The question changed sometime between January and May. Twelve months ago, professional services firms asked which AI model to adopt. Now they ask something harder: how do you let AI loose on decades of client files without it accessing the wrong matter, breaching an ethical wall, or exposing privileged material?
iManage answered Tuesday at ConnectLive 2026 in Chicago.
The knowledge work platform provider unveiled what it calls a “context fabric”—a rebuilt architecture designed to let AI agents search, reason over, and act on institutional knowledge while respecting the permission boundaries, security policies, and compliance frameworks that govern professional services work. The announcement, made 20th May in London, positions iManage as a governed layer between AI tools and the sensitive client knowledge they need to be useful.
For the 4,000 organisations using iManage—including 83% of the Top Global 100 law firms, 40% of the Fortune 100, and 79% of the Am Law 100—the shift matters. These aren’t organisations experimenting with ChatGPT anymore. They’re operationalising AI across client work, deal analysis, contract review, and litigation strategy. That requires infrastructure.
“Knowledge work is evolving faster than ever, and the value of AI depends on safely activating the knowledge, context, and expertise organisations already have,” said Neil Araujo, CEO of iManage. “iManage is helping organisations move from systems that simply store knowledge to a secure, governed foundation that actively surfaces it, connects it, and makes it usable for AI – contextually, responsibly, and at scale.”
The platform evolution arrives as iManage reports adding 90 new customer logos in 2026 and expanding cloud adoption to 78% of its global customer base. That cloud migration rate—up from lower penetration in previous years—reflects a broader enterprise shift toward managed infrastructure, particularly as AI workloads demand scalability that on-premise systems struggle to deliver.
What iManage calls a context fabric is more than document search. The system tracks relationships between files, matters, clients, and collaborators. It monitors real-time activity—who’s working on what, which documents relate to which deals, what precedent applies to current matters. That layer of metadata and relationship mapping becomes the context that AI agents need to generate useful outputs rather than generic responses.
Crucially, governance sits inside the fabric rather than bolted on afterward. Ethical walls, client-matter restrictions, and access permissions are native to how the platform operates. When an AI agent queries the system, it sees only what the requesting user would be authorised to access. That permission-aware approach addresses a core concern for regulated industries: how do you prevent AI from synthesising information across matters it should never connect?
The architecture matters for a practical reason. Professional services firms aren’t picking one AI tool. They’re adopting multiple models, agents, and applications across different teams and use cases. Some prefer OpenAI. Others want Anthropic’s Claude. Still others are evaluating Microsoft’s Copilot, Google’s Gemini, or specialised legal AI providers.
iManage positions itself as the governed knowledge foundation underneath that heterogeneous AI layer. Rather than forcing firms to choose, it provides the secure context that any AI tool can query through standardised protocols.
That strategy crystallised with Tuesday’s announcement of expanded partnership with Anthropic. Through iManage’s Model Context Protocol (MCP) Server, Claude can now access governed iManage knowledge—matter history, documents, institutional precedent—without requiring bulk exports or custom integrations. The partnership gives Anthropic enterprise credentials in professional services while giving iManage customers freedom to adopt Claude without compromising governance.
For organisations evaluating AI tools, the Anthropic integration matters as proof of concept. It demonstrates that multiple AI providers can connect to governed knowledge through standardised interfaces, reducing lock-in risk and preserving optionality as the AI landscape evolves.
The platform updates previewed at ConnectLive span three areas: making knowledge accessible to AI, strengthening governance at scale, and reducing friction in daily workflows.
On accessibility, iManage MCP Server enables AI agents and large language models to search permission-aware context directly from governed knowledge repositories. Multi-region search gives global organisations unified search across jurisdictions, helping users and AI tools find relevant context regardless of where data resides. Native optical character recognition automatically makes scanned documents searchable and legible to AI—a significant capability for firms with decades of paper files converted to image-based PDFs.
Governance capabilities are evolving to handle AI at scale. Security Policy Manager will support granular client-level and matter-level restrictions for AI use, letting firms define which matters or clients AI tools can assist with. Threat Manager now surfaces AI agent activity in user reporting, giving security teams visibility into what agents access, move, or modify. That audit trail becomes critical for compliance teams needing to demonstrate oversight of AI-assisted work.
Disposition Manager, a cloud-native records management application, now allows records managers to intervene and override exceptions when automated workflows encounter issues—checked-out documents, unavailable approvers, stuck processes. The capability reflects a reality: automated governance breaks in predictable ways, and someone needs tools to fix it without escalating to IT.
Collaboration Links brings secure external sharing into the platform, letting teams share documents with clients through simple links while maintaining governance and enabling Microsoft 365 co-authoring. External collaborators can view and edit documents without iManage accounts, addressing a longstanding friction point where teams moved work outside governed systems to collaborate with external parties.
The features aren’t revolutionary individually. Multi-region search, OCR, secure external sharing—these capabilities exist in enterprise content management systems. What’s different is the integration layer designed specifically for AI workloads in regulated professional services environments.
iManage has operated in legal and professional services for over 30 years, managing documents and email for more than one million professionals. That installed base and institutional knowledge of how law firms, accounting firms, and corporate legal departments work gives it advantages in understanding governance requirements that generic cloud storage providers lack.
The ConnectLive announcements represent the first phase of what iManage describes as a broader platform evolution. Features previewed this week will roll out across 2026 and into 2027, with ongoing development in close collaboration with customers.
For attendees at ConnectLive Chicago on 19-20 May and the London event scheduled for 9-10 June, the platform demonstrations offer a glimpse of how governed AI might work in practice. Whether the architecture delivers on its promise of secure, contextual AI at scale remains to be tested in production environments.
What’s clear is the strategic bet: that the value in enterprise AI isn’t the model, but the governed context feeding it. If iManage is right, the knowledge layer matters more than the intelligence layer. If they’re wrong, firms will find ways to connect AI tools directly to content without needing an intermediary platform.
By mid-2027, the market will have answered that question. For now, 4,000 organisations are placing their bets on governed foundations over raw capability. The agents can wait.
