Today marks an important milestone in Kaito’s history.
As unsexy as it may sound (especially in our industry), we’ve transitioned from a startup having to rely on venture capital funding to one that can stand on its own feet - yes, we've achieved profitability!
On this special day, I wanted to take a moment to reflect on our journey so far - what went well, what didn’t go so well, the lessons we’ve learned - and share with you a little bit about what’s upcoming at Kaito.
First, a high-level overview of our journey so far:
- The beginning: received first dollar in the bank (h/t
@dragonfly_xyz) in June 2022 (1 FT)
- 0.7 yr: launched our first MVP in February 2023 (7 FT)
- 1.3 yr: launched our beta version and secured our first paying customer (yay!) in September 2023 (11 FT)
- 1.5 yr: launched our public version in Jan 2024 (16 FT)
- 1.7 yr: first felt PMF in February 2024 as we scaled the business 6x in 2 months (18 FT)
- 2.0 yr: achieved profitability in June 2024 after growing our ARR 30x in 6 months (27 FT)
Plotting this against other startups (h/t
@lennysan), we take pride in the fact that we’re in the upper percentile in terms of time to product-market fit (PMF), and we're among the few GenAI startups that have achieved profitability with decent scale.
1/ Reflecting on our journey, there were some sound decisions and good practices that helped us get to where we are today:
- stay really close to users. Our entire team puts users first. Looking back, I was on almost every single demo with our users (collectively more than 1000 hours, averaging 5-10 demos per day, for the past 9 months) and have Telegram group chats so that you can find me any day, any time. I knew this wasn’t scalable, but it was one of the most valuable things we did to understand first-hand what our users really want.
- user experience matters. Part of what helped us take off beyond the initial 10 power users was making the UX easier. I had the same conversation with
@tn_pendle the other day about how the UX on Pendle significantly improved from the initial days. For us, we built analytics, dashboards, and feeds that allowed our users to get insights without having to search for them. We are still pushing hard on this - AI copilot, personalization, smart push notifications are some of the near-term features that will further reduce barriers.
- think new: One of the guiding principles at Kaito is to build things no one has built before. I particularly want to emphasize this because I feel the industry isn’t short of devs but needs more teams working on new ideas / meaningful innovations that push the industry forward. To me, this means fewer vampire attacks and less “[xxx] but on [xxx] chain” type of things. Granted, this often means higher uncertainty, longer development time, and, practically speaking, a lack of valuation anchor/peers during fundraising. But I still encourage us all to think this way - I do believe market will, over time, reward truly differentiated teams, and I think we are already starting to see this.
- being agile: We made a lot of mistakes along the way, but the best thing is we remained agile - unafraid to admit mistakes, quickly correct them, and move on. Internally, we were very early users of Perplexity and were always inspired by how it evolved from a text-to-SQL engine in the beginning to a full-blown web search engine. I highly recommend
@AravSrinivas’s recent podcast with
@lexfridman.
- tech-driven & AI-first: We always had an AI-first approach. We try to be thoughtful about what we do and what we don’t do. Being AI-first also means things may take longer (sometimes much longer) to achieve the same level of base performance, but once you get there, you achieve much better scalability and performance. There are many such examples at Kaito, including our STT model, our automated data labeling, synthetic data generation, and AI-powered newsletter (Overheard on CT).
2/ We certainly made a lot of mistakes along the way:
- We decided to try a different user segment (retail) without properly adapting to their specific workflow and preference. As a result, the retail product (WHAT) we launched in June 2023 failed to achieve meaningful traction. People were trying it thinking it’s cool but the retention was horrible. Fortunately we very quickly recognized this and deprecated it.
- The initial GTM was chaotic and the pricing was too complicated. None of us had any experience with enterprise SaaS GTM, so we were admittedly trying things as we went. We experimented self-service free trials vs demos, credit card gating vs non-gated, and tiered pricing vs unified pricing. The good thing is we’ve always been very nimble and unafraid to immediately adjust based on feedback.
- We made some hiring mistakes. Over time, we realized that the more desperate you are in terms of hiring, the more prone you are to making the wrong hires. Never scale too fast.
3/ Undeniably, we had a lot of luck:
- we started as an AI-first company way before the GenAI hype and really honed in on the data infrastructure and search architecture side. This made us 1-2 years ahead of competition. ChatGPT was truly a godsend for us that 10x’ed our user experience and making many previously impossible things possible.
- as we grew, we realized there’s a huge untapped market for leveraging Kaito for growth and marketing (social listening, mindshare analytics, sentiment tracking, etc). Over time, this became an increasingly important market for us, which is something we were completely unaware of when we started the journey.
- lastly, we got
@dragonfly_xyz's seed round commitment before Luna crashed :)
4/ So, what’s next for Kaito?
- Scaling Kaito for another 10x over the next 18 months. We will continue to provide more values to our existing users, and some of the things we're most looking forward to include launching a fully autonomous research assistant (starting with our copilot this month), bringing our solutions to the developer community (we just launched 9 APIs at
kaito.ai), and revamping all retail and internal team interfaces for the industry through our OpenAI-like inference API (any questions, ask Kaito).
- Launching a new crypto-native application that will be used by not just the crypto community. We aspire to build something as widely adopted globally as
@Polymarket and believe this is crucial for the mainstream adoption for the space.
5/ Finally, we want to send our appreciation to all our early adopters, testers, users, and contributors - we are forever grateful. Please come and meet us at EthCC, Token2049, KBW, and DevCon if you are attending - we have something special for you! And for the community members who have been following our developments from a distance, we also have something planned for you in the near future. Stay tuned!