Platform · Agents
AI agents that run your programs around the clock.
Matching, nudges, prep, check-ins and follow-ups, handled by agents that know your organization and your people. Program leaders spend their time on people, not admin.
24/7
Always on
Across every program
Every
Product
Coaching, Mentoring, Communities
Context
Aware
Grounded in roles, goals and skills
Approvals
Built in
People stay in control
Agents and orchestration
Orchestration sets the rules. Agents do the work. People make the calls.
One system, three layers. Here’s how they work together when an employee is promoted.
Rules
Orchestration
Set what happens after a promotion, hire or role change, who must approve and what gets logged. Same trigger, same steps, every time.
Priya is promoted. Start the new-manager playbook, require HRBP sign-off on her mentor match, check in at day 21.
Designed by HR · governed by ITOrchestration →
ExecutionYou are here
Agents
Carry out each step for each person: matching, nudges, prep, check-ins and follow-ups, adapted to their role and goals.
Her first 1:1s are Thursday, so the AI coach offers to rehearse the hardest one. The agent matches her with Alicia, who managed that team before, and preps them both for their first meeting.
Scoped access · every action logged
Calls
People
Approve, mentor and manage. Agents bring the context; decisions about people stay with people.
Her HRBP approves Alicia as mentor. Alicia and Priya meet. Her manager joins the day-21 check-in.
Agents assist · people decide
For CHROs
Big career moments happen to all your people. Support only reaches a few.
A promotion, a new job, a new team. These are the moments people need support most, and most get none. Programs have room for a select few and run on their own calendar, not the employee’s.
Your people, not just a cohort
An AI coach supports your whole workforce without adding program headcount.
Help at the moment of need
Support arrives the week someone is promoted, hired or changes teams, not at the next intake.
Measured by readiness, not attendance
See how ready your people are as a group, not just who showed up to the session.
- Leadership cohort: 40 seats
- Next intake in March
- Progress = attendance
- Every new manager, the week they’re promoted
- Every new hire, from day one
- Progress = readiness
The agents
One agent layer. Every Rumi product.
Some talk with your people in Slack, Teams and email. Others work in the background. Each has a defined job and a defined scope.
Talks with people
Coaching agent
Reaches out before reviews, after promotions and ahead of hard conversations.
Works outWhen to reach out and what to rehearse.
Coaching
Talks with people
Prep agent
Builds the agenda for 1:1s, mentoring sessions and reviews.
Works outWhat belongs on the agenda this time.
Mentoring
Talks with people
Practice partner
Runs a roleplay of the conversation before it counts.
Works outWhat to practice and where to improve.
Coaching
Talks with people
Reflection agent
Debriefs after big moments and turns them into next steps.
Works outWhat’s worth carrying forward.
Coaching · Mentoring
Works in the background
Matching agent
Invites, enrolls and matches, then re-matches when people move on.
Works outWho fits and why.
Mentoring · Communities
Works in the background
Nudge agent
Keeps pairs, circles and goals moving.
Works outWhen a nudge helps and when it’s noise.
All three products
Works in the background
Program health agent
Watches pairs, circles and cohorts for signs of drift.
Works outWhich pairs need attention first.
Mentoring · Communities
Works in the background
Wrap-up agent
Closes each cycle with results leaders can use.
Works outWhich results matter to report.
All three products
Reasoning you can see
Every recommendation shows its reasons.
Agents work from your people data, show the reasons behind every suggestion and leave decisions about people to people.
Grounded in your data
Role, team, goals, skills and relationships from your HRIS, LMS and Rumi.
Reasons, not a black box
Admins and employees see why a mentor or next step was suggested, and what was ruled out.
Knows when to bring in a person
When a moment needs lived experience, agents bring in a mentor or one of your coaches instead of answering themselves.
For CTOs and security teams
Bounded by design.
Every agent has an owner, a defined scope and a full action log. You decide what runs on its own and what waits for a person.
- Owner
- Mentoring program admin
- Reads
- Profile, goals, skills, mentoring history
- Never reads
- Private notes, compensation
- Can do
- Propose matches, send intros
- Needs approval
- New matches, messages to executives
- Last 24 hours
- 9 matches proposed · 0 errors
Scoped access
Each agent reads only the fields its job needs. Private notes stay private.
Approvals you set
Choose which actions run on their own and which wait for sign-off.
Every action logged
What ran, when, why and which records it touched.
Pause any agent
Stop one agent without stopping the program.
Your data stays yours
Your data never trains our models.
Questions buyers ask.
- How are agents different from orchestration?
- Orchestration sets the rules: what happens, who approves, what’s logged. Agents do the work inside those rules: matching, nudges, prep, check-ins and follow-ups for each person.
- Do agents act without approval?
- Only where you allow it. You choose which actions need a person to sign off.
- What data can agents see?
- Only the fields you connect and scope to them. Private notes stay private, and fields you exclude, such as compensation, stay out of reach.
- Do agents replace mentors, coaches or HR?
- No. Agents handle preparation and follow-through, and bring in a person when experience matters.
One platform for AI coaching, mentoring, and communities.
See Rumi coach an employee, match a mentor and report results, all on one platform.