Pivoting to a 6-figure MRR AI business

Flo Crivello raised $52M for a product that had momentary success before tanking post pandemic. But as the first product floundered, he found an opportunity and pivoted to a completely new product. Two years later, Lindy.ai has a 6-figure MRR.

Here's Flo on how he did it. 👇

Post-pandemic pivots

Lindy is actually a pivot from my previous startup, Teamflow. We were building a virtual office for remote teams.

Our timing was excellent — we’d literally started working on it 3 weeks before the beginning of the pandemic. So, things went very well (for us) when Covid happened… and very poorly when it ended. 

At the same time as growth started to flatline, OpenAI released the GPT-3 API — this was around mid-2022, before ChatGPT, and certainly much before anyone talked of agents. We were very excited by GPT-3 and wondered how we could leverage it inside Teamflow. We came up with an AI meeting recorder.

At first, just summarizing our meetings. But then, our sales folks asked whether we could also make GPT-3 update their Salesforce after meetings. That’s when we realized we could make GPT-3 directly write the call to the Salesforce API.

It barely worked, but you could see the “AI employee” arrive if you squinted: LLMs weren’t just good to “generate” text, but also to take action! 

Then, ChatGPT happened, and our explorations in this area got us more and more excited. On January 1st, 2023, we pulled the trigger and decided to pivot completely to this new vision.

Today, Lindy is the “Zapier of AI” — a no-code platform letting you build AI agents to automate basically anything. Lead qualification, outreach, enrichment, customer support, meeting recording, CRM updating, etc.

We’re two years old and we already have a 6-figure MRR.

The "Idea Maze"

We were lucky enough to have raised a lot of money ($52M) for Teamflow — and we’d remained pretty disciplined in our spend, so we had most of it left over. 

Our funding story will probably make us a bit harder to relate to for a lot of folks here; I certainly don’t know how to build what we did for less than $5M, and I generally think founders raise too little money.

It costs a lot to build something great, and things always take at least 2 years to work.

We were a team of eight engineers, and spent a lot of time in what Balaji Srinivasan calls the “idea maze” before finally finding something that clicked. 

The first version of the product was literally just a chat — type what you want to do in there, and it does it. Little by little, we added the ability to create multiple “Lindies,” and to inspect and manually edit their workflows. We counted five major refactors of the product in two years before finally landing on its current form. 

We use a pretty standard stack: React, Typescript, Node, Mongo, all hosted on GCP (costs us a fortune, I hate it), GraphQL. 

The biggest technical challenge was architectural: It took us a very long time to figure out a way to “make agents work.”

Iterate fast and listen to your gut

It took us so many iterations to find something that worked. 

In hindsight, I wish we’d made the big changes faster. Like I said earlier, we went through five big technical and product paradigms before finding product-market fit, and I think we probably stayed ~2 months too long in each paradigm. That’s close to one full year wasted being insufficiently decisive! 

My learning was to make the big changes fast, no matter how unpopular, and to listen to my gut when I feel like something is fundamentally missing.

That last part is big. I’ve definitely found it helpful to listen to my intuition more. I think tech people, in particular, can fall into the trap of overthinking — we get stuck in mental mazes of our own making, instead of just looking at the simple reality that’s right there.

Usage-based pricing is the only option

We do usage-based pricing, charging “credits” which you consume as you use Lindy. We were very deliberate in adopting this pricing strategy, as we feel like it is the one best aligned with our vision of the “AI employee.” 

We really do believe that we will see the first “1-person billion-dollar business” in just the next couple years; in this world, we don’t think seat-based pricing makes any sense. 

It's also important to note that we started charging immediately. Partly because we had to — LLMs are expensive — and party because we wanted a real signal of product-market fit and willingness to pay.

It’s very easy for people to say they really want your product. Dollars are the only truth.

Revenue growth was steady at first, and accelerated once we found both product-market fit and the right go-to-market motion that went with it.

Our margins are healthy but not quite software-like, which I understand is typical in the era of AI: you actually do have unit costs, because LLMs are so pricey.

Launch often

SEO

We invested a lot in SEO from Day 1, and persevered for a long time, even though it was slow to show results.

Social media

I also have a reasonable following on X, and I constantly post about what I’m working on.

Launches

We do two or three launches per year when we have something big to talk about, and consistently get 10-50k signups from those. 

Here's our launch playbook:

1. We start preparing launches 3-7 days ahead of time and we record videos.

2. We post those videos to Twitter.

3. We ask friends to retweet!

I wish there was more to it than that. I've just spent an insane amount of time on Twitter over the last decade or so, and that's turned into an audience. 

Don't bother with IRL events

I’ve gone to conferences, sponsored hackathons etc, and have generally found this “IRL” stuff next to useless — it’s hard to justify this investment when you can spend the same amount of time posting on Twitter and getting 500k - 1M impressions.

Fat or lean: Choose one

Broadly speaking, I feel like there are two broad categories of strategy for startups: the lean startup and the fat startup. 

My understanding of the lean startup is to try as many things as possible, as fast as possible, and see what works. I don’t mean that in a derogatory way at all — it’s a recipe that seems to work, when you do it well. 

I’ve just not found this to work for me, because I really do need to feel excited and find meaning in what I’m doing, or I just can’t bring myself to do anything for more than a few months. 

So, I’ve found the “fat startup” model to work better in my case. Go after pure “execution risk” — do something ambitious and fundamentally hard to do, but that you know will work if you can somehow do it.

One analogy for the two different models is climbing Everest vs. digging for gold. Climbing the Everest is the “fat startup:” it’s really hard, you’re not 100% sure whether you have the raw ability to do it, but at least you know exactly what you have to do. It’s fundamentally a bet on yourself. And you know that if you do all these things, you will eventually reach the top. 

Digging for gold is less dangerous, more uncertain, and to some extent more luck-based. Which is why the game is fundamentally about maximizing shots on goal: Just dig in as many locations as possible. 

So, my advice to people is to:

1. Understand which game you want to play.

2. Actually play it.

Both games are fine; but playing a lean startup game in a fat startup way — raising a ton of VC money when you have a lot of market risk and don’t quite know what you’re doing — is a recipe for disaster. 

Be excited

I’ve become less apologetic about wanting to work on something that I’m excited about. I used to feel like this was overly self indulgent — like my only concern should be to build a successful business, even if it’s boring. 

Now, I realize that there is an existential dimension to these decisions for a founder. Your company isn’t just a regular job — if you’re lucky, you’re going to spend 20 years on this! You have to be able to find meaning in it, and you’re not going to be able to go through the rough patches if you don’t have that to carry you.

Take care of your psychology

On that note, it’s going to sound cliché, but I have, over time, realized the importance of taking care of your own psychology.

I try to work out at least three times a week. I spend time in nature when I can. And I try not to cut too much on social events and hanging out with friends, no matter how busy work gets.

What's next?

We’re looking to 10x the business this year — and we’re on track!

We’re going to do that partly through scaling go to market — hiring more sales people, more customer success people, doubling down on SEO and other marketing programs, etc…

And partly through massive improvements to the product — we’ve got at least four major new features coming this year, each of which we expect will make a real difference in the business.

You can follow along on Twitter! I'm also on LinkedIn. And check out Lindy.ai.