I Built A $300K/Year AI Business For Less $50 Bucks
Hello! Who are you and what business did you start?
I’m one of the co-founders of My AskAI. We sell AI customer support agents to SaaS businesses to help them deflect and decrease their support ticket volume.
You add your help docs (Zendesk, Intercom etc), connect any other knowledge sources (Drive, Notion, Confluence) and then we create your AI agent trained only on your docs to answer your user’s questions.
If the AI can’t answer it passes the conversation over to your human agents so it’s win-win!
You can use our widget on your site or one of our integrations with Zendesk, Intercom or HubSpot to get AI answers to your customers. Businesses tend to pick us over those companies’ internal offerings based on price (we’re c.5x cheaper) and our response quality.
Right now we’re making $25k/mo and competing with some of the biggest customer support platforms.
What's your backstory and how did you come up with the idea?
When I was young I knew I wanted to start my own business.
I’d watch Dragons’ Den and The (UK) Apprentice religiously, trying to absorb whatever I could (probably a lot of bad advice in hindsight…).
But I wasn’t super confident, I didn’t have anyone around me who was entrepreneurial and I never felt I had the ‘right’ idea, one that I could commit to.
I wasn’t one of these kids who started a lemonade stand when he was at school (however I did make a tidy sum selling kids’ car seats on eBay when some new legislation came out).
I followed the usual middle-class kid’s path: went to university, got a job in finance (an accounting firm), and then settled in for the long haul.
But I couldn’t shake the entrepreneurial urge. I set up teams and initiatives at work to try and use new technology, break into FinTech and a bunch of other things.
I even took some time out to start my own sports app with a friend to help find others to play sports with, but never got anywhere.
After 9 years I had a realisation.
I asked myself:
“If it all went to plan where I was and I achieved everything I wanted in that career, would I be happy?”
When I realised the answer was a resounding no, I started drafting my resignation.
I was one step away from being a Partner in the firm.
So that was it, I had 3 months’ notice and 3 months of cash beyond that to re-invent myself, start a business or find something else to do.
Turns out it is harder than you think (or than I thought) to go from being an accountant to a tech product manager without any previous experience. Who knew?
So I started contracting at tech companies, first a renewable energy startup, then an InsurTech.
I would work for 6 months contracting, save, then try out my own ideas for 6 months, or until I ran out of cash, then go back to contracting.
It sounds pretty inefficient but I knew I worked at my best when I dedicated myself to something, rather than splitting my attention. It also put a bit of healthy pressure on me and gave me a deadline to come up with something.
During that time I started following more “builders” on Twitter (X). Seeing people every day who were creating businesses from nothing, on their own, and taking on VC-backed companies was hugely motivating to me. I’ve learned more from that community than any book or course I have ever taken.
After a couple of years of freelancing, a friend from one of the companies I was working at asked if I’d meet his friend to give him advice on contracting as a Product Manager.
He’d founded and run a VC-backed travel InsurTech startup which had been a victim of COVID-19 and just had to shut down the company so was looking for his next role.
We met and immediately hit it off, talking about this new technology called “GPT-3” which had just launched.
Very quickly we started coming up with ideas, thinking about how we could capitalise on this exciting new tech. (To this day I still don’t think I ever shared any wisdom about contracting).
First, we made a tool that used fine-tuned AI models to write university personal statements. We got a few sales but at that time OpenAI were manually approving use cases, and they didn’t like ours, so we never properly launched.
From that, we moved on to making a tool called the No Code AI Model Builder (catchy eh?). The idea was you could upload a bunch of training data and it would spit out a fine-tuned model, all without having to code at all.
We made it in a few days as part of a Ben’s Bites Hackathon and put it up on Product Hunt to see what people thought.
Within 24 hrs we had made 2 sales and had 2k visitors on the site. Within a week we were featured on a huge German tech blog.
We started to develop it further but soon found that the friction required to make fine-tuned models was just too high.
What people actually wanted to do was dump data into something and get magic out. This was further validated by all the one-off “Chat with your podcast/book” projects developers were starting to spin up on Twitter.
We knew there had to be a platform for them.
So in 2 weeks (half of which was while on holiday in Thailand), Alex managed to spin up an MVP using the no-code platform, Bubble.
It allowed someone to upload a file and then start asking questions, all in natural language.
Both having professional services backgrounds, our minds raced with applications for it. Being able to query huge documents and sites to get answers without having to digest all the content.
We were confident in our intuition that this could be a hugely useful product, but we still wanted to validate demand.
So we added a pre-launch page as a fork within our No Code AI Model Builder product, asking users what they wanted to use the product for. If the criteria matched we would push them to a pre-launch sign up for what would become My AskAI.
In those 2 weeks we got $5k in pre-launch sign-ups (50 at $99).
This was more than the total revenue we had made for our current (functional) product.
We knew we had something viable on our hands, and we had around 3-4 months of savings each to make it into something that could support us living in London.
Take us through the process of building the first version of your product.
When we are building a new product from scratch we don’t like to go overboard with planning.
I will often draft up some low to mid-fidelity wireframes on Figma (I am definitely not a designer) that show the flow of the product, button placements etc.
This takes maybe a day in total with feedback and back and forth.
Both having backgrounds in Product Management we are quite good at the “MVP” mindset of ensuring we don’t go too heavy on features. We also do quite a decent job of keeping each other in check here too.
Once we have the basic features designed and scoped out, Alex will start building pretty much right away using Bubble.
This is one of the awesome things about using a no-code tool like Bubble - you can build and change things very quickly so it is like a functional wireframe.
I think this is one big takeaway for me - whether you are going to use code or not long term, building with no-code for your v1 is such a great edge. It allows you to respond quickly to user feedback and keep them engaged.
Once a rough v1 is ready, I’ll start testing. I basically try to find any way I can to break it.
We record all bugs and features on a basic Notion board to keep things tidy and ensure the scope doesn’t go out of control.
Then we’ll get 5-10 potential users of the product (either from an existing user base, mailing list, asking people on Twitter or using a user feedback platform) and do a bit of user testing.
We’ll be looking out to see if they understand what the product does, how it works and how to use it without input from us.
All in all, the v1 of My AskAI probably cost us $50 or so in software and 2 weeks in build time.
Right now we’re profitable, with 85-90% gross and 50-55% net margins. We get about 35k visitors to our site each month and have around a 3% conversion rate.
Describe the process of launching the business.
Before we launched we had already got $5k in revenue from our pre-launch sales.
So the launch was a question of how we would turn that one-off revenue into recurring revenue (as we had an ongoing OpenAI bill to fund).
To launch we:
- Posted on Twitter, where we had been trying to ‘build in public’
- Shared on LinkedIn and a few other platforms
- Launched on Product Hunt
- Posted it to as many AI directories as we could find
(For context here, this was February 2023, probably the peak of the AI hype (to date) so we already had the wind in our sails).
We went live and saw sales pretty much immediately and within 2 months we were up to ~$7k MRR.
At this point, we were still both living off our savings and figured that we would need to get to approximately $10k MRR to live.
Fortunately, we also managed to sell our previous products on MicroAcquire (the No Code AI Model Builder and the university statement generator for a combined $25k).
Doubly fortunately, our business costs were pretty negligible, with only a few SaaS subscriptions ~$100/mo and OpenAI costs (which scaled with our revenue).
I think if we were to go back and do it again, one thing we’d do differently would be to produce better-quality video content as part of the launch.
Other products that launched at a similar time got a lot more publicity partly because they shared better launch videos that showed off their product at a time when AI virality was a lot easier to come by.
We instead shared screenshots, which were nowhere near as powerful for displaying what was quite a new and exciting form of tech.
Since launch, what has worked to attract and retain customers?
Since launching 15 months ago we have tried multiple growth channels to find a repeatable channel.
To be honest, we are still looking,
One thing we have noticed in the AI space is that a lot of channels are temporary.
What I mean is that they work for a period before quickly being exhausted.
For example, pretty much all of these worked for a period before quickly becoming saturated or us realizing they weren’t a good fit for a B2B SaaS business:
- Influencer posts (LinkedIn/Twitter/TikTok) - we’re still yet to see good examples of these working well for B2B businesses and a lot of “AI influencers” seem to result in vanity views and engagement from our experience (we’ve done quite a few tests!)
- Newsletter sponsorships - we will see an initial bump, but this quickly falls off, they’re probably better for longer-term awareness than conversions.
- Reddit ads - these are very hard to do authentically, Reddit is a brutal place.
- AI directories - when the hype was at its peak these were great ways for people to find out about new AI tools, but now only the highly curated ones drive any traffic.
- Replying on Twitter to bigger accounts - this was a bit of a growth hack for a while, responding to large influencer accounts, but again, the hype has died down quite a lot now so you don’t get the same reach.
Instead, we’ve now narrowed our focus down to:
- SEO: we have an agency we work with for blog content and are trying to create more side product tools and directories to drive traffic.
- Social: we have found our best clients tend to come through LinkedIn, so we are going to be spending more time there using social selling tools like Breakcold.
- Google Ads: we have experimented quite a bit with this and never cracked it, competitors like Intercom and Zendesk drive the prices way up but we think there could still be some value in longer tail terms.
- Product marketing: we try to make sure we are selling our own product with email flows, newsletters, changelog updates.
- Reddit: we have a bunch of alerts set up using f5bot that tell us when someone is talking about a topic of relevance to us so we can jump in, this tends to work well as long as you are providing value in any engagements.
Most don’t speak about it, but AI products are pretty high churn. Customer churn rates in their teens are not unusual (even more so for B2C AI products).
Lots of people are still trying to figure out what they can use them for or just experimenting with the quality of their output.
We spend A LOT of time talking to and working with customers. This differentiates us from others by providing a much higher level of service than most competitors.
We also ship super quickly (thanks to our no-code stack). New features and fixes are released pretty much daily. This ensures we keep close to what our customers want and remain an “actively-developed” product (a rarity in the AI space).
Two more big things that have helped reduce our churn rate significantly are:
- Qualifying and focusing on a tighter subset of users. This meant we could build a product centered around their use case and make users more likely to stick around. Qualified users are also less likely to be “experimenting” and more likely to be looking for a long-term solution.
- Going upmarket. We have way more Enterprise customers now who we work very closely and have longer-term contracts with.
How are you doing today and what does the future look like?
Right now we’re profitable, with 85-90% gross and 50-55% net margins. We get about 35k visitors to our site each month and have around a 3% conversion rate.
We are focusing on B2B SaaS businesses but the massive prize at some stage would be to look towards e-commerce businesses too.
Most of the big chat providers are starting there (e-commerce) with their AI offering but volumes are much higher and the AI needs to be more integrated into other tools to be useful. We think our product today is much better suited to SaaS businesses.
We’ve also just released a localisation feature so you can now use My AskAI in a bunch of different languages.
Today, both Alex and I work from home. It’s pretty much 8am - 6:30pm on weekdays in terms of working hours now, with us only working outside of those hours if we want to or if we have an urgent issue to work on.
This has been great for Alex, especially as a new Dad.
All of our interaction is through Slack, that’s where most decisions are made.
If we get to any bigger decisions or things we disagree on we will tag them #meetup and then go through these once per week when we meet IRL somewhere for lunch and a half-day meeting.
This approach works well for us and means we can get through a lot of smaller decisions very quickly and autonomously.
We believe that given the way AI is progressing currently we should always be able to find ways to automate before hiring and would love to scale to be a $1m ARR+ business with still just the 2 of us.
Entrepreneurship is hard, it’s even harder when you don’t have people around you. Surround yourself with people you can learn from and people who motivate you. Go to meetups, engage on Twitter and Reddit and get inspired by the people one step ahead of you every day.
Through starting the business, have you learned anything particularly helpful or advantageous?
This is the first business I’ve had with serious traction so inevitably there have been a ton of lessons learned, here are a handful:
- Don’t underestimate the “follow-up”. We have landed contracts after no response on the 7th no-response email. People are busy, if you’re not a fit they’ll often say so.
- Focus, say no but don’t be so inflexible you miss out on random opportunities. While keeping a tight product focus is important, if we were super-strict we would have missed a piece of work that came about through an inbound LinkedIn DM that resulted in c.$100k of revenue for around a weeks’ work.
- Everything is harder if you don’t position yourself. If you don't have clear product positioning, saying what you are, but also, what you aren't for, everything gets a lot harder. Deciding on copy or language to use, marketing channels and strategies, design, and prioritization of features all become very difficult.
- Don’t let your ego get in the way of decisions. We were probably late to deliver a few features (choosing others over them) just because we weren't big fans of them ourselves, despite several people asking.
But we have also made some good decisions along the way.
This is largely because wherever possible we use data to make them. Regular A/B tests of our landing page (we use VWO) and even switching the core focus of our product to Customer Support in October last year are a couple that come to mind.
We were definitely lucky timing-wise. AI has boomed.
But we also put ourselves in a position to be lucky by exposing ourselves to this new technology and acting on it
We’re both pretty disciplined, focused people and have gotten into a few good habits:
- We always reduce products or features down to something that can be built in one day max. If it gets traction we go from there.
- Even though it is just the 2 of us we are strict about our roadmap and fixing bugs.
- We have also made a habit of ensuring when we build a new feature we tell people. We update our docs, add to our changelog and share in our newsletter. It’s such an easy win to make the most of the things you have already built.
What platform/tools do you use for your business?
OpenAI - We wouldn’t have a business without it (plus use ChatGPT multiple times/day)!
CarbonAI - They power a lot of our knowledge integrations, are super responsive and just great guys building a terrific product.
Portkey - This has been a real life-saver trying to debug LLM requests at times.
Slack - Alex and I are both in London but only meet up one-half day a week the rest of the time all our comms is managed through Slack and the occasional Google Meet, it’s also great that pretty much all of our discussion and decision-making is therefore documented for us to refer back to.
Notion - Pretty much all documentation for our business is stored here and it is where we both work from our respective tasks/roadmaps.
Typefully - We both use Typefully for social media scheduling, clean interface, doesn’t do too much and actually gets us to post (which does not come naturally to either of us!)
Bubble - Our entire front end of the product is built on this, it means we can keep pace with AI developments and respond quickly to user feedback. I can’t imagine building another way now.
CustomerIO - They make building communication and email flows super easy and have given me (as the non-technical cofounder) far more control over our customer comms.
Loom - Massive timesaver for us given how we work majorly asynchronously. We use it for sharing bugs with one another, recording feature demos and explaining things to users.
Browserless - Similar to Carbon they have been a key part of infrastructure since early on.
What have been the most influential books, podcasts, or other resources?
My First Million - Not so much direct influence in what we have done but it is insanely motivational hearing about other entrepreneurs and learning how they are building and selling their businesses (same reason I love watching the Starter Story videos while I eat lunch)
Zero to Marketing - A great newsletter by Andrea Bosoni that will teach you that marketing is not one activity i.e. social or email marketing, but that everything is interconnected and each business’ marketing strategy is unique to them. You can be inspired by others but rarely can you copy and paste an approach.
Twitter - I’ve learned a ton from following all these people, everything from marketing, positioning, SEO, design, MVPs, and prioritising, genuinely the closest thing you will get to a “Twitter University” is if you follow these people and listen to what they have to say:
- Andrea Bosoni
- Marc Louvion
- Sveta Bay/Dan Kulkov
- Damon Chen
- Andrew Gadzecki
- Pieter Levels
- Danny Postmaa
- Olly Meakings/Wilson Wilson
- John Rush
- Arvid Kahl
- Tim Bennetto
- Tibo
Advice for other entrepreneurs who want to get started or are just starting out?
In some ways, I’d love to start something all over again.
I’d love to use everything I have learned from doing it this time. You realize how little you knew before you had built something.
These are a few things I’d be telling “past me”:
- Entrepreneurship is hard, it’s even harder when you don’t have people around you. Surround yourself with people you can learn from and people who motivate you. Go to meetups, engage on Twitter and Reddit and get inspired by the people one step ahead of you every day.
- Don’t get hung up on things customers will never see. Forget about the tech stack, if you can build something with Zapier do it, if you can code do that. Don’t learn something new if you already can solve the problem with what you know.
- If you can, partner up with someone who is influential in your chosen market. Someone who already has connections and a network. Distribution is one of the toughest things and having someone who already has it is easily worth its weight in gold (or equity).
- Video is getting more and more important so embrace the cringe and start trying to put your face on a camera and get comfortable talking to people through a screen recording.
- Revenue per employee will only increase over the next 5-10 years, we can build much much more with way, way less, so don’t get hung up on hiring and look for tools to help you instead.
- Don’t over-research an idea, sometimes it is quicker just to build and put it out there rather than get hung up interviewing prospective users.
Where can we go to learn more?
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
Download the report and join our email newsletter packed with business ideas and money-making opportunities, backed by real-life case studies.
