Do I need Claude or ChatGPT to use the agents?

No. The agents run inside Juno, with the same tools, knowledge and actions as the Juno MCP already set up: no connector to install and no prompt to write. If your team already runs Juno from Claude, ChatGPT or Copilot, the same actions are there, and the Juno MCP page shows how.

Does this replace my team?

No. No agent decides what matters, what goes first, who is excused or the human moments; those stay with your people. The agents take the drafting, the assembling, the searching and the counting, and each one hands a named decision back: the manager approves the spend, a person confirms a grade, you approve what goes out.

Which agents run today?

Jaden, Jasmine, Jemma and Jeremy are live. Jensen, who measures, is emerging: it works, and it is still early. Jamie, who coaches in the moment of work, is on the roadmap and is not a live capability. Each agent's status is printed beside its name on this page, and it changes here when it changes in the product.

What happens when one gets it wrong?

It stops at the approval. Nothing an agent prepares publishes, assigns or notifies before a person says yes, so a wrong draft is corrected or turned down there. A new automation is created paused until someone enables it, and an open question can be graded by a person, or graded by the AI and confirmed by a reviewer.

Can I add checkpoints?

Yes. Checkpoints are placed per phase when the journey is configured: several, one or none, wherever a person should look before the next step runs. Whatever you place, what an agent prepares still waits for a person before it publishes, assigns or notifies, so a checkpoint adds a review and never removes one.

An agent works within what the person using it is allowed to do. Which capabilities are available depends on your plan and your admin's settings, and tools you aren't permitted to use don't appear to your assistant. Your security team can review the certifications, sub-processors and policies in the Trust Center.

Ready-made AI agents, each with its own job, inside Juno.

Each agent has a role and the same tools, knowledge and actions as the Juno MCP: it builds content, finds answers and courses, plans development, checks what people can do and measures what changed. No connector to set up, no prompt to write, and nothing reaches a person before someone on your side says yes.

Book a demo

4.7 / 5G2

1Trigger & Signal

A request typed in Juno
A document you import
A change in your HRISa new hire, a promotion
An event in Junoa due date, a failed quiz

2Juno, the Super Agent

Juno

Jaden builds and ships
Jasmine finds and funds
Jemma plans
Jeremy validates
Jensen measures

Human approval

3Delivery & Adoption

A course or a lesson
An answer
A plan
A budget request
A check
A report

4Measure

What needs a decision comes back to a named person

Not every team has an LLM wired to its systems yet.

You know AI can build the course, find the answer and pull the report, but nobody has the week it takes to set up connectors and write the prompts to get there.

One request, four agents, one approval

Juno

Dana moves to Sales next month. Get her ready, find a negotiation course we can pay for, and tell me at day 60 if she's selling.

Jemma drafted her plan, Jaden her Sales ramp-up path, Jasmine two negotiation courses and a budget request, Jeremy her scenarios to grade. One approval:

Draft · I'll wait for your OK

I'll assign "Sales ramp-up path" to 1 person.

To1 learner · DanaStepsAll steps · from move dateDueJan 4, 2027 · day 60

Assign to 1 personCancel1 matching now

DraftDana's plan, by Jemma · Budget request, negotiation course

Example data

Six agents, six jobs, one team

Jaden
live
builds and ships
What it's handed
a document or a release
What it produces
a course, a journey, an audience
The decision that stays yours
you approve before it goes out
Jasmine
live
finds and funds
What it's handed
a question or a course request
What it produces
the answer, the course, the budget request
The decision that stays yours
the manager approves the spend
Jemma
live
plans
What it's handed
a goal or a new role
What it produces
a plan with steps, checkpoints, a mentor
The decision that stays yours
the employee and the manager agree it
Jeremy
live
validates
What it's handed
a scenario, an open question
What it produces
a grade and feedback
The decision that stays yours
a person confirms
Jensen
emerging
measures
What it's handed
a question the dashboard wasn't built for
What it produces
the answer, by person and team
Jamie
roadmap
coaches in the moment of work

Where they work

Over 40% of agentic-AI projects will be abandoned by 2027, for lack of results

Gartner

So each agent has one job, a status and a person who decides.

The tools, the context and the actions, already set up

Juno's own tools and actions
The same the Juno MCP exposes: find content, build it, assign it, remind, report.
Your content
What already sits in Juno is found and reused before anything new is built.
Your company context
Your people, teams, roles and user attributes, read from the platform you run.
The triggers inside Juno
Not started · not completed · a set number of days before the due date · a failed quiz · a changed user attribute, which is how an HRIS change reaches them.
What you import
A document, a file, an export: the agents start from what you bring.
Already run Juno from Claude or ChatGPT? The same actions are there →
Every prompt →

What no agent decides

What matters
Which skills and outcomes count in your company is your people's call.
What goes first
An agent drafts; a person decides which change, team or request comes first.
Who is excused
An exemption or an exception is decided by a person, never by an agent.
The human moments
The welcome, the feedback, the hard conversation stay between people.
The decision, named
Publishing, assigning and notifying wait for a person to say yes.
The job title, named
The manager approves the spend; the employee and the manager agree the plan.

Tools you aren't permitted to use don't appear to your assistant. Trust Center

Who conducts

Juno, the Super Agent

The one place to ask, every agent behind it.

Questions people ask.

Do I need Claude or ChatGPT to use the agents?: No. The agents run inside Juno, with the same tools, knowledge and actions as the Juno MCP already set up: no connector to install and no prompt to write. If your team already runs Juno from Claude, ChatGPT or Copilot, the same actions are there, and the Juno MCP page shows how.
Does this replace my team?: No. No agent decides what matters, what goes first, who is excused or the human moments; those stay with your people. The agents take the drafting, the assembling, the searching and the counting, and each one hands a named decision back: the manager approves the spend, a person confirms a grade, you approve what goes out.
Which agents run today?: Jaden, Jasmine, Jemma and Jeremy are live. Jensen, who measures, is emerging: it works, and it is still early. Jamie, who coaches in the moment of work, is on the roadmap and is not a live capability. Each agent's status is printed beside its name on this page, and it changes here when it changes in the product.
What happens when one gets it wrong?: It stops at the approval. Nothing an agent prepares publishes, assigns or notifies before a person says yes, so a wrong draft is corrected or turned down there. A new automation is created paused until someone enables it, and an open question can be graded by a person, or graded by the AI and confirmed by a reviewer.
Can I add checkpoints?: Yes. Checkpoints are placed per phase when the journey is configured: several, one or none, wherever a person should look before the next step runs. Whatever you place, what an agent prepares still waits for a person before it publishes, assigns or notifies, so a checkpoint adds a review and never removes one.
Who can see what?: An agent works within what the person using it is allowed to do. Which capabilities are available depends on your plan and your admin's settings, and tools you aren't permitted to use don't appear to your assistant. Your security team can review the certifications, sub-processors and policies in the Trust Center.
How much do the agents cost?: Pricing is published on the Pricing page.

Ready-made AI agents, each with its own job, inside Juno.

Not every team has an LLM wired to its systems yet.

One request, four agents, one approval

Six agents, six jobs, one team

Jaden

Jasmine

Jemma

Jeremy

Jensen

Jamie

The tools, the context and the actions, already set up

Juno's own tools and actions

Your content

Your company context

The triggers inside Juno

What you import

Already run Juno from Claude or ChatGPT? The same actions are there →

What no agent decides

What matters

What goes first

Who is excused

The human moments

The decision, named

The job title, named