Session 3 — Human + Agent Workflows: Who Decides, Reviews and Escalates?
Are you a COO, CHRO or operations leader deciding how employees and AI agents work together? This session examines where agents can act independently, where human judgement remains essential and how accountability is maintained.
Session 3 of 4 in The Agentic Enterprise, a WORK-SELF webinar series on turning agentic deployments into measurable ROI.
Thursday, 19 November 2026 · 2:00 PM GMT · 9:00 AM EST · 30 minutes
As AI agents move from generating output to taking action inside enterprise workflows, a new operating question appears: who remains responsible for the outcome?
Organisations need clear decision rights, review points, handoffs and escalation rules. Human judgement cannot simply be added after an agent has been deployed. It has to be designed into the workflow from the beginning.
This session explores how enterprises can structure human and agent responsibilities as autonomy increases.
Designing Accountability Into Agentic Workflows
As AI agents take on more responsibility, enterprises need to establish which decisions remain with employees, what agents can complete independently and when work must be reviewed or escalated.
These responsibilities cannot be managed through automation rules alone. Agents require Human Context that reflects employee authority, task ownership, review requirements and organisational accountability.
Human Context Infrastructure provides a governed foundation for coordinating these responsibilities, helping organisations introduce greater agent autonomy without losing human oversight or creating additional manual rework.
WORK-SELF’s Maya Enterprise delivers this operating context to approved agents within the enterprise workflows they already support.
What we’ll cover
Where human judgement should remain inside increasingly autonomous workflows
How to define decision rights, ownership and review requirements
How approvals, handoffs and escalation should work between employees and agents
How governed human-agent workflows can scale without creating another layer of manual review
About the speakers
Wolf Magdelinic
Founder and CEO, WORK-SELF
Wolf has more than 20 years of experience building AI-assisted workflow automation and enterprise technology. He founded and exited Overbond AI, led an applied machine learning research lab involving more than 30 PhD researchers, and is the inventor of a granted US AI patent.
Previously at PwC, Deloitte and CIBC, he holds an Electrical Engineering degree from the University of Toronto and an MBA from Ivey Business School. His work has been featured in The Economist, Reuters, Bloomberg, Forbes and the Financial Times.
Wolf leads the development of WORK-SELF’s Human Context Infrastructure and is co-authoring The Human Feedback Loop: Agentic Enterprise Transformation with Prof. John O’Dwyer.
Who it’s for
CIOs, COOs, CAIOs, operations leaders, AI governance leaders, risk leaders and executives designing new operating models around AI.
About WORK-SELF
WORK-SELF builds Human Context Infrastructure for the agentic enterprise.
Maya Enterprise (a WORK-SELF offering) brings authorised employee, role, task, policy and workflow information into a governed Human Context layer.
Through employee-specific Work Contracts, permissioned Context Capsules and Model Context Protocol (MCP) interoperability, Maya provides approved AI agents with the context required to respect decision authority, human oversight, escalation rules and accountability.
Operating alongside existing enterprise applications and AI systems, Maya helps organisations coordinate human and agent workflows, reduce employee rework and measure the value of completed work.
Part of The Agentic Enterprise, a four-part WORK-SELF webinar series.