Growing an AI orchestration platform to $3k MRR in 4 weeks

Santanu Dasgupta built an AI orchestration platform and launched it into a crowded market. Four weeks later, Meerkats.ai's MRR is over $3k.

Here's Santanu on how he did it. 👇

20 years in GTM

I’ve spent 20 years working on go-to-market for SaaS companies, across the US, Europe, and India — starting out as a developer at an Oracle mobile database spinout in the Bay Area, before moving into GTM and growth.

I later worked with startups and enterprises on scaling demand generation and revenue at Gartner Consulting and Tata Consultancy Services, where I saw firsthand how much GTM still depends on manual work, fragmented tools, and agency-heavy execution.

That insight led me to start an AI orchestration product called Meerkats.ai. We’re essentially building a digital growth agency in software. Meerkats replaces a lot of the repetitive work SDRs, marketers, and agencies typically handle — capturing and enriching relationship data, generating leads, running campaigns, and following up — all from a simple chat interface.

Startups' GTM teams are selecting Meerkats.ai over platforms like Claude Managed Agents and OpenAI Agent Harness, as they want flexibility in model choice (bringing the cheapest model that is appropriate for the task). Also, they may not have the in-house talent to build in Agent Harness, so they are offloading the entire task to Meerkats as a “Service as a Software” or even outcome-based service.

We launched the platform four weeks ago and are currently doing $3k+ MRR. To keep the lights on, we got funding from the University of Chicago Polsky Center as part of the Alumni New Venture Challenge. We also offered Agency services as part of the platform to fund operations. And we got generous credits from Azure, OpenAI, and Anthropic towards model costs.

Building V1

We built the initial system with the following components:

1. A spreadsheet UI for viewing lead lists and an enrichment platform.

2. Infrastructure to scale enrichments to 50K+ rows in under 30 minutes.

3. Integrations with third-party servers, and enabling dynamic code execution based on user requirements.

4. Autonomous Agents with full manus-like capabilities, including memory management, sandbox code execution, evals, tool selection, Skills.md, and files for a knowledge base, helped us create an end-to-end native AI platform. The entire setup is sometimes called an Agent Harness and enables Agents to work reliably at scale across multiple workflows. These systems can plan and execute open-ended tasks, pick the right tools and models for the task, handle execution errors, and request human attention when required.

The initial product leveraged AI models like Claude, Gemini, Codex, and agentic frameworks such as Claude Skills SDK, Codex, LangChain, Crew AI, AutoGen, and Google Cloud Platform (GCP). It also drew inspiration from the capabilities of OpenClaw.

As far as the rest of our stack, it's:

• Supabase

• Google Cloud Platform for scaling

• Fly.io for sandbox code execution

• React frontend and Node.js server

• MCP servers and CLI endpoints in GCP containers

We chose Supabase because it offers the wonderful feature of Row Level Security, as well as powerful features like built-in authentication, real-time updates, and MCP server hosting. It felt much more modern than Mongo, hence we migrated.

Business model and growth channels

Our business model is consumption-based, factoring in the number of enrichments and actions an LLM performs and the task's complexity. We aim to make it very easy to start a growth flow, usually at the top of the funnel, and then help agencies or businesses with other revenue-impacting workflows. We are also seeing demand from customers for agencies to package our service as an “Outcome-as-a-service” where the customer pays the agency for a particular outcome or task completed — similar to what they would get if they hired a VA, but an Agent completes the task.

We have three growth channels:

• We combine cold and targeted outreach to agencies in our network. We specifically target agencies that run marketing campaigns for their clients without using AI.

• We run educational events online and offline to educate agencies on how to best leverage AI Agents for revenue.

• We attract users through LinkedIn posts.

If I were starting over, I'd do more content marketing and build bigger audiences up front.

Threats and opportunities

Suggested header: Why content marketing is crucial for startups

The biggest challenge for us was how rapidly AI capabilities developed — the competitive landscape and differentiation changed completely.

Fortunately, we made the right architectural choices, releasing the right product at the right time.

The key lesson was this: Navigating rapid tech changes is tough for startups, but even tougher for incumbents. Startups and indie hackers should view this disruption as an opportunity, not a threat.

In my case, automation technology has been relatively stable over the past 10 years, with players like Make.com, n8n, Zapier. This stack is now considered legacy with AI orchestration, new models, and orchestration frameworks.

Find the bottlenecks

Here's my advice:

1. Understand agentic behavior: Focus on agents that handle processes end-to-end, not just AI wrappers.

2. Identify customer problems where AI solutions can drive revenue and cut costs.

3. Find tasks that are repetitive and are done manually — convert them to Agents. Start with one high-value task that is painful, directly impacts revenue, or customer experience. Use that as a wedge to rapidly expand to other tasks.

What's next?

My future goals are to build an AI-native company at scale with minimum headcount that delivers significant value to customers through rapid growth, selling more to existing customers, and reducing wasteful spend/headcount. To get there, we'll use our own agents for customer support and growth strategies.

I also want to improve agent reliability through evals and CI/CD pipelines.

You can learn more about our growth story on our blog, or connect with me on X and LinkedIn. And check out Meerkats!