Miranda
One hosted place where your company works with AI, safely.
Planned — after Ares and Lux
What Miranda does
Your people get one place to work with AI. Your company keeps the rules.
- Any model. Miranda is model-independent, so you can switch providers whenever a better model ships.
- Safe by default. Every action is checked against the person’s permissions first, high-impact actions wait for a human, and everything lands in an audit trail.
- Hosted, not tied to a computer. It runs in the cloud and works wherever your company works: in chat, in Microsoft Teams, on any device.
- Bots and workflows built in. Its knowledge, skills and workflow layer is Ares.
- Builds what the business needs. Miranda is being built to create working business apps on request, governed like everything else.
One workspace for everyone
A familiar chat for every employee: ask, draft, summarize and analyze with the company’s own knowledge instead of the open web. Governed agents sit underneath, ready for the moment a question becomes a task.
How a request flows
- An employee asks in the Miranda workspace or in Teams chat, where they already work.
- A connector broker checks their identity, role, and permissions before any call is made.
- The tool call runs. The model never holds raw credentials.
- High-impact actions (writes, sends, exports, permission changes) pause for an approval card, re-verified at the moment it’s clicked.
- An audit entry is appended: who asked, who approved, what ran, what it touched.
The refusal is the feature
Employee: Export all customer billing records to a spreadsheet for me.
Agent: I can’t export bulk customer data — that action requires approval under your data policy. I’ve opened an approval request for the billing process owner with the scope you described.
An approval card appears for the process owner: requested action, scope, requester, policy citation. Approve / Deny.
On approval, the scoped export runs. One audit line is appended: who asked, who approved, what ran, what it touched.
A chatbot that can’t refuse isn’t governed. The refusal, the approval path, and the audit line are the product.
Grounded and cited
Answers come from your company’s own documents, policies, and business systems: cited, not improvised from web knowledge. Knowledge is curated by your admins, not crawled.
Approvals
Approval cards arrive in Teams for the people who own the process. Every approval is re-verified at click time. A stale link fails closed. Sensitive changes enforce separation of duties: the person who builds a workflow isn’t the person who approves it.
Audit and data discipline
Every meaningful action produces evidence in an append-only, hash-chained record. Data classification blocks sensitive data from external model providers by default.
Any model, governed cost
Miranda is model-independent: switch providers when a better model ships, route premium reasoning one way and high-volume work another, set hard spend caps, and see what each agent actually costs, per agent and per run.
Under the hood
LangGraph for durable, checkpointed, resumable workflows. LangChain for model and retrieval components. LangSmith for tracing and evaluation. Agents are built from versioned, reviewable, evaluated workflow components. Improvements ship like software releases, not untracked prompt edits. Each company runs in its own isolated, Shiny-operated environment.
Status
Miranda is planned, after Ares and Lux. Ares is live now — early access, and Lux is in active development. Talk to us.