcript, which only the coach can read, and when the call ends it is saved as a transcript file and a summary in the session.

Google Calendar

A coach can connect their Google account to the app, so that their sessions appear in their calendar and the times offered to clients avoid the times they are busy. Nothing is read from or written to Google until the coach connects, and only the coach connects an account. A client never does. When you connect, Google asks you to allow four things, and the app uses each one for a single purpose. Your email address shows which Google account is connected. The list of your calendars lets you choose which ones count when the app checks your availability. Your free and busy times in the calendars you chose keep the app from offering a time when you are busy. Google gives us only the start and end of each busy period, never an event’s title, description, place or guests. A calendar the app creates in your Google account, called Insight Lab, is where the app adds, moves and removes your sessions, with the client’s name, the date, the time, the link to the call and a link to the session in Insight Lab. The app can read and change only that calendar, never your other calendars. We keep the access Google gives us, encrypted, along with your email address, the calendars you chose and the identifiers of the Insight Lab calendar and its events. We do not keep your busy times: the app asks Google when it shows available times, and uses the answer only for that. Data from Google is never sent to an AI model, and never used to train one, ours or anyone else’s. We do not sell it, use it for advertising or share it with anyone. No one at Insight Lab reads it, unless you ask us to for support, or to keep the service secure, or because the law requires it. You can disconnect at any time in Settings, or remove the app’s access in your Google Account’s security settings. When you disconnect, or your account is deleted, we delete the access and everything listed above that we kept. The Insight Lab calendar stays in your Google account, for you to keep or delete. Insight Lab’s use and transfer of information received from Google APIs to any other app will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

Emails

The app sends email through Resend. It sends sign-up and password emails, and a short message when someone makes a decision that affects you, such as an invite, an agreement to accept, a new task, a session request, or a session that was moved or cancelled. Nothing about the coaching itself is sent by email, and the app sends no marketing email. The waiting list gets one email, when we launch. It will carry an unsubscribe link and our postal address, and unsubscribing takes effect immediately rather than within the days the law would allow us. If we ever want to send you something else, we will ask first.

The anti-spam check

The waiting list form, and the app’s sign-up, sign-in, password reset and public session request forms, are protected by Cloudflare Turnstile, which stops robots from creating accounts and filling the list with addresses that are not real. To tell you apart from one, Turnstile receives your IP address and some technical signals from your browser, and Cloudflare processes them on our behalf. It is not an analytics tool and it does not follow you around the web. Cloudflare describes what it does at cloudflare.com/privacypolicy.

Analytics and cookies

The website and the app use PostHog, on PostHog’s EU cloud, to see which pages are used and where people get stuck. Until you choose, PostHog runs without cookies: it stores nothing in your browser, does not record your visit and forgets you when you close the tab. A banner asks whether you accept analytics, and declining takes one click, the same as accepting. If you accept, PostHog keeps an identifier in a cookie and in your browser’s storage, so it can recognise you on later visits and across the website and the app, and it records sessions to show how pages are used. In those recordings everything you type and all the text on the page are hidden, so a recording shows layout and movement and no words. The call screen is never recorded. While you are signed in, PostHog receives your account id and whether you are using the app as a coach or as a client, and never your name or email. PostHog drops your IP address when an event arrives, and does not capture what your browser’s console prints. Events are kept for one year and recordings for 30 days. For a coach who accepted analytics, our server also sends PostHog a record of each AI request, linked to the coach’s account id: the task, the model and provider, the number of tokens, the cost, how long it took and whether it worked. It also tells PostHog whether the coach accepted, edited or discarded an AI draft, and whether they confirmed, declined or undid something the assistant proposed. It never sends the requests, the AI’s answers or anything from the coaching itself. Your answer is kept for one year in a cookie called il_analytics_consent. You can change it in the app under Account, Usage analytics, or on the website with Analytics settings at the bottom of the page. Turning analytics off stops the recording and clears what PostHog stored in your browser. The app also sets the cookies it needs to work: your sign-in session, your language, and your choices of theme, sidebar, timezone and whether you are working as a coach or as a client. Some layout choices are kept in your browser’s storage. None of these are used for analytics or advertising, and there is no advertising pixel on either site.

Error and support reports

When the app hits an error, it sends a report to Sentry, in Sentry’s EU region. The report says where the error happened and what broke. Before it leaves, the page address is cut down to its path, you are reduced to an account id, email addresses are removed and anything you typed is dropped. Only the live app sends reports. Sentry keeps each report for 30 days. If you report a problem from inside the app, the report is saved in our database and posted to our team’s support channel on Discord. It carries what you wrote, your name, email and role so that we can answer you, the page you were on, your browser and any recent errors. The report in our database is kept until your account is deleted, and the message in Discord until we delete it. Either one is deleted sooner if you ask.

Where it is stored, and who can see it

The database and the uploaded files are in Supabase, hosted in Ireland. The website and the app run on Vercel, and the app’s server runs in Vercel’s Dublin region. Background work, such as reading a file or closing a session, runs on Vercel Workflow. Workflow keeps the state of each job in Vercel’s Washington, D.C. region. That state holds only identifiers and outcomes, never the coaching itself, and Vercel deletes it no later than 30 days after the job ends. The companies in the table below handle data on our behalf. The AI providers are reached through Vercel’s AI Gateway, so our agreement for them is with Vercel. Google, if you sign in with it or connect your calendar, and Discord, which holds our support channel, take part under their own privacy policies. Several of these are American companies, so some of the handling happens outside the European Union. The data processing terms that Vercel, Supabase, LiveKit, PostHog, Sentry, Resend and Cloudflare publish cover these transfers with the European Commission’s standard contractual clauses. Sentry, Resend, Cloudflare and Discord are also certified under the EU-US Data Privacy Framework. Inside the app, rules in the database itself decide who can read what. A coach reads only their own clients, a client reads only what their coach shares with them, and no coach can read another coach’s records. We do not sell personal information, and we do not share it for anyone else’s advertising. Not now, and not in the twelve months behind us.

Who handles your data

Why we use it

We use your account and profile to provide the service you signed up for, which the GDPR calls performance of a contract. The waiting list, analytics and the alpha’s AI records rest on your consent: the box you ticked, the banner you accepted and the alpha agreement. Transcription rests on the client accepting it, on every call. The anti-spam check and error reports rest on our legitimate interest in keeping the service secure and working. For the records a coach keeps about a client, the legal basis is the coach’s, and it comes from the agreement the client accepted. You can withdraw your consent at any time, and that does not affect anything done before you did.

How long we keep it

Your account and profile are kept until the account is deleted. A coach can delete their account from Account in the app. The deletion waits thirty days, and the coach can cancel it at any time until then. When it runs, everything the coach kept goes with it: clients, processes, sessions, notes, documents, files, the Person Map and the redacted AI copies. A client cannot delete their account from the app yet. Write to hello@s2agency.ai and we will do it. Deleting a client’s account removes their sign-in and their profile. It does not by itself delete what the coach keeps about their work together, because those records are the coach’s. If you want those deleted too, say so, and we will pass the request to your coach as described above. When a coach deletes a client, the client’s records go with them, including the redacted AI copies that mention them. A deleted assistant conversation is erased for good thirty days after it was deleted. Usage records hold no content and are kept indefinitely. Supabase keeps a daily backup of the database for seven days, so something deleted can remain in a backup for up to seven days after it is gone from the app. An address on the waiting list is kept until we launch and write to you, or until you ask us to delete it, whichever comes first. If we decide not to launch, we delete the list rather than keep it for something else.

What you can ask for

A copy of what we hold about you, a correction, or deletion. One email to hello@s2agency.ai is enough, you do not have to give a reason, and we will not treat you differently for asking. For records a coach keeps about you, we will work through your coach, as described above. This applies wherever you live. The laws below give some people a formal right to it, and we would rather not make everyone else ask nicely.

Your rights under the GDPR

We are established in the European Union, so the GDPR governs everything on this page, wherever you live. Alongside the rights above you can object to what we do with your data, ask us to restrict it, and take it elsewhere in a portable form. Where we rely on your consent, you can withdraw it at any time. If you think we have got it wrong, you can complain to your national data protection authority. In Portugal that is the CNPD, and in the UK the ICO.

If you are in Brazil

The LGPD, Brazil’s data protection law, gives you the rights in its article 18: to confirm that we process your data, to see it, to correct it, to have it anonymised, blocked or deleted when it is unnecessary or excessive, to take it to another provider, to have data processed with your consent deleted, to know who we share it with, to know what happens if you do not consent, and to withdraw your consent. To use any of them, write to hello@s2agency.ai. If you think we have got it wrong, you can complain to the ANPD, Brazil’s national data protection authority.

If you are in the United States

California, Colorado, Connecticut, Virginia and a growing list of other states give you the right to know what is held, to have it deleted, and to opt out of its sale or of targeted advertising. There is no sale to opt out of, and no advertising built on your data. To use any of the rest, write to the address above. If you think we have got it wrong, a Californian can raise it with the California Attorney General.

Children

Insight Lab is not meant for anyone under 16, and we do not knowingly hold their data. If you think someone under 16 has given us their data, write to hello@s2agency.ai and we will delete it.

If this page changes

We will update the date at the top. If a change is material, such as a new company handling data we already hold or a new use for it, we will tell you before it applies: coaches and clients by email or in the app, and the waiting list by email.