We demoed our AI agents live during the first of five Degreed in Action webinars and the chat filled up within minutes. The questions that came through were practical. Where do the agents run, and who stays in control? What happens to the content your team spent years building?
L&D leaders asking those questions are past the “should we” stage and planning how agents fit into their week. These are the answers they wanted to hear most.
How Degreed Agents Work
Degreed agents have three parts. Context is what Degreed knows about each person: their skills, role, and development history. Tools, including MCP, let people and agents take action inside the systems your workforce already uses. Agents use that context and those tools to do specific jobs, like drafting a pathway or cleaning up a catalog.
Every interaction adds to the context, so each agent starts the next conversation with a sharper picture of the person it’s helping.
Where They Run and Who’s in Charge
The first wave of questions came from admins, and they asked two things a few different ways.
Is this another tool to roll out? And what can it do without us?
1. Agents Work in the Tools Your Employees Use
“Are these [agent] capabilities available through Maestro only?”
Degreed agents connect through MCP to the AI tools your organization runs today, including Microsoft Copilot, Claude, and Gemini. Degreed customers are also able to use agents within Maestro. Employees and admins get help in the same window they open every morning, and IT secures one connection instead of a whole new platform.
2. A Human Always Approves
Agents don’t make changes on their own. Any write action waits for a person on your team to approve it, so the agent handles the legwork and you keep the final call.
3. Admins Set the Rules Agents Follow
“What controls or settings would organizational admins have when setting up these agents?”
Admins can set the guidelines the learning agents work within, like which sources come first and how long content should run, so organizations can help define agent behavior and standards. Agents also use existing Degreed configurations, like skill or role standards, endorsed content, or a skill taxonomy, so Degreed admins can set it once within Degreed and agents will have that context. You set it once and every agent applies it, so a pathway built by an HRBP, manager, or individual learner follows the same standards as one your learning team may build.
Building Learning That Fits Your Organization
The next set of questions came from people who are responsible for building learning pathways. They wanted to know if an agent would also build them to organizational standards.
4. Your Own Content Can Come First
You can tell an agent which sources and providers to prioritize. Based on your preferences, your organization’s content can show up first, which means your workforce gets to the material that matters to your business faster.
5. Check What You Have Before Building New
The Instructional Designer agent looks for existing pathways before it drafts a new one. It helps keep duplicates out of your catalog, and your team’s past work gets reused.
6. Agents Go Beyond Today’s Pathway Curation
“Will the Instructional Designer agent produce similar results to the current Maestro Quick Action for curating a pathway?”
The current pathway curation in Ask Maestro is workflow-based. Our new agents are more flexible, and add the ability to review uploaded files, connect to skill standards, and assign directly to groups.
7. Agents Work Before Your Skills Data Is Perfect
“How will this agent work if your organization doesn’t have skill targets set up?”
Agents do their best work with skill targets in place. If yours are still in progress, agents use role standards or ask you where you want your team to land on each skill. You can start now and sharpen the data as you go.
8. Managers Can Build Learning for Their Own Teams
Learning teams can’t hand-build a plan for every team in the business. The Team Transformation agent lets managers and business leaders build learning for their own people, based on that team’s actual skill gaps. A manager can go from gap to assigned pathway, with sections, tags, and a due date, in one conversation in Copilot. Because the agent follows the standards your learning team already set, spreading the work out doesn’t cost you consistency. L&D sets the guardrails and the business moves at its own pace.
Less Catalog Upkeep, More Strategic Work
Catalog maintenance can eat up budget in a way that most teams never plan for. One enterprise customer told us they spend hundreds of thousands of dollars a year reviewing their learning content by hand. That’s budget and time their team could put toward strategic work.
9. Finding Broken and Outdated Content Across Your Catalog
“Does the Content Opt agent read the third-party content providers too?”
The Content Optimization agent scans your catalog for broken links and obsolete content, including connected third-party content it can reach. Content behind SSO can’t be checked yet. When a broken link sits inside a pathway or plan, the agent can remove it or suggest a replacement, and any change waits for your approval.
10. “Obsolete” Has a Clear Definition
“How is it defining obsolete?”
The agent flags content by engagement and age. An asset with almost no learner engagement or followers can become a candidate, and so can an old tutorial when a newer one exists in the same catalog. Flags are candidates, not decisions, and a person approves any change before it happens.
Overall, Control Came Up in Every Question
Leaders want agents to take on curation and cleanup while their teams keep authority over standards and final decisions. We built our agents to work exactly that way.
Your team gets back the hours it spends building and auditing by hand, and puts them toward the transformation work the business needs to keep up with the rate of change.
Vamos manter contato.
Quero assinar a newsletter mensal que oferece insights exclusivos, anúncio de eventos e novidades sobre as soluções da Degreed.