Built for your question · 4 sections from 2 pages

Back to original
1 · From Run AI agents safely (opens in a new tab)

Responsible AI

Use AI agents with the controls regulators expect

Decide what each agent can read and do, require approval for every send and keep a full audit trail.

Agent Hub's agents: 1,942 actions today and 12 waiting for review, beside an approval request for a Campaign Agent send.
2 · From Hourglass for financial services (opens in a new tab)

Customer stories

How banks, credit unions and insurers grow with Hourglass

Regulated teams use Hourglass to serve members and clients faster, with every message reviewed.

Illustrative stories for this demo.

See customer stories

“Our branch and contact center teams used to switch between four systems to answer one question. Now the Customer Agent drafts the reply from the customer’s record, and our staff approve it before it goes out.”

Priya Raman

VP of Customer Experience, Mercantile Bank

  • 38%

    faster replies to customer questions

  • 2x

    more customers enrolled in online banking

3 · From Hourglass for financial services (opens in a new tab)

Financial services

Built for regulated teams

Hourglass gives relationship, marketing and service teams room to move, with the controls your compliance team asks for.

A compliance review queue: campaigns and outreach waiting for approval, replies about money sent to a person, and every decision logged.

Approvals where you need them

Add review steps to campaigns, outreach and agent replies, and keep a full history of who approved what and when.

  • Review steps for campaigns, outreach and replies
  • Every decision logged with who and when
  • Ready when your auditors ask
Learn more about security and control
A member questions card: six months of questions answered from approved content and passed to a person.

The service members expect

Answer routine questions from approved content in chat and email, and pass anything about money or accounts to a person.

  • Answers from approved content only
  • Money and account questions go to people
  • Chat and email in one queue
Learn more about the Customer Agent
4 · From Run AI agents safely (opens in a new tab)

The governance playbook

Four controls for every agent

Set them once in Agent Hub, and change them as trust grows.

  1. Set permissions

    Choose the data each agent can read. Limit actions by team and role.

  2. Require approval

    Approval steps for messages and record changes. Different rules for different risks.

  3. Review every action

    An activity log with the data each action used. Exports for your audit team.

  4. Pause any agent

    Switch an agent off in one click. Change its scope without starting over.

Agent settings for the Prospecting Agent: research allowed, drafts allowed and every send waiting for a person's approval, with a Pause agent button.