Founder Stories: Steve from Lorikeet

Steve Hind is the co-founder of Lorikeet- an AI support agent capable of handling complex, multi-step support requests while freeing human support workers to handle higher value tasks.

Since connecting Lorikeet with their first engineering hire at the end of 2023, they’ve gone on to onboard Eucalyptus and other multinational tech companies as paying customers and grow their team to 17 staff in a year!

Read on to learn about:

• How Lorikeet’s customers are using AI agents increase customer satisfaction handle
• Steve's journey to build Lorikeet and raise funding
• How they found product market fit in under 6 months
• Their philosophy for making great new hires
• What questions job seekers should ask when assessing an opportunity at a startup

‍

Tell us a bit about yourself?

I’m Steve. One of the co-founders of Lorikeet. We build AI agents that solve complex customer support queries for businesses.

When you deploy AI to customer support, the way many other systems work is that they look up some material and summarise it back to the customer. What we’ve been able to do with Lorikeet is enable the AI to solve much more complex multi-step processes.

For example take a request like “Where is my refund?” With a support request like that, there is going to be a lot of conditionality about the answer. It will depend on the status of your account; the status of some transfer or a multi-step process, for example you may also have to help them cancel something, etc. That’s the core of what Lorikeet helps our customers do:. Train the AI to solve complex multi-step processes.

Now if you are a simple SaaS or e-commerce business, you may have none of these complex requests - all of your customer’s questions will be - “How do I do this thing” and you can answer it by summarising the steps on how to do it from your help centre.

Once you get into sectors like finance and health however, much more of what you do will be highly complex. And that complexity can be procedural, but it can also be regulatory. For example, there are certain things you can only say to certain people in certain ways. In these cases, attributes like guardrails and the steerability of AI become not just important, they become make or break.

We have a number of customers who tell us: “Look, I assumed we'd never be able to use AI in our support operations until we understood the way your system works.” The combination of guardrails and that really high quality makes it a pretty unique offering in the market.

What inspired you to start the business?

I love building software. Ever since I started doing it, it’s always felt like magic. The work to understand what users need, to know what problem the business is solving and then to craft the software to solve it in a way that's scalable and repeatable has always seemed really fun. I felt like I learned the ropes watching really great founders and leaders do that at companies like Stripe and Watershed and was itching to give it a go for myself.

What did you do before founding Lorikeet?

I've done a bunch of things. I got a law degree. I worked as a management consultant. I went to the US to do an MBA and ended up staying there for nine years after. I worked at a hedge fund and was doing currency research and trading and then was itching to get into startups. Moved over to Stripe and into product management and then went from Stripe to a startup called Watershed that was founded by a bunch of ex-Stripe employees and worked on a cool carbon measurement product there.

After trying a lot of things in my career I settled into product and this love of building software and in parallel I sort of taught myself to code. I would never give myself a job as an engineer, but at least I got to the point where when we were building the early Lorikeet product, I was regularly shipping features, though now I'm at the point where I do it only very occasionally.

You founded Lorikeet with Jamie, your co-founder. How did you get together to launch it?

Jamie and I have a bunch of very close friends in common, but hadn’t spent a lot of time together until about 2023. We connected and there was this great confluence of interests. I had worked a lot with operational teams as a product leader and was really obsessed with how AI could unlock them being more effective. Jamie had spent the last few years at Google Brain as a research tech lead on factual grounding in large language models, i.e. preventing hallucinations. He was itching to get back to building something and away from research. That combination made things come together very naturally.

Also by virtue of having so many friends in common, from the beginning I think we had very similar priorities and values and clicked very quickly. It’s been really fantastic working together over the last year and a half or so.

What triggered you to both start the business?

One of the tricky things when you’re starting a business is you’ve gotta think about the financial pros and cons. I think for a lot of people, whether you want to do it can be irrelevant because things just might not stack up financially.

There are probably two things that are important to ask. Jamie and I are both people with kids and mortgages. So one of those things is having had a bit of a career runway to build up a bit of excess capital. For me that was being able to sell some appreciated stock I had in Stripe. For Jamie that was spending a good amount of time at Google and building a relatively stable financial position.

The other one is if you have a background that investors are excited to fund, the terms on which you can be a founder are actually pretty good. You can raise money pre-product and pre-traction. You can pay yourself a salary that stops you being in financial stress from day one.

On those financial terms, it felt like a great idea because if the business were to fail, we would have made less money, but we're not in financial stress to go take the risk and so that seems like a zero downside, lots of upside bet to take.

Tell us a bit about your first round of funding and how you approached it?

I was still in the US when I started exploring funding. I knew I was going to move back to Australia and I knew bootstrapping a local network would be really important. I got a bit of advice from angel investors I knew in the US and approached a handful of Australian funds, just really informally, saying: here's who I am, here's what I'm interested in, I'm thinking about raising some money. Can we chat?

I think I went from that initial outreach to term sheets and offers in a couple of weeks. It's just kind of lucky that it was a fast process. I guess my big reflection is at the pre-seed stage, a lot of whether you're fundable is determined before you ever reach out.

The investors who offered me money explicitly said at the time: we don't really understand the product. I was pitching a slightly different vision to what we ended up building with Lorikeet - But you know, I guess I had the kind of experience that made them want to take a bet. That's why I think it's a bit pre-determined even before you reach out.

What was your initial product idea and how did it evolve?

What we were interested in initially and started prototyping around was a meta productivity tool for operations teams. We wanted to  help operations teams understand what their processes were so that they could improve them.

The thing we learned though was that although meta productivity insights about the work is very interesting, the thing that is most helpful is actually just doing the work. So that's what led us to hone in on a more specific persona, which is support teams as opposed to all operations teams, and then having a more direct impact which is: do the support work.

What did you do after you raised your first round of funding?

The first thing we did was just go sit with the patient support team at Eucalyptus, a global telehealth company. We were really fortunate they were willing to host us in their office, a few days a week and we just watched them work. We prototyped really quickly, like I think we rolled out a product to them in less than two weeks from when we started. We weren’t spending months observing and debating academically. It was like: Watch. Ship something. Look at how it gets engaged with. Look at the feedback we get. Tweak it. Ship a new thing and experiment.

That tight loop of having the discipline to build and ship was really useful to understand what we needed to understand. It took us two months to figure out the original idea we were pursuing was not seeing the type of demand you would want.

And that's what led us to say: OK. I think we need to attack the problem more directly by going straight to doing the work, instead of focusing on meta productivity. Once we had that, it was almost night and day. The amount of demand and interest was a step change different and it was a signal we were on the right track.

How did you get into Eucalyptus’s offices to pilot your ideas?

It’s always the case when you're doing B2B that you want to leverage the relationships that you have. I’d worked for a couple of years as an advisor to some of the exec team on product topics. When I told them I was starting the business and talked about the concept, they said: Ok, we'd be up for experimenting with you. That's what got us connected.

How have things gone since you pivoted?

So we made that pivot after two months. We then took about two months to get the first version of the product deployed. We just wanted to get it live and start learning from reality as fast as we could. From there we expanded the design partners we were working with over a couple of months period and fleshed out the feature set. By about the middle of 2024, we had a feature set that clicked and started to win deals pretty consistently. I think that’s when you feel like you've gone from offering market-fit, meaning people are very excited about the thing we’re offering - to product-market fit, meaning we have a product that can deliver that offering.

Since then we’ve been focused on scaling up and gone from very few customers to dozens of customers. Our revenue is up about 4x in <4 months since launch.

It's amazing that you’ve gotten to revenue within such a short time.

Which is an intentional strategy. I think there is no better test in B2B of whether you're doing something valuable than asking people to pay for it, and ideally asking people to pay an amount that feels a bit uncomfortable. That's going to smoke out really quickly the difference between your idea being interesting and your idea being valuable. Part of the reason we were able to identify and pivot from our previous idea quite quickly is that we were asking people to pay. We weren't just skating by having people compliment us on having something interesting. We were forcing it. Like, are you confident enough that this will create enough value that you will pay. I think it’s a great way to smoke out whether you're on the right track.

What impact do you see your product having on your customers?

I think the critical thing is the companies we’re working with are all using us as a way to scale high quality support not to save costs or cut headcount.

If companies come in with the intention of using AI to cut costs, they're going to deliver a crappy experience. But the reason is that their goal is cost cutting and not delivering a good experience. AI when used well, when the system is sufficiently good, is actually a tool to scale better experiences than you can with a human team because the AI can operate 24/7, and it can scale up and down as you get spikes in demand.

The angle of AI as a tool to scale good stuff is a really productive thing to go selling into. And we're kind of fortunate that all the customers we work with are using it for that reason.

Have you seen your customers change their support teams as a result?

We tend to work with high growth businesses. On a counterfactual basis, their support teams are smaller. Maybe they’ve doubled their business in a 12 month period, but haven’t had to to grow their support team because they’re using Lorikeet.

What we see extremely consistently is that the reduction in human manual labor and repetitive support tickets is redirected to higher leverage activity in the business. So we've not worked with anyone yet who's like I've gone and done a round of layoffs because of Lorikeet. And no one really wants to either. They all have the sense the people on the team could be doing something more valuable and they're keen to release them to do that.

By way of example, we've seen people reallocated to proactively reach out to customers on trials and help them learn the product and convert, so they go from a cost activity to a revenue generating activity. What we’ve seen is instead of expecting a support representative to close 300 tickets a day, they are expected to close 50 a day and they are given the 50 that are really complicated. Then your CSAT overall goes up because the AI's handling a whole bunch of requests very well at the same or higher CSAT. The 50 each day that are really hard, instead of getting done in a big rush, are getting done by people more thoroughly. So the CSAT for those ones comes up as well. That's the kind of pattern we typically see.

I'm sure there are huge companies that are going to roll out AI and just lay people off. But it’s because their attitude is saving cost, not improving experience. And I bet their AI will be crappy as a result.

What does your team currently look like?

Right now we're a team of 17 split into three camps. In product and engineering, we have eight people. We have a general business operations team where people take on projects to scale functions we don't have yet like sales, growth and success and incubating those functions. Then we've just started building what we call a forward deployed engineering team. These are folks with strong technical backgrounds who are really excited to work day in, day out with customers and they sort of work on training the AI, testing the AI, helping customers get the most out of the platform. But in a really hands on and technical motion as opposed to a more customer success motion. That function has worked really well. We’ve hired a couple of superstars there and we’re going to keep scaling it.

What’s your approach to hiring and what's worked for you?

In general when you're a startup, you've got to have a hypothesis about why you can get really great talent.

When I was working at Stripe, we didn't have to arbitrage the talent market. We just had a bunch of really great PMs showing up cold, wanting to work for us and we could go down and choose them.

When you're a startup that you basically no one has heard of, you’ve got to have some hypothesis on how you are going to win the best talent and there are good ways of solving that and bad ways of solving it. Are you going to give them a fancier title so they'll get to be like a director at your company, but they wouldn't be at some other company? I think that's a really bad solution.

On the flip side, maybe the arbitrage you’re going to do is to offer them more autonomy. Or more space to have impact or a broader role. Or a more challenging role. And so once you have that hypothesis, you go to the market and you're like, hey, what is it that's going to make great people work for us?

To give an example, when we were hiring engineers early on, we said: strong customer orientation and a desire to work closely with customers is really critical for us and we're going to put a lot more weight on that in the hiring process than a typical engineering role would. That’s how we’re going to be really effective in finding people to become founding engineers and I think by and large we succeeded in doing that. But there was sort of this intentional hypothesis about what we are going to do differently to the talent market as a whole.

Back at the end of 2023, SkillsRobin connected you to Peter. What convinced you to hire him, for example?

I think that’s a great example. I said, you know, we want to put a lot of weight on customer facing orientation, ability and hunger.

Someone like Peter, who is self-taught as an engineer, but only had a relatively small amount of full-time experience as an engineer. But when you meet him was just off the charts in terms of like drive - like this is someone who just sat down during COVID and said I’m going to change careers, I'm just going to make it happen by myself and I’m going to do it really quickly. But it was also clear that he had strong design chops. He really wanted to talk to customers and work closely with them. Built some really interesting side projects. That was our angle on the market.

I think a lot of other companies would have looked at him and said - not enough engineering experience - pass. We said actually there are these big things we want to put more weight on around drive and customer orientation that make it a must hire for us.

What advice would you give to job seekers who are interested in working with a pre-series A startup like yours?

I think the biggest difference between the talent market in Sydney and in Silicon Valley is that in Silicon Valley everyone is just obsessed with “What is the best company and then how do I get into the best company?”

In Sydney, people are way too focused on everything that’s not that question: title, compensation, role, brand, etc.

Ultimately you have to view joining a startup as this ticket to have incredible success if the company is successful. Now there are some base things that have to be true - like the founders and leadership of the company have to be good people with high integrity. I'm not saying you should join a company that's successful but abusive, but beyond that it’s - can you get into the best company?

Related to that, there’s also some basic hygiene around the degree of transparency that founders give early joiners that seems to be violated commonly in the Sydney market. So for instance, there should be no shenanigans around equity offers. You should know how many shares you're being offered and what percentage of the total asset in shares that is. They should be willing to share cash in the bank, runway, revenue, customers. Those are sort of basic hygiene questions you want to ask. But top down it’s just going to be which company is winning and that's where you want to go.

How did you grow your team to 17 people within the space of a year?

First up, why do we scale? We scale up because we're in a competitive market. And the ability to deliver close to customers is critical. The degree of team growth is a bit of an outgrowth of what the business is like. And then I think we've been fortunate to scale quickly because we’ve never had someone reject an offer.  We've had 100% offer acceptance.

We screen pretty hard early in the funnel then have people come and do in person work trials. We’ve had some people we haven't made an offer to after a work trial, but it’s actually a relatively low rate because we screen quite heavily earlier on. So to hire 17 people, we haven't had to have a crazy wide middle pipeline. We haven’t had to do hundreds of interviews. I think it’s been quite effective.

What’s next for Lorikeet?

We now have a winning product. We can get into competitive deals with companies that have hundreds of millions of dollars in funding and have people say: “Oh, the Lorikeet product is better!” We've got to get into more deals, get the product in front of more people and that's a big priority.

Then we have a lot of really high ROI opportunities that we need to put engineering behind. That's expanding the range of support platforms we can support. Increasing like the depth and quality of the AI, making the overall experience of configuring and managing the system more automated. Those are really exciting. We've got to bring people in to help us do that. And so, you know, we would like to roughly triple our engineering team in the next 12 months. In terms of customers, we've had customers in the US from the beginning and I expect most of our growth will be in the US. It's the biggest market. It’s the fastest buying and best paying market. We’re in pilots at the moment with a number of unicorns and public companies in Australia and the US, so I think we will probably continue to move up the size axis in terms of growth. But we do work with a number of startups and we get a lot of value out of that in terms of learning and speed of deployments. That will continue to be a part of our business, but as we're finding the product can compete effectively in bigger and bigger spaces we'll keep pushing on that as well.

‍

Learn more about Lorikeet: https://www.lorikeetcx.ai/

Interested in the developer side of the story? Check out our interview with Peter from 2023 when their journey started

‍