The New Garage: OpenClaw And India’s DIY AI Agent Boom

OpenClaw, an open-source agentic automation framework that recently surged in developer popularity before being acquired by OpenAI

Across India, data scientists, startup founders, and engineers are using OpenClaw to build personal AI “employees” – systems that autonomously solve daily challenges

OpenClaw’s appeal lies in autonomy, but costs scale unpredictably, setup is complex, and security risks are significant

Added to Saved Stories in Login VIEW SAVED STORIES .inc42-toggle-item-popup { display: none; position: relative; } .toggle-item-close { text-align: end; padding: 8px 12px 0px 10px; position: absolute; right: 0; cursor: pointer; } .toggle-items-content-main { display: block; position: relative; top: 27px; left: -204px; border-radius: 12px; background: #FFF; box-shadow: 0px 4px 24px 0px rgba(100, 100, 100, 0.25); width: 435px; height: 115px; } .toggle-items-content { display: flex; align-items: baseline; justify-content: center; padding-top: 22px; } .toggle-items-content .items-content-text .h4-saved-story{ color: #000; font-size: 20px; font-style: normal; font-weight: 700; line-height: normal; text-transform: capitalize; margin: 2px 0 10px 6px; } .toggle-items-content .items-content-text .myInc42-plus-dark { width: 100px !important; } .toggle-items-content .items-content-text .myInc42-light { width: 80px !important; } .toggle-items-content .items-content-text img{ height: 22px; } .view-my-feed-btn { width: 100%; text-align: center; display: flex; justify-content: center; } .view-my-feed-btn a { width: auto !important; } .view-my-feed-btn button { border-radius: 4px; background: linear-gradient(180deg, #DA1B4D 0%, #E23026 100%); color: #fff; font-size: 12px; display: inline-block !important; min-width: 162px; width: 162px !important; height: 34px !important; font-style: normal; font-weight: 700; line-height: normal; padding: 10px; cursor: pointer; } .CustomIconStyled { position: absolute; right: 180px; top: -80px; } .SubDropdownModelShare .sub-arrow-icon { width: 76px; height: 80px; position: relative; overflow: hidden; box-shadow: none; } @media (max-width:767px) { .toggle-items-content .items-content-text .h4-saved-story{ margin: 4px 0 10px 6px; font-size: 18px; } .toggle-items-content { align-items: center; } }

A new class of AI users is emerging not inside corporate labs, but in home offices and weekend experiments. At the center of this wave is OpenClaw, an open-source agentic automation framework that recently surged in developer popularity before being acquired by OpenAI, with founder Peter Steinberger joining the Sam Altman-led AI startup.

Previously known as Clawdbot, then Moltbot, OpenClaw promises to be the ‘AI that actually does things’, which ranges from managing your calendar to building a Reddit-style social network full of other AI agents.

Across India, data scientists, startup founders, and engineers are using OpenClaw to build personal AI “employees” – systems that autonomously solve daily challenges. Meta’s chief AI officer Alexandr Wang even said that personal agents are AI’s next opportunity, in a recent interview with Economic Times.

From Budget Bots To Virtual Companies

Jaipur-based engineer Raman Choudhary left his role as a senior software engineer in Bengaluru to join his dentist father’s venture and address inefficiencies he saw firsthand.

In November 2025, he founded Dent Node, a dental lab management platform designed to streamline how labs handle patient scans from oral scanner companies like iTero and 3Shape. Since each scanner operates on separate portals, labs working with multiple dentists must manually log in, download, and organise files. Dent Node simplifies and automates this fragmented workflow.

His one-man venture found a supporter in an AI agent. Soon after discovering Openclaw, Choudhary established a personal agent to automate the workflow. The agent today handles portal logins, navigation, and file downloads automatically, reducing manual effort, while offering timely updates to Choudhary on a WhatsApp chat. He also installed ready-made ‘skills’ to create better marketing content.

“If OpenClaw lacks a function, you install a skill, whether for influencer-style content, video editing, analytics, or management automation. This modular architecture expands its capability without requiring complex workflow engineering,” he said.

On the other hand, fourth year BTech student Aditya Pratap has built a multi-agent orchestration system where OpenClaw routes tasks across eight specialised agents handling sales outreach, GitHub monitoring, database queries, and daily briefings. “The system integrates with Slack, GitHub, Gmail, PostgreSQL, Twitter, and WhatsApp but runs in sandboxed environments to avoid exposing company data,” said Pratap.

Beyond work, Bengaluru-based Soham Mullick, lead data scientist at cloud database company Teradata, built a customised agent to track updates from the Union Budget 2026 announced on February 1. He used Openclaw to make a budget tracker that pushed updates every 15 minutes tailored to Karnataka announcements and salaried-class impact. “It definitely helped me keep a track of such a complicated topic almost real-time without being glued to TV,” said Mullick.

He is now experimenting with inbox automation that auto-labels low-priority emails for deletion. But experimentation isn’t cheap: Mulllick spent nearly ₹1,000 in API credits in a single day, even while using discounted providers.

Even high-profile founders are experimenting. Harshil Mathur, cofounder of Razorpay, built a mini ‘Jarvis’ using OpenClaw and MCPs with tightly restricted permissions. His system reads sleep and recovery data, generates morning briefings, orders food on Swiggy, auto-sends gifts, and even detects when he is in bed to turn off lights.

The most interesting insight came from correlating Swiggy food history with wearable health tracker WHOOP’s recovery metrics. In his words, stitching personal data with real-world actions feels like ‘programming reality’.

Travel Platforms Stranded In The Fog Of War

New Ventures, Old Hands: India’s Startup Founders In Encore ...

Indian Startup IPO Tracker 2026

“What amazed me most is that it connects to WhatsApp. You just tell it what to do, and it goes and does it,” Razorpay cofounder Mathur told Inc42.

“It is like an untrained intern, extremely intelligent, but inexperienced.”

In his view, the system can reason through complex problems on its own. If given a task, it figures out a path and attempts execution, occasionally fumbling but often succeeding in surprising ways. He believes the future is close at hand — one where everyone has a personal AI agent acting like an executive assistant.

“It’s costing me a bit more right now because I’m using the Opus model, which is the most expensive option,” Mathur said of the cost incurred. “Initially, I was spending almost $100 to $200 a day because I was using it for everything.”

To reduce costs, he implemented model routing — assigning a cheaper model for simpler tasks and reserving Opus only for complex reasoning, which can bring costs down by nearly 70%, reducing daily spend to around $30–$50. The Razorpay CEO added that further efficiencies are possible. Developers can host their own models to cut costs dramatically, in some cases running systems at near-zero marginal expense.

Agents as Organisational Simulators

US-based AWS senior engineer Manu Mishra is attempting to simulate an enterprise itself. Mishra said, “The idea is to simulate an enterprise structure where a single human acts as the CEO, supported by a fleet of OpenClaw agents functioning as virtual employees.”

When someone deploys the solution, they assume the role of CEO. Based on the configuration, they can define roles such as VP of Engineering, VP of HR, VP of Marketing, and so on. The CEO interacts with the VPs, who can then break down high-level instructions into smaller tasks and delegate them to multiple agents — similar to how a real organisation operates.

Built using OpenClaw and AWS CDK, he’s experimenting with multi-agent memory to mimic real organisation for solopreneurs.

Hyderabad-based engineer Akash Avvaru is already running AI ‘employee’ agents for his freelance work where he offers web development and consulting services. Using Openclaw, along with other AI tools, Avvaru has built eight core agents that together run his micro-agency.

The lead generation agent scrapes platforms like LinkedIn, qualifies prospects, pulls verified data via sales engagement platform Apollo.io, scores leads, and pushes them to my CRM. Another agent writes personalized CEO-level cold emails, while others manage infrastructure through Resend, warm up 40 email accounts, track deliverability, and send over 2,700 emails without hitting spam. Additional agents handle CRM updates, content, LinkedIn posts, reporting, and calendar management. “It’s not just automation — it’s an AI-run micro-agency with me as the CEO,” Avvaru said.

OpenClaw’s appeal lies in autonomy — it interacts with browsers, dashboards, and apps like a human, using screenshots and dynamic navigation. Its emerging skills ecosystem allows modular extensions. But costs scale unpredictably, setup is complex, and security risks are significant.

It may be noted that OpenClaw itself is open source and free. However, the costs come from infrastructure and AI usage. Several estimates show that basic personal setups cost roughly $6–$13/month, whereas Small business workflows run $25–$50/month with mixed model routing. Heavy automation with premium models can exceed $200/month.

Further, OpenClaw is raw infrastructure which is powerful but also flaky and insecure by default, as RazorPay’s Mathur puts it. It is not for non-technical users and should never be given access to anything you wouldn’t be comfortable leaking, he added.

The open-source AI agent framework recently gained notoriety for an incident where it deleted the personal email inbox of Summer Yue, the Director of AI Alignment at Meta. Yue tasked OpenClaw with suggesting which emails in her overflowing inbox should be archived or deleted.The agent however, trespassed critical instruction to confirm before acting and began indiscriminately deleting tens of thousands of emails.

According to advisory and consulting firm Greyhound Research, while Openclaw may not be insecure by intent, it collapses the boundary between intelligence and execution faster than governance is evolving. Most hobby developers see an assistant.

“What they are actually deploying is an orchestration runtime with inherited identity, persistent memory, tool invocation rights, and cross system reach,” said Sanchit Vir Gogia, founder and CEO.

He further added that from a regulatory standpoint, particularly in India, this risk has legal implications.

With OpenAI now in the picture, questions remain about pricing, openness, and enterprise guardrails. OpenAI’s alignment with OpenClaw shifts competition from pure model capability to control over orchestration that includes how models connect to tools, memory, identity, and workflows.

Even if OpenClaw remains open source, roadmap gravity, integration defaults, and ecosystem incentives could tilt neutrality toward its primary backer, increasing concentration risk at the orchestration layer, Gogai said.

© Inc42

visit website