Workplace AI has spent a long time proving it can write the email, summarise the meeting and produce a slightly too-enthusiastic project plan. This week, the industry’s ambitions moved further down the org chart: AI is being positioned to coordinate work, solve IT problems and connect processes across an entire business.
That sounds like the productivity leap everyone has been waiting for. But this week’s stories also underline the awkward detail that determines whether agents become genuinely useful or just highly sophisticated chaos machines: they need context, permissions and someone accountable when things go sideways.
TL;DR
-
Meta is making a direct enterprise play with a platform spanning AI, business applications, infrastructure and security.
-
Omnissa is bringing agents into desktop and endpoint management, promising faster IT action with human guardrails.
-
JAAM Automation argues that companies must understand how work moves today before giving AI more responsibility tomorrow.
Meta Wants to Own More Than the Chat Window
Meta’s Enterprise Platform launch is a clear bid to become part of the machinery behind how businesses run — not simply another AI tool employees open alongside their existing stack.
The initial offering includes the security-focused Muse agent, Meta Business Agent, Muse API and Muse Code. Meta is also grouping AI, infrastructure, business applications and security under a dedicated enterprise organisation led by Chief Enterprise Platform Officer Chirantan “CJ” Desai.
The productivity story here is bigger than another copilot. Enterprises are increasingly looking for ways to connect models, agents and workflows without stitching together a dozen point products themselves. Meta wants to offer more of that foundation in one place.
That puts it into an increasingly crowded race. As UC Today explored in its look at Google’s push to bring AI agents into enterprise workflows, the battleground is moving from individual assistance towards orchestration: getting AI to work safely across the systems where work already happens.
Can IT Move at AI Speed Without Losing Control?
Omnissa’s latest AI announcements make the agentic productivity case in unusually practical terms. Horizon Delegate can perform approved actions inside a virtual desktop while the user is away. Its new DEX agent is designed to identify employee IT issues and recommend resolutions, while separate agents target vulnerability defence and the Windows application lifecycle.
This is not glamorous AI. It is arguably more valuable for that.
Few employees care whether their IT support process is “agentic”. They care whether the software works, access is restored and a recurring issue is fixed before it eats another hour of their day. For IT teams, automation could also reduce the backlog of patching, packaging and troubleshooting work that rarely makes anyone’s strategic roadmap.
But Omnissa’s story raises the right question: when agents can act faster than people can review, where does oversight sit? The company’s answer is governance and human-in-the-loop controls — and it needs to be. As UC Today has noted, AI agents create an observability problem as well as an automation opportunity.
The aim should not be autonomy for its own sake. It should be faster resolution with a clear audit trail, sensible permissions and a human able to intervene before “AI speed” becomes “AI-sized incident.”
The Lesson: Automation Cannot Fix a Workflow Nobody Understands
JAAM Automation’s new platform, M. by jaam, takes a more grounded route into agentic work. Its pitch is not that businesses should hand everything to AI immediately. It is that they should first create a working understanding of how processes move between people, systems, knowledge and decisions.
Andrew Murphy, JAAM’s Chief Strategy Officer and Co-Founder, puts it neatly: “If we’re going to ask AI to take on more of that work, it needs to understand how the business actually works.”
That should be the lesson of the week.
Most business processes do not run cleanly through one platform. They are held together by approvals in Teams, details in spreadsheets, exceptions known by one experienced employee and a great deal of “that’s just how we do it.” Automating that without context risks making a broken process run faster — right up until it breaks somewhere more consequential.
JAAM’s gradual approach, moving from analysis and assistance towards greater automation, is less headline-grabbing than an autonomous-agent promise. It is also closer to how responsible adoption will work in most enterprises.
Productivity leaders should start with the workflows most likely to generate measurable ROI: high-volume tasks, repeatable decisions and processes slowed by manual coordination. Then they should give agents responsibility progressively, rather than mistaking an impressive demo for an operating model.
The shift is clear. AI is no longer just helping people complete work. It is being asked to help run it. The businesses that benefit will be the ones that understand their workflows well enough to know where an agent belongs — and where it absolutely does not.
