Microsoft Rebuilds Copilot Around Work, Adds Coding and Autopilot Agents, and Its Stock Jumps 4%

Microsoft has made a decisive choice about where it wants to compete in artificial intelligence. On Friday, September 25, the company unveiled a sweeping overhaul of its Copilot assistant that merges its consumer and workplace versions into a single application aimed squarely at businesses, and introduces new agents designed to do work rather than simply answer questions.

Investors welcomed the move. Microsoft shares rose about 4 per cent to around $516 in afternoon trading, according to 24/7 Wall St. Oppenheimer analyst Brian Schwartz raised his price target to $570 from $515 and maintained an outperform rating, arguing that customers are increasingly standardising on Microsoft as their primary enterprise AI platform.

Stepping back from the consumer chatbot race

The most striking part of the announcement was what Microsoft said it would not do. The company has effectively ceded the consumer AI companion market to rivals such as OpenAI's ChatGPT, Google's Gemini and Meta's Muse.

"We're not going to build a Copilot that's like your personal companion," said Charles Lamanna, a senior Microsoft executive. "That's just not what people want from Microsoft. We help you get stuff done."

It is a notable shift for a company that had invested heavily in consumer Copilot features and marketing. By focusing on the workplace, Microsoft is playing to its strengths: its dominant position in office software through Microsoft 365, its relationships with large enterprises and its security and compliance infrastructure.

Three pillars of the new Copilot

The redesigned Copilot is built around three main capabilities.

Home is a central hub for conversational chat and task delegation. It includes what Microsoft calls "Office in Copilot", which embeds Word, Excel and PowerPoint directly within Copilot so users can work on documents collaboratively without switching applications.

Code is a tool powered by GitHub Copilot technology that lets employees, including those without programming skills, build apps, dashboards, workflows and software using natural language prompts. It extends the idea of "vibe coding", where users describe what they want and AI writes the code, into the enterprise, with the governance controls that large organisations require.

Autopilot is a persistent digital agent that can carry out complex, multi-step tasks in the background within a company's cloud environment. It is a revamp of an agent called Scout that Microsoft introduced in June. Autopilot has its own identity in the corporate directory and operates under permissions controlled by users, an approach designed to address concerns about AI agents acting without oversight.

According to Reuters, Code will begin rolling out in early access at the end of the month, with a preview for Microsoft 365 Premium and Pro subscribers later in 2026. Autopilot will enter private preview at the end of the month.

Tackling the enterprise agent problem

Microsoft's emphasis on identity, permissions and governance reflects the main obstacle to AI agent adoption in large companies.

"I think it's been very hard for many organizations to bring agents into the enterprise because of security, compliance and governance concerns," said Annie Pearl, corporate vice president of Copilot product, in comments reported by Reuters.

Enterprises are enthusiastic about the productivity potential of agents that can complete tasks such as preparing reports, reconciling data or managing workflows. But they worry about agents accessing sensitive information, taking unauthorised actions or making errors that are difficult to trace. By giving agents directory identities and tying them to user-controlled permissions, Microsoft is applying the same security model it uses for human employees.

A new way to pay for AI

The overhaul also introduces a new pricing structure. Everyday chat will be available through a fixed-rate subscription, while long-running agentic tasks will be billed on a pay-as-you-go basis. Microsoft is adding "FinOps for AI" governance tools to help companies track and manage AI usage costs, including insight into how much individual employees are spending.

The shift to usage-based pricing for agents reflects the economics of AI. Agentic tasks can consume far more computing power than simple chat queries, and a flat subscription fee may not cover those costs. Charging for usage aligns revenue with costs, but it also creates uncertainty for customers about their bills, which is why cost management tools are an important part of the package.

Why investors cheered

Microsoft has faced growing pressure to show a return on its enormous investment in AI infrastructure. The company is spending more than $115 billion on capital expenditure in 2026, much of it on data centres. Investors have been looking for evidence that Copilot and other AI products can generate revenue on a scale that justifies that spending.

Before Friday, Microsoft's stock had lagged the broader technology sector, rising about 7 per cent this year compared with roughly 37 per cent for the sector, according to 24/7 Wall St.

The company has reported more than 30 million paid Copilot enterprise subscriptions as of June, representing about a third of its Microsoft 365 commercial user base. The new usage-based model for agents creates a path to additional revenue from those customers as they adopt more advanced AI capabilities.

Oppenheimer's Schwartz argued that the enterprise focus positions Microsoft well. As businesses consolidate their AI spending, many are likely to prefer a platform that integrates with the tools they already use and meets their security requirements.

The competitive landscape

Microsoft's retreat from consumer AI does not mean less competition. In the enterprise market, it faces rivals including Google, which offers Gemini across its Workspace productivity suite; OpenAI and Anthropic, which sell directly to businesses; and specialist software companies building agents for specific functions.

The Code tool, in particular, enters a crowded field of AI coding assistants and app-building platforms. Microsoft's advantage lies in GitHub, the world's largest software development platform, and in its ability to bundle AI capabilities with existing enterprise licences.

What it means for Indian enterprises and IT services

For Indian businesses, many of which run on Microsoft 365, the new Copilot could make AI agents more accessible. Tools that let non-technical employees build apps and automate workflows may be particularly valuable for mid-sized companies without large IT teams.

The shift also has implications for India's IT services industry. As AI agents take on more routine tasks, from building simple applications to processing data, service providers will need to move toward higher-value work such as designing AI systems, integrating them into business processes and managing governance.

A bet on work, not companionship

Microsoft's Copilot overhaul is a clear strategic bet: that the biggest prize in AI lies in helping businesses get work done, not in building a digital companion for consumers. With new coding and autonomous agent capabilities, enterprise-grade governance and a pricing model tied to usage, the company is positioning Copilot as the operating layer for AI at work.

The market's initial response was positive, but the real test will come as Code and Autopilot move from preview to general availability. If enterprises adopt them at scale, Microsoft's decision to walk away from the consumer chatbot race could prove to be one of its smartest moves of the AI era.