How to Start an AI Automation Agency With Lindy 3.0

TL;DR

You can start an AI automation agency by inserting yourself into small and medium businesses and charging $500 to $5,000 a month, or $300 to $5,000 one-time plus $2 to $700 monthly. Lindy 3.0 removes the technical friction by letting anyone build an AI agent from a single prompt, so non-engineers can automate customer support, lead generation, and refunds.

Transcript

This is a service you could offer for a monthly fee just using Lindy on the back end. It's so easy to create AI agents. It's almost addictive. It's literally just you type words and just does it for you. You can have the entire flow completely end to end managed by AI agents. There's a dozen Silicon Valley companies right now raising tens of millio... Read More

Key Insights

• Lindy 3.0 lets anyone create an AI agent from a single prompt, meaning you type or dictate what you want to automate and the agent is built for you in a minute or two, removing the need to be an engineer or learn a no-code tool.
• An AI automation agency works by inserting yourself into small to medium-sized businesses, automating certain processes with cheap easy tools, and charging between $500 and $5,000 a month or $300 to $5,000 one-time plus $2 to $700 monthly ongoing.
• Computer use is the new capability where agents get a virtual mouse and keyboard to operate their own cloud computer, so they are no longer limited to available MCPs or integrations and can log into internal tools that lack an API.
• The biggest opportunity is being the expert in the room for business owners who know they need AI but have no idea where to start, since the technical friction that used to trip up non-technical founders has been removed.
• Creating AI agents becomes almost addictive once it is easy, and the host reports there are a dozen Silicon Valley companies raising tens of millions that are basically just an AI SDR agent.
• AI agents pair a dumb data source with intelligence on top, so the founder uses Google Sheets as his personal CRM while Lindy handles adding contacts, retrieving investors, and intercepting flight emails to suggest who to meet.
• One customer built a 400-node Lindy that acts as a fully autonomous SDR, finding leads, qualifying them, sending personalized emails, handling objections, learning from them, and A/B testing subject lines, built 80% by prompt and 20% by manual tweaking.
• The capabilities discussed are literally 30 days old, and the founders argue there is no excuse anymore not to use AI agents because acting sooner puts you ahead of your peers while it is still very early.

Questions & Answers

Q: How much can you charge running an AI automation agency?

According to the host, you can charge small to medium-sized businesses between $500 and $5,000 per month for automating their processes. Alternatively, you can structure pricing as a one-time fee of $300 to $5,000 plus an ongoing monthly charge of $2 to $700. The model works by inserting yourself into these businesses and using very cheap, easy-to-use tools to automate certain processes they use, positioning yourself as the expert in the room for owners who know they need AI but do not know where to start.

Q: What is Lindy 3.0 and what are its main updates?

Lindy 3.0 is described as the biggest update in the company's history as an AI automation platform. It has three big updates: team collaboration so you can invite team members and share your agents, an agent builder that lets you create an agent from just a prompt, and computer use. The two the founder is most excited about are the agent builder and computer use, which together let anyone build and deploy agents without being technical or knowing the tool.

Q: How does the agent builder let you create an AI agent from a prompt?

When you log into Lindy there is a text box where you either type your prompt or press a mic button to dictate it. The founder recommends dictating because you can give more detail and even ramble, and agents are good at making sense of what you said. You describe what you want to automate, such as customer support, content generation, or lead qualification, and a minute or two later Lindy creates the agent. If it is not exactly right, you send another prompt to iterate and it improves.

Q: What is computer use for AI agents?

Computer use is the idea that agents can now use their own computer in the cloud, having a virtual mouse and virtual keyboard. This means agents are no longer limited by the MCPs or integrations they have access to. As long as a tool is accessible from a web browser, the agent can use it, even an internal company tool. The founder notes teams who wanted to automate customer support were previously blocked because their internal refund tools had no API, and now agents can just log in and use those tools directly.

Q: Can an AI agent issue refunds automatically from support emails?

Yes. The host gives the example of a prompt saying that every time an email arrives to a support address asking for a refund, the agent should go into Stripe and issue the refund. The founder confirms this is exactly the kind of use case that works. However, the prompt cannot be too broad; the agent needs to learn and know how to respond to certain specific inquiries rather than handle everything vaguely, so you define the conditions and the action clearly.

Q: How can you use Google Sheets as a personal CRM with AI agents?

The founder uses Lindy to manage his personal CRM backed by Google Sheets. He sends a prompt like help me manage my personal CRM, and Lindy clarifies its plan: adding a contact means adding a row to Google Sheets, and retrieving a contact means pulling data from the sheet. After he confirms, the agent is created in seconds. He also added a workflow that intercepts flight confirmation emails from his inbox and emails him who he should meet up with at his destination, treating Sheets as a dumb data source with agent intelligence on top.

Q: What was the most impressive Lindy the founder has seen?

The founder describes a customer's Lindy with roughly 400 nodes as by far the most insane one he has ever seen. The user built about 80% of it using the prompt-to-agent feature and 20% by manually tweaking it. It functions like a fully autonomous SDR that finds leads online, qualifies them, decides who to engage, researches them, and sends personalized emails. It then handles back-and-forth, asks why uninterested people declined, handles and learns from objections, and A/B tests emails and subject lines.

Q: Why do the founders say now is the time to start using AI agents?

They stress there is no excuse anymore not to use AI agents because the technology has removed the friction that once required being an engineer or learning a no-code tool. The capabilities discussed are literally 30 days old, so it is still very early and people underestimate how early it is. The sooner you act, the more you will be ahead of your peers. They also note a dozen Silicon Valley companies are raising tens of millions building what is basically just an AI SDR agent, signaling the size of the opportunity.

Summary & Key Takeaways

• The host argues one of the biggest current opportunities is starting an AI automation agency: you insert yourself into small and medium businesses, use cheap easy tools to automate their processes, and charge $500 to $5,000 monthly, or $300 to $5,000 one-time with $2 to $700 ongoing per month.

• Lindy 3.0, called the biggest update in the company's history, adds team collaboration, an agent builder that creates an agent from a single typed or dictated prompt, and computer use that gives agents a virtual mouse and keyboard to operate cloud computers and internal tools without APIs.

• Real examples include issuing Stripe refunds from support emails, managing a personal CRM in Google Sheets, and intercepting flight confirmation emails to suggest local contacts. One user built a 400-node autonomous SDR. The founders stress the technology is only 30 days old and it is still very early.