March has been a busy month for AI announcements, and the flurry of updates carries real implications for how Indian companies should plan their technology investments. Rather than chasing every headline, smart leaders need to separate signal from noise.
This roundup cuts through the marketing speak to tell you what actually matters for your business decisions in the coming quarters.
OpenAI Expands Enterprise Controls, Eyes Regulated Industries
OpenAI announced expanded data residency options for its Enterprise tier, now including servers in Singapore and Frankfurt. For Indian companies in banking, healthcare, or government contracting, this addresses a persistent compliance headache — where exactly does your data live when you use AI tools?
The company also introduced granular audit logging, letting IT teams track exactly which employees accessed what AI features and when. This matters if you’re building toward ISO 27001 certification or navigating RBI’s technology risk guidelines.
Pricing remains steep at the enterprise level, but OpenAI signalled volume discounts for companies committing to annual contracts above 500 seats. If you’ve been running pilots on pay-as-you-go plans, March might be a good time to negotiate.
Google DeepMind Pushes Multimodal into Production
Google’s Gemini models received a significant update focused on what the company calls “agentic workflows” — essentially, AI that can take actions across multiple software tools rather than just answering questions. Early access partners report the system can now reliably navigate enterprise software like SAP and Salesforce when given proper API credentials.
For CTOs evaluating automation projects, this changes the build-versus-buy calculation. Tasks that previously required custom integration work might now be achievable through Google’s pre-built connectors.
The catch: Google Cloud Platform commitment remains a prerequisite for the most capable features. If you’re already on AWS or Azure, switching costs still make this a harder sell than the demos suggest.
Anthropic and Meta Take Different Paths on Safety
Anthropic released Claude’s latest iteration with expanded “Constitutional AI” features — a system that lets enterprises define their own rules about what the AI should and shouldn’t do. A fintech company, for instance, can instruct the model to never provide specific investment advice, with those guardrails enforced at the model level rather than just the application layer.
Meta, meanwhile, continued its open-source push with new Llama model weights optimised for on-device deployment. Indian startups building mobile-first AI products should pay attention — running models locally means no per-query API costs and better performance in areas with unreliable connectivity.
The divergence is strategic. Anthropic is betting that enterprises will pay premium prices for controllable AI. Meta is betting that widespread adoption of its open models will drive developers toward its broader ecosystem. Both approaches create opportunities for Indian companies, depending on whether you prioritise cost control or compliance certainty.
Smaller Players Worth Watching
Beyond the giants, several mid-sized AI companies made moves that could affect your vendor shortlist. Cohere launched India-specific pricing for its enterprise search product, acknowledging that US dollar pricing has been a barrier for local adoption.
Mistral, the French AI company, announced partnerships with two Indian system integrators for on-premises deployments — relevant if your compliance requirements prohibit cloud-based AI entirely.
Perplexity expanded its enterprise tier with features aimed at replacing internal knowledge bases, positioning itself as a competitor to tools like Notion AI and Confluence’s new AI features. Early reports suggest it handles technical documentation better than general-purpose chatbots.
What This Means for You
If you’re a founder or tech leader in India, March’s announcements suggest three concrete actions.
First, revisit your AI vendor contracts before Q2 planning. Several companies are offering better terms for annual commitments, and the competitive pressure means negotiation leverage you didn’t have six months ago.
Second, evaluate on-premises or hybrid deployment options if you’ve been avoiding AI due to data residency concerns. The gap between cloud-only and self-hosted capabilities is narrowing faster than most analysts predicted.
Third, assign someone on your team to track agentic AI developments specifically. The shift from AI-as-chatbot to AI-as-workflow-automation will reshape how companies think about headcount and process design over the next eighteen months.
The companies announcing today are shipping products in ninety days. Your planning cycles should account for that pace.