up to $20

Privacy policy

Last updated: September 30, 2026

plori ("we") runs plori.ai and api.plori.ai: cloud AI agents, each in its own environment. This page describes what we collect, where your data goes when agents work, and how to get it removed. It is written to be read.

The short version

  • We collect what the service needs to run: your email, your agents' data, usage records, and service logs.
  • The website has no ads. Cloudflare Web Analytics measures page traffic without cookies. We record limited product usage on our own servers to understand how people use plori. The X advertising pixel measures visits from our X ads.
  • We never sell your data. We do not use your content to improve or train models unless you turn that on in Settings; it is off by default, and it only ever applies to activity from after you turned it on.
  • Your agents talk to model providers and web tools. Those calls leave our infrastructure and are listed below.
  • Deleting an agent wipes its data. Email us to delete everything.

What we store

Account. A registered account has a verified email address. You can sign in with Google or a one-time code sent to that address, so there is no password to store. When you use Google, we store Google's stable account identifier and the verified email it returns; we do not keep the Google access token or profile data. An anonymous trial is a random identifier kept in your browser, with no email attached.

Agent data. What your agents need to work: conversations and instructions, the files on the agent's disk, agent settings, memory notes, and environment variables you set. Environment variables marked secret and connected provider keys (BYOK) are encrypted at rest and are never shown back once saved.

Billing. Payments are processed by Dodo Payments, our payment provider and merchant of record. We keep transaction references, your plan, and the credit ledger. Card numbers never touch our servers.

Usage. Runs, tool calls, token counts, and credit spend, kept for metering, billing, and quota enforcement.

Logs. Service logs and request metadata, including IP addresses, kept on our own infrastructure for security, abuse prevention, and debugging. Diagnostic traces of agent runs can include run content; access is restricted to operators, and they are used only for reliability work.

Google user data

plori touches your Google account in two separate places, both optional, and this section says exactly what each one gets.

Signing in with Google. The sign-in request asks for openid, email, and profile. We keep Google's stable account identifier and the verified email address it returns, and we use them for one thing: recognizing your account when you come back. We do not store the access token, the refresh token, your name, or your picture.

Connecting a Google Sheet to a workflow. If you build a workflow that writes to a spreadsheet, plori asks for a single scope, https://www.googleapis.com/auth/spreadsheets, and uses it only to read and write the spreadsheet your workflow names, while that workflow runs. We request no Google Drive scope, so we cannot browse, list, or open anything else in your Drive. The tokens for this connection are encrypted at rest and are deleted when you disconnect the account, delete the workflow, or delete your plori account. You can also revoke plori's access at any time from your Google account permissions page.

What we receive from Google APIs is used to provide the feature you asked for, and for nothing else. We do not sell it, do not transfer it except as required by law, and never use it for advertising or profiling. We never use it to develop, improve, or train any AI or machine learning model. The "Help improve plori" switch described under Model training does not cover data from a connected Google account, whatever it is set to. Only operators who need it for support or debugging can reach it.

plori's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

Where data goes when agents work

A plori agent does real work, and some of that work leaves our servers:

  • Model calls. Prompts and completions go to the model provider serving the request: OpenRouter for hosted routing, or the provider whose key you connected (OpenAI, DeepSeek, Groq, xAI, Together AI, or OpenRouter).
  • Web search and fetch. Agent web searches are served by Tavily and page fetches by Jina, so the query or URL is shared with them.
  • Wherever you point your agent. Agents can use the network. If you tell your agent to call an API, clone a repository, or visit a site, that traffic goes where you sent it, under that service's terms.

Service providers

We use a small set of providers to run plori: Vultr (cloud servers and storage, United States), Cloudflare (web hosting, DNS, networking, and cookieless web analytics), Google (optional account sign-in and user-authorized workflow connections), Dodo Payments (payments), Resend (email-code sign-in and account email), Tavily (web search), Jina (web fetch), the model providers listed above, and Sentry (website error reporting).

We store product usage records on our own servers. These include a limited set of website events, such as page views and interactions with the agent message box, and service records such as agent creation and runs. We limit website event properties to values such as the page path and referring site. They do not include prompts or files. Once you have an account, some usage records can be linked to it.

For MCP connections, we record authorization attempts and limited tool-use metadata. This includes the MCP client, tool name, outcome, duration, and whether an agent run or workflow started and completed. We do not put tool arguments, results, prompts, or files in these analytics records. A browser identifier or your account may link an MCP authorization attempt to an earlier visit to our MCP page. We use these records to measure and improve the service.

Sentry receives a report only when the plori website itself crashes in your browser: the error message, the stack trace, the page you were on, and your browser version. We strip email addresses and anything credential-shaped out of every report before it is sent, and we never attach your name, email, or account details to it. It never receives your prompts, your agent's output, or your files.

Model training

By default, none of your content is used to improve or train models. That is the setting until you change it.

If you want to help, Settings has a switch called "Help improve plori": while it is on, we may use your conversations, agent activity, and workflow runs to improve the models and agents behind plori. Rating a run good, ok, or bad after it finishes is ordinary product feedback and is not that consent; the rating alone never makes the run's content usable for training. Some details, because they matter:

  • It is opt-in and off by default. Nothing you did before turning it on is ever used; consent applies only to activity from that moment forward.
  • You can turn it off anytime, with the same switch, as easily as you turned it on. Activity from after you turn it off is not used. To be honest about the limit: data that already went into a completed training run cannot be removed from that model.
  • Your choice changes nothing else. Agents work the same, plans cost the same, whether this is on or off.
  • Secrets are excluded. Credentials you enter through secret cards and secret environment variables never enter this data in the first place.
  • Google account data is excluded. Anything we receive from a connected Google account, including spreadsheet contents, is never used for this, on or off.
  • Before any of this data is used, we remove personal information such as names, email addresses, and phone numbers that appear in content.
  • Anonymous trials cannot turn this on; only a registered account can give or withdraw this consent, and we keep a record of each change so your choice is provable.

Retention and deletion

  • Deleting an agent wipes its disk, conversations, and derived data.
  • Anonymous trials expire after 7 days and their data is removed.
  • Delete your account, or ask us to, and everything goes except records we are required to keep for tax and accounting.
  • Logs and diagnostic traces are kept for a limited period and then deleted.

Security

Each agent runs isolated in its own environment with its own storage identity, internal services authenticate to each other with mutual TLS, traffic to plori.ai and api.plori.ai is encrypted in transit, and provider keys and secret environment variables are encrypted at rest. No system is perfect: if we learn of a breach that affects your data, we will tell you.

Your rights

Email us to access, export, correct, or delete your data. If your jurisdiction grants specific privacy rights (for example under GDPR or CCPA), requests go through the same address and we honor them.

Cookies and tracking

The dashboard keeps your session token in your browser's local storage. We also use local storage for a device identifier and referral information. A first-party attribution cookie lets the sign-in flow read that referral information. A separate first-party plori_device cookie lets the MCP authorization page read the same browser identifier. We may use it to link that authorization attempt to an earlier visit to our MCP page. These cookies do not authenticate you. We use the X advertising pixel to measure visits from our X ads. The pixel sends page visits and browser information to X. X may use advertising cookies to associate a visit with an ad interaction. See X's conversion tracking documentation.

Our servers also send registration and trial-start confirmations to X. These include an X ad click identifier when available. Registration confirmations include a hashed account email. Anonymous trial confirmations include the IP address and browser user agent. The conversion API does not receive the raw email.

Cloudflare Web Analytics measures page traffic on plori.ai without cookies or fingerprinting. We send a limited set of product events to our own servers and store them there. These records help us understand page visits and product use.

Children

plori is not directed at children under 16, and we do not knowingly collect their data.

Changes

We will post any changes to this policy here and update the date at the top. If a change is material, we will say so in the dashboard or by email.

Contact

[email protected]