What this founder learned from failing to scale his genAI startup

Sign up for the Daily Newsletter, sent exclusively to our premium subscribers. We break down the big and messy topics of Asia’s tech and startup community. Get the newsletter in your inbox everyday with a premium subscription.

Hello reader,

As a writer and storyteller, I firmly believe that everyone has an interesting tale to tell, one that can capture the imagination – as long as it’s told right.

That’s the role I see for partnership content. My team and I put together disparate elements into a roughly 1,000-word article that’s not just easy to understand but also educational and – dare I say it? – entertaining, regardless of the subject matter.

I say all of this to point out that today’s featured story is a great example of such concise and useful content. In it, a genAI founder takes us through the lessons he learned from building but failing to scale his solution, and it’s all laid out to make it easy to understand, despite the complexity of the subject matter.

If publishing partnership content is something you want to do but don’t fully know how to go about it, hit me up (winston@techinasia.com). We got you.

Today we look at:

• One genAI startup founder’s experience trying to crack the scaling problem
• The completion of TikTok and Tokopedia’s mega merger
• Other newsy highlights such as Investree agreeing to terminate its CEO and Grab’s proposed acquisition of Trans-cab being put under the microscope

Premium summary

You have to try before you’ll know

Image credit: Timmy Loen

I’m guilty of being an overthinker and overplanner.

However, while I do still see the value in meticulous planning, I’ve long since realized that, for some things, you just have to go. Some lessons and insights can only be learned by making an honest attempt. That’s what Ian Tan did with Scribbler, his podcast summarizer solution.

• Neither here nor there: Initially, Tan and his partner charged users US$10 a month for unlimited use. This didn’t work, as it meant that Scribbler could only profitably serve customers who ask for less than 25 summaries per month, and those that went beyond that would very quickly create negative margins.
• Not very niche niches: Summarizing tools powered by genAI are more common than you’d think. For instance, there are no less than 40 podcast-related products on a popular AI tools directory.
• Part of the machine, not the whole machine: Tan’s opinion is that large language models are a new computational building block rather than a platform unto themselves and that it’s better to look at how existing solutions can be improved with them instead of tackling new, unsolved problems.

Read more: What I learned about genAI from failing to scale my AI startup

When giants get together

For 24 hours only – get your passes at super discounted prices!

Quick bytes