Exclusive: UiPath CMO Michael Atalla on AI at work
UiPath CMO Michael Atalla discusses the company's evolution from task automation to 'agentic business orchestration' on its five-year IPO anniversary. He argues that most AI initiatives fail due to coordination problems rather than technology shortcomings, and that while job anxiety is legitimate, human judgment remains essential in AI-augmented workflows.
Summary
In this interview, UiPath CMO Michael Atalla reflects on the company's five-year journey since its IPO, tracing its evolution from robotic process automation to what it now calls 'agentic business orchestration' — coordinating AI agents, automation, and humans across end-to-end workflows. Atalla draws on his 15 years at Microsoft, where he led Office 365's transition from on-premise to cloud, to argue that the core challenge in AI adoption is not the technology itself but the failure to redesign workflows around it. He warns that companies are repeating the 'lift and shift' mistake from the cloud era.
On AI pilot failures, Atalla identifies coordination as the root cause of the widely cited 70-80% failure rate. He explains that pilots typically run in isolation — one agent here, one automation there — with no connection to broader business goals, causing ROI to evaporate. He argues that successful organizations treat AI agents as components of a governed workflow system rather than standalone tools.
Regarding AI's impact on jobs, Atalla acknowledges that entry-level role anxiety is legitimate and data-backed, citing a nearly 20% drop in entry-level dev jobs since 2024. However, he pushes back on the notion that human involvement becomes optional, emphasizing that AI cannot exercise judgment, taste, or moral reasoning — captured in his framing that an LLM cannot ask 'should we?' He argues roles are changing shape rather than disappearing, with new demand emerging in workflow design and AI governance.
Atalla also outlines UiPath's internal deployment philosophy: automation handles structured, repeatable tasks; agents handle ambiguity like non-standard invoice interpretation; and humans retain ownership of decisions with real accountability. He is skeptical of 'full autonomy' narratives, arguing the near-term reality is agents operating inside governed, observable workflows with more cognitive responsibility but not unchecked independence.
Key Insights
- Atalla argues that the primary reason 70-80% of agentic AI initiatives never leave the pilot stage is a coordination failure — agents and automations run in isolation with no connection to broader business goals, causing ROI to disappear in the gap.
- Drawing on his Microsoft experience, Atalla claims enterprises are repeating the cloud-era 'lift and shift' mistake with AI — deploying tools without redesigning the underlying workflows, which he identifies as the core reason AI investments underdeliver.
- Atalla contends that AI agents are well-suited for handling unstructured data and context-aware decisions within defined processes, but that deterministic rules-based work still runs better on traditional automation, and decisions with real accountability must stay with humans.
- Atalla asserts that an LLM fundamentally cannot ask 'should we?' — it has no motivation, taste, or instinct for risk — and uses this to argue that human involvement does not become optional as AI becomes more capable, even as the nature of roles changes.
- Atalla claims that framing AI as something happening to employees rather than something they will build with is where workplace anxiety originates, and argues this framing mistake is largely avoidable through deliberate communication choices by leadership.
Topics
Full transcript available for MurmurCast members
Sign Up to Access