WORK-SELF launches Maya Human Context MCP Server for enterprise AI agents

WORK-SELF on June 9, 2026 announced Maya Human Context MCP Server, a governed layer that gives enterprise AI agents permissioned, employee-specific context before they initiate, escalate or hand off work. The company says the system is designed to improve human-agent collaboration while limiting oversharing and adding auditability in sensitive workforce settings.

Why it matters: - Enterprise AI agents can already retrieve data and execute tasks, but many still miss the human context needed to collaborate safely and effectively. - WORK-SELF is positioning Maya Enterprise as the runtime layer that tells agents how to work with people, not just with systems. - The pitch matters for companies trying to automate work without increasing confusion, manual rework, or risky escalation decisions.

What happened: - WORK-SELF announced the Maya Human Context MCP Server, described as the next evolution of Maya Enterprise. - The launch took place on June 9, 2026 in London. - The product is built to give enterprise AI agents governed, permissioned, employee-specific human context before they initiate, escalate or hand off work. - CEO and Co-Founder Wolf Magdelinic said MCP gives agents a standard way to connect to enterprise tools and data, while Maya gives them a governed way to understand the humans they work with.

The details: - Maya returns a permissioned Context Capsule and Work Contract for approved enterprise agents. - The Context Capsule includes the minimum necessary task, role, organisational, culture, transition-readiness and work-preference context for a specific purpose and duration. - The Work Contract defines human-owned decisions, agent-owned tasks, escalation rules, review style, autonomy thresholds, interruption cadence and completion rules. - The platform also includes a Review Map that tells employees what to read, skim, ignore, decide or delegate. - An Autonomy Dial shows workflow mode as Manual, Copilot, Delegated, Autopilot, Training or Focus. - A Digital Andon signal lets humans or agents stop or escalate when risk, missing context, overload or unclear decision rights appear. - A Learning Loop updates the Work Contract after accepted, edited, escalated or rejected AI outputs. - WORK-SELF says Maya is built on a proprietary identity graph, 80,000+ identity profiles, 2.2 billion scenario permutations and a founder-held 17-claim US AI patent. - The company says the system also draws on 1,000+ Maya AI Agent consultations and 151 full Career Audits. - Maya Staging, HRIS/API integrations and MCP connectors can attach target operating model documents, org charts, policies, role architectures, knowledge bases, CRM records, project data, learning materials and workflow documentation. - The platform includes a Cohort of Influence methodology that incorporates reference organisations and bodies of work across workforce transformation, change management, skills intelligence, executive coaching, behavioural science, career identity and AI governance. - WORK-SELF says governed human context is visible, consented, correctable, purpose-limited, auditable and shared with agents only when relevant to a task.

Between the lines: - WORK-SELF is framing Maya as a category move, not a point tool, by arguing that enterprise AI now needs a human-context layer alongside the standard MCP-style tool connection layer. - The company is also drawing a hard line around governance, signaling that workforce context can be useful to AI agents without becoming a hidden surveillance system. - That positioning could resonate with employers that want AI orchestration in sensitive workflows but still need employee trust, legal review and audit trails.

What’s next: - WORK-SELF says early use cases include financial services, professional services, customer service and operations, and enterprise transformation programmes. - In financial services, the company sees use across investment research, compliance monitoring, client advisory, model governance and internal mobility into AI oversight roles. - In professional services, it targets client proposal workflows, research synthesis, deliverable review, account planning and quality-control handoffs. - In customer service and operations, the platform is aimed at AI triage, human escalation pods and policy-aware handoffs. - For transformation programmes, WORK-SELF points to pre-announcement readiness diagnostics, redeployment planning, workforce transition coaching and CHRO visibility across affected cohorts. - The governance model includes employee-visible profiles, granular consent, purpose limitation, data-loss prevention, immutable audit logs, context expiry and separation between private employee context and manager-facing support signals. - Where biometric-aware inputs or deeper transition-state signals are used, WORK-SELF recommends explicit consent, legal review and no use of private employee context for performance management without a separate lawful basis and human review.

The bottom line: - WORK-SELF wants to make human context a governed enterprise AI primitive, not an afterthought.