Privacy & Terms

Eco is a "second brain" that captures the tacit knowledge behind your decisions — the how and why, not just the what — so it can one day answer questions in your place. You write notes; Eco turns them into structured knowledge and can be asked about them later.

Eco can also be used in a company/pilot context, where a team's specialists opt in to share knowledge with a shared successor agent. That mode adds extra rules covered below (see "Sharing & consent" and "Meeting & third-party data").

• Account info: name, email, and your login credentials (handled by Supabase Auth).
• Content you create: notes, the knowledge entries extracted from them, tags, projects, and comments.
• Vector embeddings of your notes/entries, used to power search and the agent — stored in the same database, not a separate third-party vector service.
• Usage signals directly tied to the product working (e.g., which entries were cited in an answer, so royalties/attribution can be computed).
• If you import a meeting transcript: the transcript text, and — only with your explicit confirmation — a mapping of who's who in that meeting.

We don't collect more than the product needs to function, and we don't sell your data.

Your data is stored in a PostgreSQL database managed by Supabase, hosted in the South America (São Paulo) region. Vector search runs in the same database — there is no separate vector-database vendor. The app itself runs on Vercel. Backups and infrastructure security are managed by these providers per their own security practices.

To turn your notes into structured knowledge, generate answers, and compute embeddings, your content is sent to Google Geminiat the moment of that request (writing an entry, asking a question, etc.). We don't control Google's infrastructure, but our usage does not opt your content into being used to train Google's general models. We also use Resend to send transactional email (e.g., account-related notices). No other third parties receive your content.

Every table holding your content is protected by Row Level Security policies enforced directly by the database — not just application code. In practice: your notes, entries, and conversations are readable and writable only by you. A background job that needs to write system-level records (like royalty accounting) runs under a restricted server-side role that never touches your browser or session.

By default, nothing you write is visible to anyone else — not even a teammate. An entry only becomes available to a successor agent (in the pilot/company context) when you explicitly mark it as shared, one entry at a time (or in a batch you trigger yourself). Every share is logged with a timestamp as a lightweight consent record. You can revoke sharing on any entry at any time.

If you import a meeting transcript, Eco can detect who's who and asks you to confirm which speaker is you — knowledge is only extracted from your own words; other participants' words are treated as context, never learned as yours.

A separate, optional "group dynamics" analysis exists for company pilots. It is off by default, requires the company to explicitly enable it, and requires you to confirm that every participant was told the meeting would be analyzed this way before anything runs. Results are phrased as hypotheses, never attributed to a person by name — only roles or the group as a whole.

• Access and export what Eco has learned from you.
• Correct or edit any note or entry at any time.
• Clear Eco's knowledge (in Settings) — resets what the agent has learned from you.
• Request full account deletion, including a hard delete of your notes and entries.

To exercise any of these, email jvalmd00@gmail.com.

We keep your content for as long as your account is active. Deleting a note removes it, but by design an entry you already shared (and consented to) can outlive the note it came from — that's the explicit "teach once" model. "Clear Eco's knowledge" or a full account deletion request overrides that and removes it for good.

Eco is in active beta. Features can change or break, and we'll do our best to keep your data intact but can't promise the same uptime or polish as a mature product. Use it accordingly — don't rely on it as your only copy of anything critical yet.

• You own what you write. Using Eco, you grant us the limited right to process it (as described above) solely to provide the service to you.
• Don't upload someone else's sensitive personal data without a lawful basis to do so — the meeting-analysis consent flow exists precisely to keep this honest.
• Don't use Eco to try to extract, scrape, or misuse other users' shared knowledge outside the product's own sharing controls.
• We can suspend accounts that abuse the service or put other users' data at risk.

Answers Eco gives you are generated from your own notes using a large language model. They can be incomplete or occasionally wrong, and citations point back to your source notes so you can verify anything before acting on it. Eco is not a substitute for professional, legal, medical or financial advice.

We'll update this page as the product changes and update the date at the top. For anything not covered here, or to exercise your rights, email jvalmd00@gmail.com.