Y Combinator CEO Spends Nearly ₹2 Crore a Year on AI Agents. Is This What the Future of Business Looks Like?
Published
Artificial Intelligence is no longer confined to answering questions, generating content or helping people write emails. It is gradually becoming something much more capable: a digital assistant that can understand context, work across different tools and take action with a degree of independence. And one of Silicon Valley's most prominent startup leaders has offered a fascinating glimpse into that future. Garry Tan, CEO of Y Combinator, has revealed that he spends between $100,000 and $200,000 annually on AI tokens. That translates to roughly ₹90 lakh to ₹1.8 crore, with the upper figure often described as approximately ₹2 crore. But the amount is not even the most interesting part of his revelation. It is what his AI agents actually do.
What AI Does Garry Tan Actually Use?
Garry Tan's AI setup is not built around a single chatbot. It involves multiple tools and AI agents working together. One of the key tools in his workflow is Claude Code, Anthropic's AI-powered coding agent. It helps developers work on software projects through natural-language instructions, understand code and complete complex development tasks. Tan has also developed and publicly shared a collection of Claude Code skills called gstack. It contains specialised instructions that allow AI to approach different tasks from different professional perspectives, such as a CEO, engineer, designer or code reviewer. Another important part of his setup is OpenClaw, an open-source platform for building and operating AI agents. Tan has described himself as an early adopter and a power user of the platform. And then there is Stray Light, an internal AI agent used at Y Combinator. Unlike a general-purpose chatbot, Stray Light is designed around YC's organisational knowledge, including information about founders, startup applications and correspondence. These tools serve different purposes: • Claude Code: Helps with coding and software development. • gstack: A collection of specialised AI skills built around Claude Code. • OpenClaw: Helps operate AI agents that can work across different tasks and tools. • Stray Light: Y Combinator's internal AI agent, built around the organisation's knowledge. The important point is that Tan is not simply opening an AI chatbot whenever he needs an answer. He is building an ecosystem of AI tools and agents that can assist with different aspects of his work and personal life.
When AI Knows More Than Just Your Next Question
Artificial Intelligence is no longer limited to answering questions, generating content or helping people write emails. It is gradually becoming something much more capable: a digital assistant that can understand context, work across different tools and take action with a degree of independence. And one of Silicon Valley's most prominent startup leaders has offered a fascinating glimpse into that future. Garry Tan, CEO of Y Combinator, has revealed that he spends between $100,000 and $200,000 annually on AI tokens. That translates to approximately ₹90 lakh to ₹1.8 crore, with the upper figure often described as nearly ₹2 crore. But the amount is not even the most interesting part of his revelation. It is what his AI agents actually do.
What AI Does Garry Tan Actually Use?
Garry Tan's AI setup is not built around a single chatbot. It involves multiple tools and AI agents working together. One of the key tools in his workflow is Claude Code, Anthropic's AI-powered coding agent. It helps developers understand code, work on software projects and complete complex development tasks through natural-language instructions. Tan has also developed and publicly shared gstack, a collection of specialised Claude Code skills. These skills allow AI to approach different tasks from different professional perspectives, including product strategy, engineering, design, code review and quality assurance. Another important part of his setup is OpenClaw, an open-source platform for building and operating AI agents. Tan has described himself as an early adopter and power user of the platform. And then there is Stray Light, an internal AI agent used at Y Combinator. Unlike a general-purpose chatbot, Stray Light is designed around YC's organisational knowledge, including information about founders, startup applications and correspondence. Here is a quick look at the different parts of his AI ecosystem: • Claude Code: Assists with coding and software development. • gstack: Provides specialised AI skills for different professional tasks. • OpenClaw: Helps build and operate AI agents. • Stray Light: Y Combinator's internal AI agent, built around the organisation's knowledge. These tools serve different purposes, but they point towards the same idea: AI is evolving from a tool that responds to instructions into a system that can actively assist with work.
When AI Knows More Than Just Your Next Question
In a conversation with author Walter Isaacson on The Builders podcast, Tan described how extensively he relies on AI agents for professional and personal tasks. According to his account, his AI agents have access to his emails, Dropbox files and Apple Notes, including personal therapy notes going back approximately 20 years. They also assist him with coding, research and preparation for important conversations. Imagine having an assistant that understands your previous conversations, remembers important documents and can connect information from different parts of your life. That is the kind of experience Tan is exploring. He describes AI agents as something like a personal assistant that knows you, helps with everyday activities and can sometimes anticipate what you might need next. Y Combinator's internal agent, Stray Light, takes a similar concept into an organisational environment. It brings together extensive company context to help the accelerator work with information about founders and applications. This represents a significant shift in how organisations can think about artificial intelligence. AI is moving from being a tool that employees occasionally open to becoming a more integrated part of how work gets done.
The Difference Between an AI Chatbot and an AI Agent
For years, businesses have experimented with chatbots and AI-powered writing tools. These systems can answer questions, draft messages, summarise documents and generate ideas. AI agents introduce another dimension: the ability to work through tasks and workflows with greater autonomy. Consider a simple example. A conventional chatbot can help a business owner draft a follow-up email to a potential customer. An AI-powered agent, when connected to the appropriate business systems and given the necessary permissions, could help identify the lead, review previous interactions, prepare a personalised follow-up and assist with the next step. The distinction is important. A chatbot primarily responds to instructions. An AI agent can help move a process forward. This does not mean every AI agent can independently access every system or make decisions without oversight. Its capabilities depend on integrations, permissions, instructions and safeguards. Nevertheless, the direction is becoming increasingly clear: the next phase of AI adoption will be about completing work, not merely generating answers.
Why Businesses Need More Than Just AI Tools
Imagine running a growing business. Customer enquiries arrive through WhatsApp. Sales teams maintain lead records in spreadsheets. Employees coordinate through different communication channels. Financial updates sit in separate files, while managers spend hours collecting information before making decisions. The challenge is not necessarily a shortage of people or effort. It is the absence of connected information. A business owner may have all the necessary information somewhere, but finding it, understanding it and acting on it can consume a considerable amount of time. This is precisely where AI agents could make a meaningful difference. Instead of treating every department as an isolated function, businesses can begin connecting information and workflows across their operations. For example: Sales and CRM: AI can assist teams in organising leads, reviewing customer interactions and identifying follow-up opportunities. Human Resources: AI-powered workflows can help organise employee information, support recruitment processes and simplify routine HR activities. Finance: Connected financial information can help business owners monitor transactions, outstanding payments and operational expenses. Customer Communication: AI can assist with WhatsApp-related workflows, customer queries and timely follow-ups. Management: Connected business information can help managers understand what is happening across different departments without repeatedly asking teams for updates. The real opportunity is not simply to introduce more AI tools. It is to make the tools and information a business already relies on work together.
Meet Mingrow's 12 AI Agents: A Smarter Way to Run a Business
Garry Tan's AI setup demonstrates what intelligent agents can do when they have access to relevant information and the right tools. But this technology is not just for Silicon Valley executives or companies with massive AI budgets. Businesses also need specialised AI assistance that understands their day-to-day operations. This is where Mingrow, The AI Business OS, brings its approach to the conversation. Mingrow brings together a team of 12 specialised AI agents, designed to support different aspects of business operations. Its AI agent ecosystem includes names such as Sidd, Dan, Lucy, Serra and Corolla, alongside the other specialised agents in its team. Instead of depending on one general-purpose assistant for every business requirement, Mingrow's approach brings multiple AI agents into a connected business environment. The idea is straightforward: different business functions have different requirements, and AI assistance should be useful in the context of those requirements. Think about a business where customer information, lead management, employee records, financial activities and communication are all part of everyday operations. When these functions are disconnected, employees spend valuable time switching between platforms and coordinating information manually. Mingrow aims to reduce that operational fragmentation by bringing essential business functions together. Its platform combines: • CRM and Lead Management to organise customer information and sales activities. • HR Management to support employee-related operations. • Finance Management to bring financial activities into a connected environment. • Field Staff Management to support coordination beyond the office. • WhatsApp Workflows to make business communication more organised. • 12 AI Agents to bring specialised AI assistance into everyday business operations. The bigger idea is to make AI part of how a business functions, rather than treating it as another application employees have to remember to open. For a growing business, that can mean less time spent coordinating scattered information and more time focusing on customers, decisions and growth.
From Garry Tan's Personal AI Setup to Mingrow's Business AI Ecosystem
There is an important difference between Tan's personal AI infrastructure and the requirements of an ordinary business. Tan has built an extensive AI environment around his personal information, professional responsibilities and organisational knowledge at Y Combinator. A growing business, however, needs something more structured. It needs systems that can support customers, employees, financial activities, communication and daily operations without creating another layer of complexity. That is the opportunity behind an AI Business OS. Mingrow is working towards this direction by combining business management capabilities with a team of specialised AI agents. Rather than asking business owners to manage multiple disconnected applications, the platform aims to bring essential operations into one connected environment. The goal is not to replace the people who build and operate a business. It is to reduce repetitive coordination, improve access to business information and make everyday workflows more manageable. In that sense, the broader AI shift that Garry Tan is demonstrating is closely connected to the problem Mingrow is trying to solve for businesses.
The Future Will Not Belong Only to Companies Spending Millions on AI
There is an important detail in Tan's revelation that deserves attention. He acknowledged that spending $100,000–$200,000 annually on AI tokens is an extraordinary privilege and that the technology remains too expensive for many people to use at that scale. That raises a larger question: if AI agents are becoming increasingly useful, how can ordinary businesses benefit without building expensive, highly customised AI systems? The answer will increasingly depend on accessible, integrated business platforms. Small and medium-sized businesses may not need a personal AI infrastructure as sophisticated as that of a Silicon Valley executive. What they need are practical AI capabilities that solve everyday problems, fit into existing workflows and make business management simpler. This is where platforms such as Mingrow become relevant. The future of business AI should not be measured only by how much a technology leader spends on tokens. It should also be measured by how effectively a growing company can use AI to save time, improve coordination and make informed decisions.
The Bigger Picture: AI Is Becoming Part of How Businesses Operate
Garry Tan's AI setup offers a glimpse of what highly integrated AI assistance can look like. Claude Code, OpenClaw and Y Combinator's Stray Light demonstrate different ways AI can become part of everyday work. Meanwhile, Mingrow's 12-agent approach reflects how this direction can be brought into the everyday operations of growing businesses. The question is gradually changing from How can AI help me do this task? to How can AI help my business run better every day? That is a much bigger opportunity. As AI agents become more accessible, businesses that organise their information, connect their operations and adopt practical AI workflows will be better positioned to benefit from this transformation. Mingrow is built for that direction — bringing business operations and 12 specialised AI agents together, so business owners can spend less time managing scattered systems and more time building what comes next.
This article is intended for informational and educational purposes only. AI capabilities, integrations and levels of automation vary across platforms. Businesses should evaluate data privacy, security, permissions and human oversight before adopting AI-powered systems.