WORK-SELF expands Maya Enterprise across 12,000 apps

WORK-SELF has expanded its Maya Enterprise platform across more than 12,000 enterprise applications. It says the system is designed to reduce employee rework in AI-driven workflows.

The London-based company says Maya Enterprise adds what it calls Human Context Infrastructure to agentic workflows on platforms including Microsoft, Salesforce, ServiceNow, SAP and UiPath.

WORK-SELF is pitching the product as companies seek tighter control over the growing use of AI agents in large organisations. Its argument is that businesses can improve returns from AI not only by making agents more effective, but also by reducing the human checking, correction and coordination that often follows an automated task.

According to the company, enterprise engagements using the system have shown up to 40% less employee rework. For clients with more than 1,000 employees, better coordination between people and AI agents can amount to more than GBP £5 million in annual savings and more than 500,000 skilled hours returned to the workforce, it said.

International Data Corporation has projected that active AI agents will rise from 28.6 million in 2025 to 2.2 billion by 2030. WORK-SELF cited the forecast to illustrate the expected scale of adoption.

How it works

Maya Enterprise is designed to give AI agents organisational information such as who owns a task, who can approve decisions, when human review is needed, how much autonomy an agent should have and when work should be escalated.

According to WORK-SELF, the platform maps employees, roles, responsibilities, decision rights, policies, workflows, approvals, preferences and agent actions through a semantic data layer and a Human Context knowledge graph. It then serves that information to agents in real time through what it calls Context Capsules.

The company says its Simulation Lab models work at task level across departments and workflows. It also says its WORK and SELF archetypes have been pre-trained across 2.2 billion task-level scenarios to help define relationships between employees, agents and tasks.

Beyond the reported reduction in rework, WORK-SELF says semantic optimisation in the way context is delivered to agents has produced a further 20% improvement in workflow efficiency. It adds that Human Context-enabled automation has led to a 60% improvement in cost per unit of delivered work.

"Enterprises have invested heavily in making agents faster and cheaper. That is only part of the ROI equation," said Wolf Magdelinic, chief executive officer and founder of WORK-SELF.

"If an employee still has to check, correct and coordinate the work around the agent, productivity is leaking back out of the workflow. Human Context is designed to give that time back," Magdelinic said.

Partner network

Alongside the expansion, WORK-SELF outlined a group of technology and implementation partners tied to the offering, including CockroachDB, Takara.ai, INT Global and UAIN Solutions.

According to the company, CockroachDB provides the distributed data infrastructure for Maya's knowledge graph, context exchange and audit records. WORK-SELF says it validated its data layer on CockroachDB without changing application code, supporting regional data placement and resilient deployments.

INT Global is being used to connect governed information across fragmented enterprise systems through its OneSpace data platform. WORK-SELF says the partner brings more than 1,000 professionals across technology, data, cloud and security.

Takara.ai focuses on AI engineering and semantic optimisation. WORK-SELF says that work improved how Human Context is prepared and delivered to agents, contributing to the additional workflow efficiency gains it reported.

UAIN Solutions, a UiPath implementation partner, is supporting Maya deployments across Microsoft, ServiceNow, Salesforce and Agentforce, SAP and UiPath. According to WORK-SELF, that includes Microsoft expertise led by Microsoft AI MVPs.

Existing estates

A key part of the company's pitch is that businesses do not need to replace existing systems to adopt the product. WORK-SELF says Maya operates through a low-code interoperability layer that connects to enterprise applications, databases, data lakes, software-as-a-service platforms, large language models, custom applications, legacy systems and on-premise infrastructure.

According to the company, the underlying orchestration setup has supported more than 100 million automated tasks, more than 750,000 reusable recipes and skills, and 99.9% uptime. It adds that identity-aware execution, human approvals, traceability and MCP interoperability are built into the platform.

"Connectivity is not coordination," Magdelinic said.

"An agent can be connected to every application it needs and an employee can still be left deciding whether the work is right, who owns it and what happens next. Human Context turns agent productivity into workflow productivity," he said.