Product
Everything your team hands to Renly.
Research, documents, scheduled automations, apps on your records, code and the systems you already run, in one workspace on the web, Windows and Android. You choose what needs your approval, and every response carries a sealed record.
Work and research
Ask for the outcome, not the steps.
Agents plan the work, use your connected systems and the web, and show each step as it runs. Big questions are split across parallel researchers, and every claim points to its source.
- Research mode: choose Research for a deep, cited report built by parallel researchers that read the sources in full.
- Web and internal sources together: public research alongside your wiki, files and CRM.
- Citations you can open, on the web page or the document they came from.
- Large inputs welcome: paste a long document or drop in files, spreadsheets and images.
Create
Finished work, not a draft in a chat.
Agents hand back real files and media, saved where your team works.
Documents, decks and sheets
PowerPoint decks with charts, Excel workbooks, Word documents and PDFs, saved to your folders, SharePoint or OneDrive.
Images and video
Images for campaigns and slides, and short videos you can trim, caption, voice over and add sound to.
Data and charts
Upload a spreadsheet or export; Renly queries it, runs Python in a sandbox with no network access, and charts what it finds.
Diagrams
Flowcharts, process maps and architecture diagrams from a description, editable, and embeddable in the wiki.
Posts and pages
Blog drafts in WordPress and LinkedIn posts, published only when a person approves.
Artefacts Beta
Publish a result to your workspace for the team to open, comment on and reuse.
Automations
Routine work, on a schedule.
Turn something you ask for every week into an automation that runs on time in your timezone, pauses when a step needs a person, and reports back.
- Three kinds: a Task (one agent on a schedule), a Workflow (several agents in sequence) and an Objective Beta (a goal tracked over weeks).
- Triggers: schedules, record changes, webhooks and API calls. Agents can also schedule their own follow-ups.
- In control: approvals mid-run, retry from the step that failed, a full run history, bulk approvals, and a step limited to exact actions.
Agents
An agent for every job, with limits you set.
Start from 97 ready-made agents across 22 industries, or describe a new one in plain language. Each agent has its own instructions, connectors and autonomy.
- Connector access per agent: inherit the organisation's, or only the connectors you pick, each read only or read and write.
- Autonomy per agent: analyse only (no tools at all), suggest only, act with approval, or act autonomously, never above the maximum your admins set.
- Action rules: action by action, decide what an agent may do for you: allowed, ask first, or never. An action set to never is not even offered to the agent.
- Agents that delegate: a lead agent can hand parts of a job to specialists and collect the results.
Knowledge and memory
It knows your business, and shows you what it remembers.
A wiki your team and agents keep up to date, search across your documents, and memory of preferences and decisions that you can read and manage.
- Wiki: pages, diagrams and exports to PDF, Word, HTML or Markdown.
- Memory you control: facts, preferences and decisions, each one visible, pinnable and deletable, for you, your project or the whole organisation.
- Earlier chats, if you want: turn on "Remember earlier chats" and agents can look back through your own conversations.
Apps
The tracker your team keeps in a spreadsheet, as an app with rules.
Describe a request tracker, asset register or CRM and an agent builds it on your Records data, or start from a template. Pages, roles and write rules publish together as one version.
- Rules on the server: a change that needs a person waits for one, in the app, the Approvals inbox, email or Teams. Agents follow the same rules.
- Your data, and theirs: text, numbers, dates, choices and links, plus live Dataverse, Salesforce, SharePoint Lists, Excel and SQL through each viewer's own connection.
- Every change on the record, with the app version and the approver. Archived records can be restored. On Pro Plus and Enterprise.
Approvals
See exactly what will happen, then decide.
When an action needs a person, Renly shows the precise change: the email as it will be sent, the fields as they will be written. What runs is exactly what you approved.
- Sending is always reviewed. Email, messages, posts, publishing and e-signature requests through a connector each need their own approval. No setting removes that.
- Edit before approving, or reject with a note so the agent tries again.
- Approve anywhere: in the chat, from the automation queue, in bulk, or on your phone.
Hi Priya, as promised, the revised pricing sheet is attached. The new rate applies from 1 November.
Evidence
A sealed record for every response.
Each response carries an Evidence Passport: who asked, which agent and model acted, which tools ran, what was approved and by whom, sealed into your organisation's tamper-evident chain.
- For one response or a whole chat, exportable as a document.
- Human review on the record: approved for reliance, reviewed with qualifications, or rejected.
- Open standards: Renly is implementing the Governed Runtime Standards, open formats that anyone can verify.
sha256:41ab…9f2cWindows, Android and other AI apps
Wherever the work is, Renly is there.
The same chats, agents and rules on every surface.
Windows app
Quick ask from any app with Ctrl+Alt+R, local folders, approved commands, a governed browser and computer use with your consent. From the Microsoft Store.
Get the Windows appAndroid app
Chats, approvals, notifications, to-dos and dictation on your phone. On Google Play.
Get the Android appInside other AI apps
Claude.ai, ChatGPT, Cursor and VS Code can use your Renly agents through MCP, with the same approvals and the same record.
Connect an AI appCode Beta
A coding agent, with the brakes you choose.
Code sessions read, edit and test code in a repository on your machine through the Windows app, or in your GitHub and Azure DevOps repositories.
- Five permission modes, from asking every time to full autonomy, set per session.
- Change sets you can review file by file, checkpoints before edits, and revert in one click.
- Sub-agents investigate in parallel, read only, and report back.
Permissions for this session
Connectors
22 connectors, and any MCP server.
Agents act in your systems through permissions your admins grant, and only the connectors your plan and your admins allow.
Admin
Run it for the whole company from one place.
Admins decide which connectors and models are available, how far agents may go on their own, and whether devices can be used, then see how AI is being used, what it costs and how often people step in.
- Limits: connector switches, a maximum agent autonomy, per-tool policy for MCP servers, and a kill switch for automations.
- Identity: Microsoft and work-email sign-in, and SCIM provisioning for joiners and leavers.
- Oversight: insights on actions, cost, approvals and failures, audit logs, and events streamed to your SIEM.
Questions, answered.
Which AI models does Renly use?
Renly runs models from Anthropic, OpenAI and Google, as well as Azure OpenAI and Microsoft Foundry deployments. On Enterprise you can bring your own models and keys.
What can an agent do without asking?
That is up to you. Each agent has an autonomy level (analyse only, suggest only, act with approval, or act autonomously), admins set a maximum for the organisation, and an automation step can be limited to exact actions. Sending email, posting messages, publishing and e-signature requests through a connector always wait for a person.
Can I use Renly from Claude.ai or ChatGPT?
Yes. Renly has an MCP server, so MCP clients such as Claude.ai, ChatGPT, Cursor and VS Code can use your Renly agents with the same approvals and the same evidence record.
Which features are in Beta?
Code sessions, Objectives and Artefacts are labelled Beta in the app. They are built and in use, and still changing quickly.
Where does my data live, and is it used for training?
Your data is isolated per organisation, encrypted, and never used to train models. Agents reach your other systems only through the permissions your admins grant. More in the security overview.
See it on your own work.
Start on the free plan with a work email, or book a demo built around your use case.