Microsoft India has reported a 9% increase in sales revenue and a 20% improvement in deal closure speed after integrating AI agents across its sales teams. These agents handle repetitive tasks, enabling employees to focus on higher-value work. The shift reflects Microsoft’s evolving strategy of embedding artificial intelligence deeper into everyday operations.
Human–AI Collaboration in the Workplace
Microsoft’s approach is based on creating a dual workforce, where AI agents function as colleagues to human staff. This integration is not limited to automation; it involves active collaboration where both human and AI inputs are used to close deals, analyse client needs, and improve turnaround times.
Beyond India, this marks a global trend of organisations using AI to enhance workforce productivity rather than replace human jobs. In Microsoft’s case, the gains are measurable and part of a broader transformation plan.
Launch of MAI Models to Cut OpenAI Dependency
At the core of this transition is Microsoft’s launch of its own AI models. The new MAI-Voice-1 and MAI-1-preview are part of the company’s strategy to reduce dependence on OpenAI. MAI-Voice-1 is a natural speech model designed for performance using a single GPU. MAI-1-preview, the larger model, was trained using 15,000 Nvidia H-100 GPUs—fewer than required by rival models—making it more cost-efficient.
These models are already being used to power features like Copilot Daily and AI-generated Podcasts, and are available for testing on the LMArena platform. Microsoft sees this development as part of a five-year roadmap to assert greater control over its AI infrastructure and accelerate innovation.
Strategic Push Amid Industry Concerns
Microsoft’s shift comes at a time when experts are questioning the pace of AI investments and the risk of a potential AI bubble. The company believes building internal capabilities will provide long-term value and flexibility. With its multi-model approach, Microsoft can now customise AI experiences across user groups without being limited by third-party tools.
While maintaining its partnership with OpenAI, the tech giant is positioning itself for greater self-reliance. The challenge ahead will be to validate whether the benefits of building AI models in-house outweigh the risks and costs in the long run.
