Rene AI assistant is reached by text message rather than by opening an app, and the first two services it asks to connect are Gmail and Google Calendar. That pairing is the whole product in miniature: the mail is where commitments arrive, the calendar is where they land, and the gap between the two is the work most people do by hand every morning.
Rene launched on 16 September 2026 from Second Enlightenment Limited, a New York studio that also builds the creative workspace Sunra. The agent lives in Messages, SMS, Telegram and WhatsApp, and it describes its own job as “the coordination layer” over the apps you already use. Connecting Gmail lets it triage replies and draft them. Connecting Google Calendar lets it see a conflict before you do, propose the fix, and write the change once you say yes.
Connecting a mailbox and a diary to any agent is a permissions decision before it is a productivity one, and Rene has published more of the detail than most. This article goes through what the Rene AI assistant connects to, how it obtains access, what Google’s Limited Use rules oblige it to do with that data, which models see your text, and the checklist worth running before you point any of this at a work account. It sits in the same category as the wider wave of artificial intelligence tools moving into messaging.
Table of contents
- What the Rene AI Assistant Connects To
- How the Rene AI Assistant Gets Access
- What Google Limited Use Rules Require of the Rene AI Assistant
- The Model Layer Behind the Rene AI Assistant
- What the Rene AI Assistant Does in Practice
- Rene AI Assistant Checklist Before Connecting a Work Account
- How the Rene AI Assistant Documents Line Up With the Launch
- Rene AI Assistant Risks and Limits the Terms Spell Out
- Rene AI Assistant Questions, Answered
- References
What the Rene AI Assistant Connects To
The Rene AI assistant homepage cycles through nine named integrations, each with a one-line description of the job it does. Read together they describe an agent aimed at the coordination layer rather than at any single application.
Gmail is the first connector
The Gmail job stated in the Rene AI assistant marketing copy is triaging replies: the assistant reads what arrived, decides what needs an answer, and drafts it. The homepage walks through the pattern in one exchange — a meeting has moved, the email announcing it has been read, a reply has been drafted, and the calendar has not caught up yet. The Rene AI assistant surfaces all three facts in one message and waits.
Google Calendar is the second
The Google Calendar job is described as clearing conflicts, which is a more specific claim than creating events, and it is the harder half of what a Rene AI assistant is asked to do. The example on Rene’s own site is a 10am that moved to 11am while the calendar still said 10 — the sort of mismatch that only shows up when something is read against something else. A Rene AI assistant with both connectors can compare an inbox against a diary, which neither app does on its own.
The other seven connectors
Beyond the two Google services, the Rene AI assistant homepage names Slack for syncing updates, Notion for turning notes into next steps, Linear for filing issues, GitHub for prioritising reviews, Outlook for drafting follow-ups, Apple Health for context, and WHOOP for recovery tracking.
| Service | Job as Rene describes it | Group |
|---|---|---|
| Gmail | Triaging replies, drafting them | Mail and diary |
| Google Calendar | Scheduling, clearing conflicts | Mail and diary |
| Outlook | Drafting follow-ups | Mail and diary |
| Slack | Syncing updates | Work tools |
| Notion | Turning notes into next steps | Work tools |
| Linear | Filing issues | Work tools |
| GitHub | Prioritising reviews | Code |
| Apple Health | Checking context | Health |
| WHOOP | Recovery tracking | Health |
Grouping those nine gives three for mail and diary, three for work tools, one for code and two for health. As a share of nine connectors that is 33.3%, 33.3%, 11.1% and 22.2% respectively.
What stays the source of truth
Rene’s FAQ is unusually direct about the limits of this. Asked whether it replaces existing apps, the answer is no: “Rene connects to the tools you already use and turns their scattered context into clear next steps. Your email, calendar, notes, and work apps remain the source of truth.” The Rene AI assistant is a reader and an actor on top of those systems, not a new home for the data.
How the Rene AI Assistant Gets Access
For a Rene AI assistant the connection model matters more than the feature list, because it defines what can go wrong and who can stop it.
Delegated access, never a password
The Rene AI assistant uses OAuth-style delegated access for every connected service. Its Privacy Policy states the consequence plainly: “You do not give Rene your service password.” That is the standard pattern for a modern integration and the right one, but it is worth naming because a surprising number of assistants still ask for credentials directly.
Composio sits on the connection path
One name in the policy deserves more attention than it usually gets. Rene says it “may use Composio as a service processor to create and maintain OAuth connections and carry out actions you request in connected services”, and adds that Composio “is not used as an AI model provider for your requests”. So a third party holds and exercises the token that reaches your mailbox, with a scope limited to connection management and action execution rather than inference.
Disconnecting, and what it does not undo
Disconnecting a service stops new Rene AI assistant access and, for Google specifically, revokes the associated authorisation token. What disconnection does not do is reverse anything already done. The Terms are explicit: “disconnecting a service may not undo actions Rene already took on your behalf.” Prior conversations, artifacts, logs and records may also remain until they are deleted under the policy.
More than one account per service
The policy notes that some services allow more than one account to be connected, and that Rene may display limited account identifiers — a GitHub handle, for instance — so you can tell the connections apart. For anyone running a personal and a work Google account side by side, that distinction is the difference between a useful assistant and an accident.
| Access question | What Rene commits to | Where it is stated |
|---|---|---|
| Does it store my password? | No, delegated access only | Privacy Policy |
| Who holds the OAuth token? | Rene, with Composio as processor | Privacy Policy |
| Can I revoke it? | Yes, disconnect in the app | Privacy Policy |
| Does revoking undo actions? | No | Terms of Service |
| Is my data sold? | No | Privacy Policy |
| Is it used to train models? | No | Privacy Policy and FAQ |
What Google Limited Use Rules Require of the Rene AI Assistant
Because Gmail and Google Calendar are Google APIs, the Rene AI assistant is bound by the Google API Services User Data Policy, including its Limited Use requirements. Rene states that its “use and transfer to any other app of information received from Google APIs adheres to” that policy, and then restates the specifics.
The forbidden purposes
Four uses are ruled out by name: selling Google user data, using or transferring it for advertising, using it for profiling, and using or transferring it for “training or improving generalized or foundation AI or ML models”. Anything unrelated to the feature you requested is out too. That last clause is the broad one, and it is the reason a Rene AI assistant may read a thread to answer your question but may not mine the mailbox for anything else.
The minimum-transfer rule
Where a requested feature needs an AI provider or another service provider, Rene commits to transferring “only the minimum Google user data needed to provide that feature”. That is a data-minimisation promise applied to the inference path itself, not just to storage.
When a human can read it
Rene lists four exceptions where a person may see Google user data: you gave explicit permission for a specific interaction, access is necessary for a security or abuse investigation, access is required by law, or the data has been aggregated and anonymised for a permitted internal operation. Those are the standard four, and the value is in seeing them written down.
Deletion and revocation
Two commitments cover the exit. Disconnecting a Google account stops new access and revokes the token. Deleting a Rene account revokes the connected Google authorisation and deletes the associated Google user data, subject only to limited legal, security, fraud-prevention and backup-retention requirements. Anyone assessing a Rene AI assistant for business use should note that the backup carve-out is doing real work in that sentence — it always is.
The Model Layer Behind the Rene AI Assistant
Knowing which companies see your mail is a separate question from knowing which company holds your account, and the Rene AI assistant publishes the answer on a dedicated page.
Three named providers
The AI Provider List, last updated 18 August 2026, names OpenRouter as an AI routing provider and OpenAI and DeepSeek as AI inference providers. Routing means the model handling a given request is chosen dynamically, so the Rene AI assistant is not a wrapper around one vendor.
| Company | Role as published | Sees your prompts? |
|---|---|---|
| OpenRouter | AI routing provider | Yes, in transit |
| OpenAI | AI inference provider | Yes, for routed requests |
| DeepSeek | AI inference provider | Yes, for routed requests |
| Composio | OAuth and action processor | No, not a model provider |
Consent before any of them sees anything
The unusual commitment is the gate in front of that list. Before Rene first sends your data to an external AI provider it asks for consent, and the consent can be withdrawn later in Settings. Withdrawing it does not break the account; it disables the features that require external processing. Very few assistants expose that switch at all, and for a Rene AI assistant reading a mailbox it is the most consequential control on the page.
What actually gets sent
The policy enumerates what may travel from the Rene AI assistant to a provider: text you provide, voice or transcripts, images, files, the conversation context needed for the request, relevant contact fields, and data from services you connect. Read alongside the Google minimum-transfer promise, the intended shape is a per-request slice rather than a mirror of your mailbox.
What the Rene AI Assistant Does in Practice
The documents describe the boundaries. The product pages describe what a Rene AI assistant actually does with the access, and there are four patterns worth knowing.
The morning briefing
The flagship Rene AI assistant example is a start-of-day summary that has already done the reading. A meeting moved, the email is handled, the calendar is stale, and the fix is one reply away. This is the case where connecting both Google services pays for itself, because neither Gmail nor Google Calendar can notice the mismatch alone.
The two-agent scheduling case
The headline architectural claim is that every user has their own Rene AI assistant, and that two of them can text each other to find a slot that works for both calendars, with each person approving their own side. Gmail and Google Calendar access is what makes that negotiation possible: without a real diary on both sides, the exchange would be two agents guessing.
The booking pattern
The third pattern uses the browser rather than a connector. Asked to find dinner near a hotel, the assistant searches, checks the restaurant’s own site for availability, proposes a specific table and time, and books only after confirmation. The Rene AI assistant consistently stops at the point of commitment, which is the same shape as the calendar approval step.
The artifact pattern
The fourth Rene AI assistant pattern is creation. The homepage shows a website with a gallery built from generated images, then a matching slide deck in the same colours, both produced inside the chat. Rene’s Privacy Policy corroborates it by listing “generated outputs, summaries, reports, and other artifacts created through Rene” among the data the service holds.
Rene AI Assistant Checklist Before Connecting a Work Account
Most people will point a Rene AI assistant at a personal Gmail first, and that is a low-stakes decision. A work mailbox is not, and five questions are worth answering before anyone in a company connects one.
| Question | What to check | Where the answer is |
|---|---|---|
| Processors | Does your assessment cover Composio holding a token to the mailbox? | Privacy Policy, connected accounts |
| Inference location | Does your policy allow processing by OpenRouter, OpenAI or DeepSeek? | AI Provider List |
| International transfer | Processing may occur outside your country | Privacy Policy, international processing |
| Retention | Chat history and artifacts persist until deletion is requested | Privacy Policy, retention |
| Offboarding | Who deletes the account, and does that cancel the subscription? | Support page |
The subscription trap in the support page
One operational detail is easy to miss and annoying to discover late: deleting a Rene account does not automatically cancel an active App Store subscription. That has to be cancelled through the Apple Account settings separately. For an individual it is a nuisance; for a company reclaiming a departing employee’s tools it is a loose end.
Device permissions are separate from connectors
Connected apps are only half of the Rene AI assistant access surface. Depending on features enabled, Rene may also use camera, microphone, speech recognition, location, calendars, reminders, contacts, notifications and HealthKit data including steps, workouts, sleep, heart rate, active energy and body weight. The policy states these are optional unless a feature needs them, and they are managed in device settings rather than in the Rene AI assistant itself. Anyone who treats this as a data management question rather than a gadget question will ask the right things first.
Start where the blast radius is small
The sensible sequence is personal Gmail and personal Google Calendar first, for a week or two, with the approval prompts read rather than tapped through. Only once the behaviour is familiar does a work account become a reasonable question, and by then you will know which of the five rows above actually matters to you. Treat it the same way you would any other workflow automation rollout.
How the Rene AI Assistant Documents Line Up With the Launch
Reading the published dates together tells you how long the Rene AI assistant was in preparation before anyone outside the company could text it.
Rene’s Terms of Service carry a last-updated date of 10 June 2026. The Privacy Policy and the AI Provider List both carry 18 August 2026. The public launch was 16 September 2026. Counting forward, 10 June to 16 September 2026 is 98 days, and 18 August to 16 September is 29 days.
Each bar is that gap as a share of the longest, 98 days — so 29 divided by 98 is 30%.
What the dates imply
A Terms of Service written in June, with the Privacy Policy and provider list refreshed together in August, is the signature of a product that was in private use long before it was announced. That matches the founder’s own account of running the assistant in his texts for four months before launch.
What is still missing
No pricing page exists. The Terms describe paid plans, subscriptions, usage allowances, premium models and one-time credits; the homepage says “free to start, no credit card required”; no number is published anywhere. There is also no security page, no SOC 2 or ISO 27001 claim, and no data processing agreement on offer — all three of which a business buyer would normally want before connecting a corporate Google Workspace account.
Rene AI Assistant Risks and Limits the Terms Spell Out
The Rene AI assistant Terms are the most candid document on the site, and they set expectations that the marketing does not.
The accuracy disclaimer
The Terms state that AI outputs “can be incomplete, inaccurate, delayed, or inappropriate for a given situation” and that Rene is not a substitute for professional legal, financial, medical, safety or emergency advice. The user is responsible for checking outputs before sending messages, changing calendars or tasks, making purchases, relying on health information, or taking actions affecting other people.
Responsibility sits with the user
The Rene AI assistant instruction model is equally clear. Connecting a service authorises Rene to read, create, edit, send, delete, schedule or organise content there when a feature supports it, and you are responsible for reviewing important actions before asking for them, and for making sure your use follows each service’s own terms.
Availability is not guaranteed
Rene depends on third parties for authentication, payments, hosting, analytics, crash reporting, model routing and inference, notifications and the connected apps themselves, and disclaims responsibility for their availability or for changes they make to APIs, pricing or policies. A Rene AI assistant is therefore only as reliable as the weakest link in that chain on any given day.
Credits are not money
One line in the Rene AI assistant Terms is worth reading twice: usage allowances and credits “are not a bank account, stored value account, or currency”, and may be subject to product limits described in the app. That is boilerplate, but it is the boilerplate that decides what happens to unused credit.
Rene AI Assistant Questions, Answered
Does connecting Gmail give Rene my password?
No. The Rene AI assistant uses delegated access such as OAuth, so it never sees or stores the password for a connected service. Access is revoked by disconnecting the service in the app or by deleting the account.
Can Rene send email on my behalf?
Yes, when you ask the Rene AI assistant to and after it asks you to confirm. The FAQ says Rene will “ask before anything sensitive is sent or changed”, and the Terms describe sending a confirmed email as one of the user-facing features Google data may be used for.
Is my Gmail data used to train AI models?
No, on two separate Rene AI assistant commitments. Rene states that what you share is not used to train AI models, and the Google Limited Use section rules out using or transferring Google user data for training or improving generalised or foundation models.
Which AI companies process my messages?
OpenRouter routes Rene AI assistant requests, and OpenAI and DeepSeek run inference. Rene asks for consent before sending data to any external provider for the first time, and the consent can be withdrawn in Settings at the cost of the features that need it.
What happens to my data if I delete my account?
Deleting the account revokes connected Google authorisation and deletes the associated Google user data, subject to limited legal, security, fraud-prevention and backup-retention requirements. Deleting the account does not cancel an active App Store subscription, which must be cancelled through the Apple Account.
Does Rene work without Gmail and Google Calendar?
Yes. The Rene AI assistant lists Outlook as an alternative mail connector, and connecting nothing at all still leaves a text-based assistant that can browse, search and create. The Google connectors are what make the scheduling and triage features work.
Is there an app to install?
The agent itself is reached by text with no signup, but there is a Rene app for account settings, connected services and subscriptions, and a web app at app.rene.co.
References
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.