Untether Labs
AI AgentsBuilt in, not bolted on

Ask it. Hand it work. Approve the result.

Untether's AI agents answer operational questions from live data, prepare the change, and wait for your team to approve it. Built into the system your workforce runs on, not bolted on.

AI Agent

Where are we short next week?

One gap. Thursday 2 to 6 PM, behavioral health, short by 3.

Who can cover?
4 people are licensed, available, and under their hours.

Needs your approval

Offer Thursday 2 to 6 PM to 4 care team members

Approved. Shift offered.

Say it the way you would say it to a colleague

No report builder to learn and no ticket to file. The agent has the tools to look things up and to act, inside your permissions.

  • Where are we short next week?

    Answers from the live schedule and forecast, by team and hour.

    You decide what to do about it.

  • Sam called out. Who can cover?

    Ranks the eligible, licensed people and prepares the change.

    You approve before anything is saved.

  • Move Dev to the evening shift on Thursday.

    Checks the rules and stages the edit on the schedule.

    You review it and save.

  • Can Priya take the 14th off?

    Checks her balance, the policy, and coverage for that day.

    You approve or decline.

  • Show me overtime by team this month.

    Builds the report from your timesheets.

    You keep it, share it, or change it.

  • What is our fill rate?

    Answers in seconds, with the numbers behind it.

    No report request. No waiting.

How it fits togetherThree ways in, one system

AI is only as good as what it can see

Agents, workflows, and scheduling all run on the same rules and the same data. A tool that sees one piece can only send an alert. A model added to a spreadsheet can only guess.

Works with what you've built

Your systems

EHR, HRIS, payroll, calendar, and messaging. See integrations

Your own platform

A documented API and webhooks for the care platform or internal tools your team built. Read the API docs

Your AI tools

MCP is the standard way AI assistants connect to software. Untether supports it, so the AI tools your company adopts can work with your schedules, with the same permissions as a person.

Your data

It is yours. Export it through reports or pull it through the API.

You stay in control

The AI takes the repetitive work. Your team keeps the judgment. Untether is HIPAA compliant and SOC 2 audited. Visit the trust center

  • Every schedule is a draft until a person publishes it.
  • Every change an agent proposes shows a preview and waits for approval.
  • Agents act with the permissions of the person using them. They cannot do what that person cannot.
  • Every action is recorded in the audit log.

Build it or buy it

Your engineers can build a first version. The cost is keeping it right for years. They should build what makes your care model different.

What it takesBuild in-houseUntether
Getting a first version workingQuick to demo. The work is everything after.Live in days. Integrations set the pace.
License, labor, and pay rule changesYour engineers track and ship every change.Work rules, pay rules, overtime, and time off policy are settings.
Integrations when other systems changeYou maintain every connection.We maintain them: HRIS, payroll, EHR, calendar.
Edge casesEach one is a project.Employees and contractors, swaps, open shifts, on-call, and multiple licenses are handled today.
When payroll or coverage breaksYour engineers are on call.Our team is, alongside yours.
When the engineer who built it leavesThe knowledge leaves too.The product keeps shipping.
Mobile app for the care teamA second build, on two app stores.iPhone and Android, included.
In production with clinical teamsStarting now.Running every day at Dialogue, Iris Telehealth, Headspace, and Everly.

Build what is unique to your care model. Plug into Untether for the rest.

For engineeringSend this to your engineers
  • REST API with OAuth 2.0 client credentials
  • Signed webhooks for schedule, shift, and time off events
  • MCP for AI assistants and agents
  • Typed TypeScript client
  • Sandboxes for testing
  • Role-based access and an audit log

Developer docs

FAQs

No. Untether takes the repetitive work: building the first draft, finding who can cover, checking rules. Your schedulers review, adjust, and publish. They keep the judgment.

Nothing reaches the care team without a person. Schedules are drafts until published, agent changes wait for approval, and a published schedule can be unpublished.

No. Operations teams run Untether day to day. Engineers are only needed if you want to connect your own systems through the API.

Yes. Your tools read and write through the API, and webhooks tell them when a schedule, shift, or time off request changes.

No. Your agents connect to Untether through the API and MCP, with the same permissions as a person. They work with the schedule instead of rebuilding it.

Ask it something about your own schedule

We'll show you what Untether does and what your team approves.

Request a demo