AI is changing the way enterprise software is purchased, rather than replacing users, by becoming one.
Debate around AI and the workplace is often centered on artificial displacement: Will it replace workers? Where did it fall? In fact, some “AI-First” experiments have produced different results –Clear Reversing the customer service automation course, andDuolingoFacing a public rebound in AI-focused growth strategy.
These results give us a sense of Microsoft’s recent efficiency-driven approachLayoffs. Unlike automation (à laklarna)’s over-promise, Microsoft is reorganizing its own enterprise AI tools as “customer zero”, fundamentally changing the way computing giants write code, ship products and support customers. It’s a strategic arm shot (it’s a painful one) that reveals what’s next: AI agents are not only automating for automation outcomes, but also for deciding the decisions of the tools, processes and infrastructure used along the way.
AI Agent as Orchestration
In the past, enterprise software was selected through familiar dances: evaluation, presentation, stakeholder alignment and procurement. But today, AI agents are building applications, configuring infrastructure and selecting tools (autonomous and large-scale). Ask the agent to rotate the customer feedback portal, it may choose Next.js as the front end, Neon as the cloud database, VERCEL for hosting, and clerks for authentication as a service. No one has to Google Options, compare suppliers or meet with sales staff. Agents are just actions.
Neon lights internal telemetryshowAI agents now create databases at 4 times the speed of human developers. This model is beyond the scope of the engineering. Agents will soon assemble sales pipelines, coordinate landing traffic, manage IT operations, and select effective tools in the process.
Microsoft’s sales team Re-Org further hinted at how such procurement will happen in the future. Now, enterprise customers have a single point of contact at Microsoft, rather than a few salespeople for different products. In some ways, this may be because agent AI tools will choose their own suppliers, and the co-pilot doesn’t require five sales reps. The agent will not stop and ask, “Do you have a preferred vendor?” It will understand the task at hand and continue along its code path, moving towards the answer.
Humans are cycling AI
This evolution from executors to decision makers is provided by the Human (HITL) approach trained by AI Models.
For years, enterprise AI has been limited by expensive labeling processes, fragile automation and underutilized human expertise, resulting in failure of high-risk environments with nuanced differences such as finance, customer service, and healthcare.
The HITL system changes this problem by embedding AI directly into the workforce. In real-time work, agencies observe GUI-level interactions (click, edit, approve), capturing rich signals of natural behavior. These humans are precisely high-quality verification points that improve operational accuracy to ~99% without interrupting workflows. The result is a continuous learning cycle where the agent not only follows the instructions but also learns how to work. This also creates a dynamic life dataset tailored to the actual business processes within the organization.
This transformation offers entirely new market opportunities.
In terms of development, traditional supervised learning models are being replaced by embedded learning systems to harvest real-world interactive signals and make cheaper, faster, and more adaptable AI. This further provides a large-scale new training suite for proxy AI systems without incurring the cost of hiring human knowledge workers to shepherd AI. With lower development costs, high fidelity and better vitality, the next generation of co-pilots will automate with real-time human judgment, dominating vertical industries such as customer service, security, sales and internal operations.
Therefore, these tools will require infrastructure for real-time monitoring, GUI-level interactive capture, dynamic tagging and automated retraining, creating further platform opportunities.
Microsoft’s sense of urgency
Although the internet is full of savvy employees “AI attacks” their workflows, the reality is that most workers lack this product development acumen. (Same goes for their bosses.) In addition to a small portion of the business world with rare technology fluency, most corporate clothing will have greater value in purchasing AI tools, namely world-class talent building, customization and service to address specific workflows.
Microsoft’s sense of urgency stems from its understanding that the “build or buy” issue is changing rapidly. Technically, this “Eureka” moment is what catalyzes the operator hub of corporate AI apparel. HITL stands for the transition from read/write data integration to a more abundant, more dynamic intelligence layer –A method that reflects the actual completion of work in an enterprise.
We saw the start of a game that dominated corporate AI among the Giants in the tech world. Signals such as Openai’s investment in application layer experience (shopping agent) Get proxy developer Windsurf) Highlight a clear trend: Mastering human application interaction capture has become the basis for scalable proxy automation. As companies like Microsoft, OpenAI and others absorb critical data environments and reorganize themselves into “customer zero”, they see AI as the new chief procurement officer of their ecosystem. These companies see the value of selling shovels in the gold rush and know that AI is finally sharp enough to start digging.
Tomasz Tunguz is the founder and general manager of the theoretical joint venture. He has been managing partner at Redpoint Ventures for 14 years.
Comments expressed on fortune.com are entirely the opinions of its author and do not necessarily reflect opinions and beliefs wealth.
Read more:
- “Social AI” is a political brand. Reality is closer to digital colonialism
- Trump accidentally Shows Europe needs to build a full stack of AI industries– Avoid dangerous dependencies
- this AI cost crash is changing what is possible– The huge impact on technology startups
- When AI builds AI: The next great inventor may not be human
This story was originally fortune.com
