Work reworked Human Agentic AI
C · Weak signal Short Deloitte Global thought-leadership piece on the 'Human-Agentic Workforce'; no data, and its company examples are described without figures or sources.
Summary based on the full text·Original: about 12 minutes to read
In short
A 14-page Deloitte Global paper (2025) by its future-of-work and agentic AI leaders. It argues that adding agents to existing workflows reproduces old inefficiencies at scale, and that firms should redesign work, roles and governance around humans and agents working together. It proposes five shifts, a "systemic codependency" between functions, and a dual-speed approach with a fast lane for experiments and a stable lane for guardrails.
Key numbers
- 10xproductivity leap the executive summary invites readers to imagine; no evidence is offered for it
- 5key shifts proposed, from technology-led to human-centric transformation through to organisational readiness
- 2015year Lemonade launched with AI agents built into its operating model, the paper's main case
As reported by the source.
English·PDF·opens deloitte.com
What it reports
- Five shifts: technology-led to human-centric transformation; task-based roles to outcome-based design; digitised processes to AI-native operating models; headcount-driven growth to growth through redesign; cultural inertia to organisational readiness.
- CIOs, CHROs, chief risk officers and business leaders should run as one leadership system with shared accountability.
- Firms should set "human-agent autonomy thresholds": when agents act alone, when humans validate, when they collaborate.
- Lemonade (launched 2015) is the main case: agents "Maya" and "Jim" handle onboarding and claims, and the company is said to scale without an equivalent rise in headcount.
- Brief mentions of NVIDIA, PepsiCo and Volkswagen as examples of clear ambition, dual-speed delivery and scaling pilots.
How they got it
Opinion and advisory framing. No survey, data or method. Company examples are summarised without sources, numbers or confirmation that Deloitte worked on them.
Limits
- No evidence for the 10x productivity framing or for any outcome claim.
- Case descriptions are promotional and unsourced.
- Heavy on urgency language ("Lead or be left behind") and light on how to do the redesign.
Why it matters for companies
The one practical idea is autonomy thresholds: deciding in advance, per workflow, when an agent may act without a human check. That question belongs in every agent deployment review, and it is a joint decision for business owners, IT, risk and HR. The rest restates the common case for redesigning work rather than layering AI on top.
Summary based on the full text. Read the original for the full argument and data.Open PDF on deloitte.com