OmniaGraph privacy policy
Last updated: 3 November 2026
Who we are
OmniaGraph is a product of Rodan Analytics Limited, a company registered in England and Wales under number 13780448, whose registered office is at 27 Old Gloucester Street, London, England, WC1N 3AX ("we", "us").
We are the controller for account, sign-in, billing, website, support and security data. For the content a customer puts into its workspace, we act as the customer's processor under our terms of service, and the customer decides how that data is used.
Contact us at [email protected], or by post at the address above.
What this policy covers
This policy covers:
- the public website at
omniagraph.com - the OmniaGraph app at
app.omniagraph.com - the OmniaGraph API
- the OmniaGraph MCP server at
mcp.omniagraph.com, which AI assistants such as Claude, ChatGPT and Cursor connect to
What we collect
Your account
- Your email address and, if you give it, your name.
- The workspaces you belong to and your role in each: owner, member or viewer.
- Invitations you send or receive, with the invited email address and role.
Signing in
We do not use passwords. To sign in, you ask for a six-digit code, which we email to you.
- We store the code only as a one-way hash, with its expiry time (10 minutes) and the number of attempts (at most 5).
- We record each sign-in attempt: whether it succeeded, the reason if it failed, the workspace and a one-way hash of your email address and of your IP address.
- The sign-in page uses Cloudflare Turnstile to block automated sign-in attempts. Turnstile runs in your browser and Cloudflare receives information about your browser and connection to do this.
OmniaGraph does not store a password for you. We may also offer sign-in with a Rodan account. Rodan Analytics Limited operates that login, which has its own privacy notice, and its sign-in steps are its own. It only signs you in to an OmniaGraph account that already exists: it does not create an account and does not give you access to a workspace.
- When you sign in this way, Rodan tells us your Rodan user ID, your email address, whether that address is verified and, if it has one, your name. We keep the Rodan user ID, linked to your OmniaGraph account, and the time you last signed in this way. We record the attempt as we record any other sign-in.
- We match you to your OmniaGraph account by email address. If we cannot match you to one account, we refuse the sign-in.
- So that Rodan can recognise you, we may send it the names and email addresses of OmniaGraph users. We do not send workspace content, and your roles and workspaces stay in OmniaGraph.
Your workspace
A workspace holds what you and your colleagues put in it, and what OmniaGraph produces for you:
- monitors (the markets, companies or regulators you follow), their schedules and settings
- watchlists
- questions you ask with Ask, and the conversation threads
- findings, briefs, reports and exports produced by research runs
- private documents you upload
Findings are built from public sources such as regulators' websites, company registers and news. Each finding has an OmniaScore, a 0 to 100 score of how strong its evidence is. The OmniaScore is calculated by fixed rules, not by an AI model.
Private documents
Private documents are encrypted when we store them. Their text is stored encrypted in our database and uploaded files are stored encrypted in Amazon S3 in London (eu-west-2).
To read a PDF or use a document in research, we send its text to an AI model through OpenRouter on a route that neither keeps nor trains on the data (see "AI models" below). You can delete a private document at any time; deleting it removes it from the workspace and from every monitor that used it. Deleted files can remain in versioned storage or backups until those copies expire.
Connected AI assistants
When you connect an AI assistant to OmniaGraph, we store:
- the assistant's registration: its name, the addresses it returns to after sign-in and the access it asked for
- the access granted, the workspace and the user who approved it
- one-way hashes of the access and refresh tokens, never the tokens themselves
- when each token was issued, when it expires, when it was last used (updated at most once a minute) and whether it has been revoked
- a count of the assistant's calls and refused calls for each day, which we keep for 2 days to apply rate limits and show usage in Settings
We do not record the questions an assistant asks OmniaGraph or the results we return. Our web servers do record the address of each request; see "Server logs".
How you use the app
To improve the product, the app records some actions with your workspace and user: opening a monitor, viewing a report tab, expanding a finding, clicking a source, starting and completing an Ask, requesting a monitor run and connecting an AI assistant.
Website and app analytics
The website and the app load Google Tag Manager only if you accept analytics in the consent banner. Your choice is stored in your browser. If you decline, analytics stays off. The Tag Manager container loads Google Analytics.
Error reports
When something fails, the backend and the website and app send an error report to Sentry. Reports exclude personal data by default: we have turned off Sentry's option to send it. We send performance traces for 1 request in 10. We do not record browser sessions or screens.
Server logs
Our web servers write a line for each request, with the time, the address requested, the response code and the client's IP address.
Enquiries
If you use the contact form on the website, we receive what you enter in it. We store it and use it to reply to you.
Connecting an AI assistant
What an assistant can do
An AI assistant connects to one workspace at a time, as you, and can only do what your role in that workspace allows. You see the workspace and the access asked for on a consent screen before you approve. A connection cannot reach other workspaces.
| Your role | What a connected assistant can do |
|---|---|
| Viewer | Read only: search and read findings, monitors, reports, briefs and run status |
| Member | Also start research, create monitors, add to watchlists, use private documents and export |
| Owner | Everything a member can do, and manage the workspace |
The tools an assistant can use are:
- Read: search, fetch an item, list monitors, get a report, ask about stored findings and check a run's status.
- Change: start research, create a monitor, add an entity to a watchlist and upload a private document. Assistants are told these tools change the workspace, so they normally ask you first.
- Paid: Ask with live web research and research runs use your workspace's credits.
An assistant does not get workspace administration access unless it asks for it and your role allows it.
Where the results go
When an assistant calls OmniaGraph, we send the result to that assistant. The result then passes to the assistant's provider, for example Anthropic for Claude, OpenAI for ChatGPT or Anysphere for Cursor. The provider handles it under its own terms and privacy policy, not this one, and may keep it in your chat history. Check those terms before you connect.
We do not control what the assistant or its provider does with the results. Connect only assistants your organisation allows for this data.
Tokens
A connection uses an access token that lasts 1 hour and a refresh token that lasts 30 days. Each time the refresh token is used, we issue a new one and the old one stops working.
How to disconnect
- In OmniaGraph: a workspace owner can see every connected assistant, its access and today's use under Settings, and disconnect it. Disconnecting revokes its tokens at once.
- In the assistant: removing OmniaGraph from the assistant stops it calling us. The assistant may or may not revoke its token with us, so also disconnect it in OmniaGraph.
- Removing a user from a workspace, or lowering their role below what a connection was granted, stops that connection working at once.
Why we use your data
We use the data above to:
- provide the service you or your organisation signed up for (contract)
- run research, score evidence and write briefs (contract)
- keep accounts and workspaces secure, prevent abuse and apply rate limits (legitimate interests)
- fix errors and improve the product (legitimate interests)
- measure use of the website, with your consent (consent)
- answer enquiries and support requests (legitimate interests)
- meet legal obligations (legal obligation)
We do not sell personal data. We do not use your workspace content to train AI models, and we ask the AI models we use not to keep it or train on it (see below).
AI models
Research runs, Ask, document reading and brief writing use AI models from OpenAI, Anthropic, Google, Perplexity and xAI. We reach them through OpenRouter.
Every request we send through OpenRouter asks for zero data retention: OpenRouter may only send it to a model provider that neither keeps nor trains on it. This setting does not cover OpenRouter's own add-ons. We use its file parsing to read PDFs and its web search for research, and those add-ons run under OpenRouter's terms.
All AI processing goes through OpenRouter, to model providers with zero data retention. That includes the research behind public company and news pages such as The Humanoid Index. Web search for those pages runs inside Perplexity's search models, which keep zero data retention. Only the search of X posts uses OpenRouter's search add-on, which is not covered by the zero data retention setting and runs under OpenRouter's terms. We no longer call model providers directly.
Who we share data with
We may share personal data with service providers such as those below, which process it for us, with professional advisers and buyers of our business, and where the law requires it or to protect our rights. We may change our providers.
| Provider | What for | Data |
|---|---|---|
| Amazon Web Services | Servers (Lightsail), database and file storage (S3) | All service data |
| OpenRouter | Routing requests to AI models | Prompts with research scope, questions and document text |
| Model providers reached through OpenRouter: OpenAI (through Microsoft Azure), Anthropic (through Amazon Bedrock or Google Vertex), Google, Perplexity and xAI | Reading sources, extracting findings, writing briefs, answering questions | As above, without retention |
| Exa | Web search for sources | Search queries built from monitor scopes and questions |
| Mailgun (Sinch) | Sending sign-in codes and email updates | Email address, sign-in codes, briefs sent by email |
| Rodan Analytics Limited (the Rodan login) | Signing in with a Rodan account, where we offer it | Names and email addresses of OmniaGraph users; the Rodan user ID and email details it returns when you sign in |
| Cloudflare | Website delivery, network protection and Turnstile | IP address, request details, browser signals |
| Sentry (Functional Software) | Error reports | Error details, without personal data by default |
| Tag Manager and analytics, only with consent | Usage of the website and app | |
| PostHog | Product analytics | Usage of the website and app |
| Workplace, support and sales tools | Internal alerts, enquiries and support | Contact and support details |
OmniaGraph also reads public sources, such as GOV.UK, Companies House, the EU Official Journal, the US Federal Register and SEC EDGAR. We send them search terms built from your monitors, not personal data.
AI assistants you connect (Claude, ChatGPT, Cursor and others) are not our service providers. You choose to send results to them, and their providers act under their own terms.
International transfers
We store service data mainly in the United Kingdom. Your information may be transferred to, processed and stored in other countries, including the European Union and the United States. Where it is, we take the steps the law requires, such as relying on UK adequacy regulations, including the UK–US data bridge where a provider is certified, or on the International Data Transfer Agreement or UK Addendum issued by the Information Commissioner.
How long we keep data
We keep personal data for as long as we need it for the purposes it was collected for. How long that is depends on the type of data and on our legal and business needs.
We take into account:
- how long you or your organisation has an account, workspace or contract with us
- how long we need records for accounting, tax and legal obligations
- how long a claim could be brought, and our need to establish, exercise or defend legal claims or enforce our agreements
- our need to keep the service secure and to investigate misuse
- our need to answer queries and resolve problems
We keep workspace content for as long as the customer's workspace is open. When a workspace closes, we follow the customer's instructions and our terms. Copies in backups are removed as the backups expire.
We may keep anonymised or aggregated data that no longer identifies anyone without limit.
Your rights
You can ask to see, correct or delete your personal data, to restrict or object to how we use it, or to receive it in a portable form. Where we rely on consent, you can withdraw it at any time, for analytics by changing your choice in the consent banner.
If your organisation's workspace holds your data, we may pass your request to your organisation, because it decides how that data is used.
To make a request, email [email protected]. We may ask you to prove your identity, and we will respond as data protection law requires. Some rights have legal limits.
If you are unhappy with how we handle your data, please contact us first. You can also complain to the Information Commissioner's Office at ico.org.uk.
Security
- Connections to OmniaGraph are encrypted.
- Private documents are encrypted at rest.
- Tokens and sign-in codes are stored only as one-way hashes.
- Each workspace is kept separate, and a connected assistant can reach only the workspace it was approved for.
Children
OmniaGraph is a business service and is not meant for children.
Changes to this policy
We may update this policy at any time by posting changes here and changing the date at the top. Where the law requires it, we will tell you directly.