Assistant data.
Colleague processes the messages, files, workspace data and task context needed to provide the assistant service.
Privacy & terms
This page explains what personal data Colleague processes, why it is processed, which providers help operate the service, how long data is kept and the terms that apply when you use Colleague.
Last updated: 3 June 2026.
Colleague processes the messages, files, workspace data and task context needed to provide the assistant service.
Connected accounts are optional and only apply when you choose to enable them or ask Colleague to use them.
Colleague produces drafts, suggestions and assistance for review; you remain responsible for checking outputs before relying on them.
You can stop sharing material, disconnect optional services and ask for Colleague-held data to be deleted or isolated.
Colleague is intended for legitimate academic, research, administrative and professional productivity work, including:
Do not use Colleague for material that is sensitive, restricted or governed by an agreement, policy, ethics approval, contract or data management plan that does not allow this use. Excluded material includes:
You are responsible for deciding what to share with Colleague and for making sure that use fits your institutional policies, contracts, funder rules, ethics approvals and data management plans.
You keep ownership of material you provide to Colleague. Sharing material does not transfer ownership, but you are responsible for making sure you have the right to share source material with the service, including third-party papers, images, datasets, institutional documents and copyrighted works.
Colleague outputs are draft assistance. Review outputs before using or publishing them.
Colleague is not a substitute for legal advice, clinical advice, financial advice, institutional approval, ethics review, formal data governance review or intellectual-property advice.
Colleague does not replace institutional approval, ethics review, data governance or funder requirements. Use Colleague only where your policies, contracts, ethics approvals and data management plans allow it.
You can change what Colleague can access by unsharing files, removing it from calendars, disconnecting optional services or asking us to remove Colleague-held workspace data, memory, files or connected account access.
Deletion removes the relevant content from active use. Some technical logs, backups or records may remain for a limited period where needed for security, troubleshooting, accounting, legal compliance or abuse prevention.
If you share something by mistake, contact [email protected] with enough detail to identify it so we can help remove or isolate it.
Colleague is designed for normal professional use by individual researchers and small academic teams. Very large document-processing jobs, automated bulk processing, scraping or unusually high-volume use may need a separate agreement or a narrower task.
Colleague uses the information you provide to deliver the assistant service: understand requests, remember relevant working context, work with files you choose to share and help with academic drafting, editing, planning, analysis and admin.
Colleague can work from messages, instructions and material you choose to share. The Google Workspace account gives you a familiar way to share documents, folders, calendars and email context with the agent. Other personal-account connectors are optional and should be enabled only when you deliberately want that mode of work.
Telegram is the current chat channel for Colleague; the optional connectors are the extra services you choose to add around it.
Colleague does not sell your data. We will not use your content in public examples or marketing without permission.
When a model or connected tool is needed, Colleague aims to send only the context required for the task. The practical rule is simple: share the material that lets the agent do useful work and keep the rest outside the task.
If you request early access or contact Colleague through the website, we collect the information you provide, such as your name and email address.
We use this information only to respond to your enquiry. We do not use your details for marketing updates.
Enquiry records are normally deleted after 90 days, unless we need to keep a specific record for longer to handle an ongoing request, security issue, legal obligation or dispute.
The website may use Cloudflare Turnstile or similar anti-abuse checks. These services may process technical information, such as request metadata, to help distinguish genuine submissions from spam or automated abuse.
Depending on the circumstances, you may have rights to access, correct, delete, restrict or object to the use of your personal data and to receive a portable copy of personal data you provided. You may also be able to withdraw consent where processing is based on consent.
To exercise these rights, contact [email protected]. You also have the right to complain to the UK Information Commissioner’s Office at ico.org.uk/make-a-complaint.
Under UK data protection law, Colleague needs a lawful basis for each use of personal data. The main lawful bases are:
| Use | Data involved | Lawful basis |
|---|---|---|
| Providing Colleague and responding to your requests | Messages, files, workspace data, memory, selected excerpts and task context. | Contract: to provide the assistant service you ask for. |
| Optional connectors and connected accounts | Connection metadata, authentication data, account data and task material from services you choose to connect or share. | Contract: to provide the connected-service features you choose to use. |
| AI model processing | Prompts, messages, relevant excerpts, instructions and action results needed for a task. | Contract: to provide AI-assisted drafting, analysis and task support. |
| Security, abuse prevention and reliability | Technical logs, request metadata, access checks, security signals and telemetry. | Legitimate interests: to keep Colleague secure, reliable and protected from misuse. |
| Support, deletion requests and rights requests | Contact details, request history and records needed to handle the request. | Legitimate interests and legal obligation where data protection law requires a response. |
| Enquiries and early-access requests | Name, email address and anti-abuse metadata. | Legitimate interests: to respond to enquiries, manage access requests and prevent spam or abuse. |
| Legal, compliance or dispute records | Records needed for accounting, legal compliance, security, complaints or dispute handling. | Legal obligation where required by law; otherwise legitimate interests. |
Colleague is designed for lower-risk academic and professional work. If material outside that scope is shared by mistake, Colleague uses it only as needed to isolate or delete it, respond to the issue, meet legal obligations or protect the service.
Colleague does not make solely automated decisions with legal or similarly significant effects about you. It produces drafts, suggestions, analysis and task assistance for human review.
Colleague uses trusted third-party providers to run the service, connect to user-approved tools, deliver AI functionality and keep the platform reliable.
These providers may act as subprocessors, connected-service providers or independent services depending on how you use Colleague and which account you choose to connect.
Colleague uses Hetzner for its runtime workspace, memory and service infrastructure. Material shared with the agent's Google Workspace is stored by Google. Files in connected personal accounts remain with their provider unless you ask Colleague to work with them; for that task, relevant copies or excerpts may be synced into the Colleague runtime workspace and sent to a model or tool provider as needed.
| Provider | Role in Colleague | Data involved |
|---|---|---|
| OpenAI | AI model inference and OpenAI subscription connection | Task prompts, messages, relevant excerpts and account-authentication data. |
| Anthropic | Optional AI model fallback | Task prompts, messages and relevant excerpts if an Anthropic model is enabled. |
| Hetzner | Runs Colleague infrastructure and runtime storage | Runtime workspace files, memory notes, configuration, logs and task data processed by Colleague. |
| Cloudflare | Network, access and security | Request metadata, access checks and security signals. |
| Agent Google Workspace and optional connected Google services | Docs, Drive files, Gmail, Calendar, Sheets, Slides, Tasks, account data and authentication data you share with the agent's account or connect through Google. | |
| Microsoft | Optional Outlook and OneDrive connectors | Only Microsoft account, authentication, email or file data you choose to connect or ask Colleague to use. |
| Dropbox | Optional Dropbox connector | Only Dropbox account, authentication, file or folder data you choose to connect or ask Colleague to use. |
| GitHub | Optional GitHub connector | Only GitHub account, authentication, repository, issue or pull-request data you choose to connect or ask Colleague to use. |
| Telegram | Chat interface | Messages routed through the bot and related chat metadata. |
| Grafana Cloud | Optional telemetry and reliability monitoring | Operational logs, traces and metrics if external telemetry is enabled. |
Some providers used by Colleague may process data outside the UK or the EEA. Where this happens, Colleague relies on the provider’s transfer safeguards, such as adequacy arrangements, standard contractual clauses or equivalent safeguards where applicable.
Colleague is operated by Louise Foley trading as Colleague. For privacy questions, deletion requests or concerns, contact [email protected].