LTM is extending Microsoft 365 Copilot to approximately 140,000 employees across Larsen & Toubro Group, turning an internal AI program into one of India’s largest corporate workplace deployments. The rollout moves beyond email summaries and document drafting, embedding Copilot-powered agents into HR, IT support, sales, onboarding, learning, and software delivery.
The expansion was detailed by LTM and reported by ScanX on July 15. LTM, formerly LTIMindtree, is leading the broader L&T Group implementation after building its own Microsoft-based AI services and training more than 90% of its workforce in generative AI.
The headline number requires some context. LTM’s fiscal 2026 reporting described Microsoft Copilot adoption among more than 71,000 of its own employees, while the new figure of roughly 140,000 covers the wider L&T Group. That makes this a group-scale transformation rather than simply a larger license purchase inside LTM.
The most consequential part of the rollout is not the number of people who can open Microsoft 365 Copilot. It is LTM’s attempt to make Copilot and Microsoft Teams the conversational front end for existing corporate systems, data repositories, and approval processes.
RAIma, LTM’s HR and employee-support agent, was built using Microsoft Copilot Studio and deployed through Microsoft 365 Copilot. Employees can use natural-language conversations to find policies, apply for leave, correct attendance records, book office seats, raise service requests, and retrieve medical insurance information.
Microsoft’s April 2026 case study said RAIma had scaled to more than 78,000 monthly active users. It is available through Teams, Microsoft 365 Copilot, and LTM’s enterprise portal, including chat and voice interfaces.
LTM now reports a 70% improvement in IT and HR query resolution, alongside a 15% increase in employee engagement and HR productivity. Those are company-supplied figures rather than independently audited productivity measurements, and LTM has not published the underlying baselines or precise methodology.
Even so, the usage figures suggest that RAIma is more than a limited proof of concept. For administrators, the relevant lesson is that the most widely adopted enterprise agents may be those that eliminate routine navigation across HR portals, ticketing systems, policy libraries, and internal forms.
RAIma also illustrates the infrastructure sitting behind an apparently simple chat window. An agent handling leave, attendance, insurance, or support requests must enforce identity, role, location, and data-access rules while connecting to multiple systems of record. A polished response is useful only if the agent retrieves the correct policy version and exposes no information that the employee could not access through the original application.
A.S.K., short for Agent for Stories and Knowledge, connects Microsoft 365 Copilot to proposal archives, case studies, presentations, and other approved corporate material. Sales and presales employees can ask for relevant examples, locate the latest presentation, analyze a request for proposal, or generate a first draft grounded in existing content.
Microsoft said the agent connects with more than 40 internal SharePoint repositories and is available to over 1,000 employees across sales, presales, and enablement teams. It uses current access controls when returning personalized results, an important distinction from an unrestricted chatbot trained on a bulk export of company documents.
The design puts SharePoint governance back in the spotlight. Copilot can make forgotten files remarkably easy to discover, but it can also reveal years of weak permissions, duplicate documents, outdated presentations, and inconsistent retention practices.
Before enabling similar agents, Microsoft 365 administrators need to know whether “Everyone except external users” links, abandoned Teams workspaces, oversized security groups, and inherited SharePoint permissions still match business intent. AI does not create every oversharing problem, but semantic search can make existing exposure easier to exploit accidentally.
A.S.K. also shows why enterprise deployments are shifting from general-purpose assistants toward grounded agents. A generic model may produce an elegant sales answer, while an agent connected to approved repositories can reuse the company’s actual delivery history, terminology, and positioning. Human review remains essential, particularly when the generated material becomes part of a contract, bid, or customer commitment.
The developer program matters because its risks differ from those surrounding Word, Outlook, and Teams. Generated code can introduce insecure dependencies, reproduce flawed patterns, expose confidential context, or pass superficial testing while failing under production conditions.
Large-scale adoption therefore depends on more than giving developers access to an assistant. Organizations need policies covering which repositories and data can be submitted, how generated code is reviewed, when security scans are mandatory, and who remains accountable for the finished software.
LTM says its wider program is supported by governance covering security, privacy, compliance, and ethical use. The company has not disclosed a detailed control architecture for the 140,000-person L&T deployment, including how policies vary between subsidiaries, regulated workloads, and geographic regions.
That missing detail will be important as the project expands. L&T Group spans technology services, engineering, construction, manufacturing, energy, and financial services, where data classifications and operational risks can differ substantially. A permission model acceptable for internal sales content may be inappropriate for engineering designs, customer financial records, or infrastructure project documentation.
That approach reflects a broader change in Microsoft 365 administration. Copilot adoption increasingly touches Entra identity, SharePoint information architecture, Teams lifecycle controls, Purview compliance, endpoint policy, data-loss prevention, and Copilot Studio agent management. It cannot be delegated entirely to a workplace productivity team.
Agent creation adds another layer. Low-code tools allow business units to build useful assistants quickly, but they can also produce agent sprawl: overlapping services with unclear ownership, stale connectors, poorly documented data access, and no retirement process.
A sustainable deployment needs an inventory of production agents, named business and technical owners, connector reviews, usage monitoring, incident procedures, and expiration or recertification rules. Administrators also need to distinguish experiments from services that employees rely on for HR, sales, or customer-facing work.
LTM’s reported results indicate that workflow-specific agents can deliver more visible returns than open-ended prompting alone. RAIma resolves repetitive employee requests, while A.S.K. retrieves and repackages institutional knowledge for sales teams. Both operate within tools employees already use instead of requiring another standalone application.
The next test is whether those gains survive expansion from LTM’s technology-focused workforce to roughly 140,000 people across the broader L&T Group. At that scale, permissions, content quality, and agent governance will matter at least as much as Copilot’s underlying models—and every inherited Microsoft 365 configuration mistake becomes part of the AI deployment.
The report also identifies L&T Technology Services alongside LTM as a leader of the implementation. For administrators, the timeline clarifies that the group-wide deployment remains in progress rather than already operating at its full announced scale.
The expansion was detailed by LTM and reported by ScanX on July 15. LTM, formerly LTIMindtree, is leading the broader L&T Group implementation after building its own Microsoft-based AI services and training more than 90% of its workforce in generative AI.
The headline number requires some context. LTM’s fiscal 2026 reporting described Microsoft Copilot adoption among more than 71,000 of its own employees, while the new figure of roughly 140,000 covers the wider L&T Group. That makes this a group-scale transformation rather than simply a larger license purchase inside LTM.
Copilot Becomes a Front End for Corporate Systems
The most consequential part of the rollout is not the number of people who can open Microsoft 365 Copilot. It is LTM’s attempt to make Copilot and Microsoft Teams the conversational front end for existing corporate systems, data repositories, and approval processes.RAIma, LTM’s HR and employee-support agent, was built using Microsoft Copilot Studio and deployed through Microsoft 365 Copilot. Employees can use natural-language conversations to find policies, apply for leave, correct attendance records, book office seats, raise service requests, and retrieve medical insurance information.
Microsoft’s April 2026 case study said RAIma had scaled to more than 78,000 monthly active users. It is available through Teams, Microsoft 365 Copilot, and LTM’s enterprise portal, including chat and voice interfaces.
LTM now reports a 70% improvement in IT and HR query resolution, alongside a 15% increase in employee engagement and HR productivity. Those are company-supplied figures rather than independently audited productivity measurements, and LTM has not published the underlying baselines or precise methodology.
Even so, the usage figures suggest that RAIma is more than a limited proof of concept. For administrators, the relevant lesson is that the most widely adopted enterprise agents may be those that eliminate routine navigation across HR portals, ticketing systems, policy libraries, and internal forms.
RAIma also illustrates the infrastructure sitting behind an apparently simple chat window. An agent handling leave, attendance, insurance, or support requests must enforce identity, role, location, and data-access rules while connecting to multiple systems of record. A polished response is useful only if the agent retrieves the correct policy version and exposes no information that the employee could not access through the original application.
A.S.K. Turns SharePoint Archives Into Sales Material
LTM’s second flagship agent targets a familiar enterprise problem: years of valuable content stored across SharePoint sites but difficult to locate when a sales deadline arrives.A.S.K., short for Agent for Stories and Knowledge, connects Microsoft 365 Copilot to proposal archives, case studies, presentations, and other approved corporate material. Sales and presales employees can ask for relevant examples, locate the latest presentation, analyze a request for proposal, or generate a first draft grounded in existing content.
Microsoft said the agent connects with more than 40 internal SharePoint repositories and is available to over 1,000 employees across sales, presales, and enablement teams. It uses current access controls when returning personalized results, an important distinction from an unrestricted chatbot trained on a bulk export of company documents.
The design puts SharePoint governance back in the spotlight. Copilot can make forgotten files remarkably easy to discover, but it can also reveal years of weak permissions, duplicate documents, outdated presentations, and inconsistent retention practices.
Before enabling similar agents, Microsoft 365 administrators need to know whether “Everyone except external users” links, abandoned Teams workspaces, oversized security groups, and inherited SharePoint permissions still match business intent. AI does not create every oversharing problem, but semantic search can make existing exposure easier to exploit accidentally.
A.S.K. also shows why enterprise deployments are shifting from general-purpose assistants toward grounded agents. A generic model may produce an elegant sales answer, while an agent connected to approved repositories can reuse the company’s actual delivery history, terminology, and positioning. Human review remains essential, particularly when the generated material becomes part of a contract, bid, or customer commitment.
Developers Form the Largest Daily User Base
LTM says more than 23,000 developers now use AI tools daily, supported by over 1,300 AI experts and enablement leads. Its fiscal 2026 reporting previously cited about 21,000 developers using AI assistance, so the higher figure appears to reflect continued adoption or a broader reporting period.The developer program matters because its risks differ from those surrounding Word, Outlook, and Teams. Generated code can introduce insecure dependencies, reproduce flawed patterns, expose confidential context, or pass superficial testing while failing under production conditions.
Large-scale adoption therefore depends on more than giving developers access to an assistant. Organizations need policies covering which repositories and data can be submitted, how generated code is reviewed, when security scans are mandatory, and who remains accountable for the finished software.
LTM says its wider program is supported by governance covering security, privacy, compliance, and ethical use. The company has not disclosed a detailed control architecture for the 140,000-person L&T deployment, including how policies vary between subsidiaries, regulated workloads, and geographic regions.
That missing detail will be important as the project expands. L&T Group spans technology services, engineering, construction, manufacturing, energy, and financial services, where data classifications and operational risks can differ substantially. A permission model acceptable for internal sales content may be inappropriate for engineering designs, customer financial records, or infrastructure project documentation.
Adoption Is Now an Operational Discipline
Training more than 90% of LTM’s workforce in generative AI gives the deployment a stronger foundation than the familiar strategy of distributing licenses and waiting for usage. LTM has also built a network of specialists and enablement leaders who can identify suitable workflows, support employees, and establish reusable implementation patterns.That approach reflects a broader change in Microsoft 365 administration. Copilot adoption increasingly touches Entra identity, SharePoint information architecture, Teams lifecycle controls, Purview compliance, endpoint policy, data-loss prevention, and Copilot Studio agent management. It cannot be delegated entirely to a workplace productivity team.
Agent creation adds another layer. Low-code tools allow business units to build useful assistants quickly, but they can also produce agent sprawl: overlapping services with unclear ownership, stale connectors, poorly documented data access, and no retirement process.
A sustainable deployment needs an inventory of production agents, named business and technical owners, connector reviews, usage monitoring, incident procedures, and expiration or recertification rules. Administrators also need to distinguish experiments from services that employees rely on for HR, sales, or customer-facing work.
LTM’s reported results indicate that workflow-specific agents can deliver more visible returns than open-ended prompting alone. RAIma resolves repetitive employee requests, while A.S.K. retrieves and repackages institutional knowledge for sales teams. Both operate within tools employees already use instead of requiring another standalone application.
The next test is whether those gains survive expansion from LTM’s technology-focused workforce to roughly 140,000 people across the broader L&T Group. At that scale, permissions, content quality, and agent governance will matter at least as much as Copilot’s underlying models—and every inherited Microsoft 365 configuration mistake becomes part of the AI deployment.
Update: L&T Targets Full Copilot Deployment by December 2026 (July 15, 2026)
Whalesbook reports that L&T has distributed nearly 100,000 Microsoft Copilot licenses so far, with the remaining rollout expected to be completed by December 2026. The planned 140,000 users represent about 80% of the conglomerate’s workforce.The report also identifies L&T Technology Services alongside LTM as a leader of the implementation. For administrators, the timeline clarifies that the group-wide deployment remains in progress rather than already operating at its full announced scale.
References
- Primary source: scanx.trade
Published: 2026-07-15T05:02:06.919000+00:00
LTM Leads L&T Group's AI Workplace Transformation with Microsoft 365 Copilot
LTM is leading one of India's largest enterprise AI workplace transformations by deploying Microsoft 365 Copilot across the L&T Group for approximately 140,000 employees. The rollout, supported by AI tools RAIma and Agent A.S.K., has delivered a 70% improvement in IT and HR query resolution...scanx.trade - Official source: adoption.microsoft.com
- Official source: news.microsoft.com
From prompts to partnership: How LTM’s Rajesh Kumar collaborates with Microsoft 365 Copilot - Source Asia
LTM CIO Rajesh Kumar collaborates with Microsoft 365 Copilot to increase productivity and efficiency at worknews.microsoft.com
Last edited:
