Ryan Johnson @RyanWorkforceAI

Great post. This is why measurement matters so much in AI adoption; ateam can be “using AI” heavily and still not know whether work is actually improving. For workforce and learning teams, the useful questions are pretty practical: are people making better decisions, reducing rework, catching errors earlier, and learning from the system over time?

Learning does not always happen where the learning system lives. When people need help, they often go to: • a manager • a colleague • a chat thread • a shared doc • whatever source feels fastest That matters for AI. If we want AI agents to support learning at work, they cannot just sit on top of the LMS. They need to meet people closer to the moment of need. The practical question for L&D and HR teams is not only: “Do we have good learning content?” It is: “Can people find the right guidance when the work is actually happening?” That is where workforce learning gets more useful. #LearningAndDevelopment #AgenticAI

@morganlinton And then there is Ultracode 😱

@thsottiaux Couldn't imagine a more amazing place to work!

@toddsaunders That is actually pretty amazing - this is an elite example how agents are transforming work.

AI agents do not become useful just because they can take action. They become useful when teams define the operating rules: • What can the agent do on its own? • What knowledge can it trust? • What still needs human review? • When should work stop and escalate? Failing to define these is where many pilots get stuck. The tool works. The workflow looks promising. But no one has clearly defined the boundaries. For L&D, HR, and enterprise teams, the practical goal is: Turn AI agents from a smart demo into safe, usable work partners. The goal is not more automation. The goal is better decisions at work. #AgenticAI #WorkforceAI #AIGovernance

@danshipper Thanks Dan, been a fan for years and this is good stuff!

@adocomplete @Cloudflare Been working with Fable 5 on Cowork, and this has been my experience too!

I like the flywheel framing, especially because it makes the hard part visible. The autonomous enterprise is not just AI taking action its the org getting better at capturing judgment and flipping it into usable processes... learning from each loop without losing human accountability.

@emollick Just seeing this post, but it is a real issue for workforce AI adoption. If teams cannot trust the measurement, they cannot tell whether AI is improving work or just creating better-looking reports...

@alliekmiller Great post Allie. This is exactly where workforce teams need to pay attention. AI does not just change tasks, but how trust, ownership, and judgment work inside the organization. Imo the org design question is becoming a learning question.

@morganlinton It is pretty amazing - been doing the same. Have had issues with route card/pheromone system and canonical claim promotion with my research pipeline, and it is not only fixing, its actively improving... Fable 5 on Ultracode is insanely good.

Great points in this post- this gap matters a lot for workforce teams. Buying AI is not the same as changing how work gets done. The hard part is redesigning: • workflows • decision rights • handoffs • review points • skills • accountability That is where adoption either becomes real or stays cosmetic.

Just seeing this post. Very important. This is where HR and L&D teams need to be especially careful. When AI agents touch people data, the issue is not only accuracy. Teams need clear rules for: • what the agent can access • what it can recommend • what needs human review • how mistakes are caught • who owns the decision Basically, trust, identity has to be designed before anything else - along with the data layer, you could argie its the #1 thing enterprise teams should focus on.

This great stuff @github, and also a useful framing for workforce learning. AI does not just change the task. It changes how people build skill around the task: • what they practice • what they review • what they trust • when they ask for help • how they know the work is good That is the real upskilling challenge.

Great post. This is where agentic AI gets practical. Imo the #1 rule: Agents are only as useful as the data and workflows underneath them. For workforce teams, the same lesson applies: • connect trusted knowledge • define ownership • reduce handoffs • measure whether work improves The operating model matters as much as the tool.

This is a really important findings for enterprise teams. Well done @AnthropicAI team. As AI agents move closer to real workflows, security is no longer just a technical control. People need to understand: • What the agent can access • What instructions may be risky • When outputs need review • When to stop and escalate That is a workforce learning problem too.

One pattern I’m watching in workforce learning: Agentic AI does not remove the need for skill development. It changes what skill development has to cover. It is no longer enough to teach people how to use a tool. Teams now need judgment around: • What should the agent do? • What should a human still review? • Which sources should the agent trust? • When is the output good enough? • When should work stop and escalate? That is the shift. The future of agentic AI upskilling is not just tool training. It is helping people make better decisions inside agent-assisted work. #WorkforceAI #LearningAndDevelopment #AgenticAI