Adecco's EZRA pairs AI coach with human experts to scale manager

EZRA Loop blends an always-on AI partner with ICF-accredited coaches, aiming to democratise management coaching at enterprise scale.

A brightly lit office or control room features a large triple-panel display showing complex data visualizations above rows of white desks with dual monitors.

EZRA, the enterprise coaching arm of the Adecco Group, has launched Loop, a product that pairs a persistent AI development partner with a dedicated human coach for each enrolled manager. The launch marks one of the more concrete attempts by a major workforce-services group to operationalise the much-discussed "human-plus-AI" model at scale, moving the concept from strategy deck to billable product.

The product centres on Cai, an AI system that retains context across conversations, tracking a manager's stated priorities, progress and organisational environment. When Cai identifies a development need that falls outside its competence, it routes the manager to one of EZRA's 4,500 ICF-accredited coaches. The division of labour is not arbitrary. EZRA's data, drawn from more than one million sessions with 200,000 people, suggests AI outperforms human coaches on task-oriented skills such as goal-setting and prioritisation, while human coaches retain an edge on interpersonal dimensions including emotional intelligence and executive presence.

The numbers behind the hybrid claim

EZRA reports that teams led by coached managers resolved customer problems 18% faster. More striking is the combined-mode figure: managers receiving both AI and human coaching showed a 37% higher likelihood of measurable behavioural change than those receiving either alone. The company frames this as evidence that the orchestration of human and AI capability, rather than substitution in either direction, is where the performance delta sits.

Denis Machuel, CEO of the Adecco Group, described the logic: "In the hybrid human-AI workplace, it is essential to intentionally design how people and technology interact." The quote is careful to position Loop as a design question rather than a cost-reduction exercise, which matters commercially: enterprise buyers remain sensitive to any framing that implies coaching is being automated downward.

The convergence angle: workforce orchestration as an emerging sector

The launch sits inside a fast-moving competition for what is becoming a distinct market category: agentic workforce tools sold to HR and L&D (learning and development) buyers. Microsoft, which is an EZRA customer, is simultaneously pushing Copilot-embedded coaching nudges through its Viva suite. Vendors including BetterUp, CoachHub and Torch are iterating toward similar hybrid models. The differentiator EZRA is staking out is persistent memory and session continuity between the AI and the human coach, a design choice that demands tighter data architecture than most point solutions offer.

For the cross-sector strategist, the more significant signal is structural. Workforce-services firms such as Adecco have historically monetised labour placement and training volumes. A subscription coaching product built on a proprietary AI layer represents a margin profile that is categorically different: lower variable cost, recurring revenue, and a dataset moat that compounds with every session. Adecco's stock listing on the SIX Swiss Exchange means investors can model this transition directly; the question is how quickly Loop can scale its enrolled manager base to the point where the AI layer begins to offset the cost of maintaining a 4,500-strong human coach network.

There is a broader capital allocation read-across here as well. The enterprise HR-tech category attracted significant venture investment between 2020 and 2023, then cooled sharply as rate rises compressed growth multiples. The current wave of AI-native or AI-augmented HR products is drawing renewed corporate balance-sheet investment rather than VC, with established players such as Adecco, SAP (via SuccessFactors) and Workday each making internal bets rather than acquiring early-stage startups. That pattern suggests incumbents believe they can build fast enough to defend category position without paying acquisition premiums, a thesis Loop will help test.

The regulatory landscape adds a secondary dimension. Coaching and behavioural data held by an AI system that surfaces it to a human coach sits in a grey zone across GDPR and emerging AI-Act obligations in the EU. Adecco is headquartered in Zurich and serves clients in 60 countries; how it structures data residency and consent for Cai's persistent memory will be an early compliance test case for the wider category.