Too Slow

Today was a day of mixed emotions with regards to Yakk.

Earlier today, Duolingo announced their new Bots feature, and before I even opened the app, I knew it was either completely, or in some significant way, exactly the same thing I have been aiming for with Yakk.

While it came as a shock, it was simultaneously completely unsurprising. The truth is, I've been far too slow with Yakk progress. I've made every mistake in the book (despite fair warning from people who know better), from obsessing over the perfection of every little thing, to rebuilding it three times.

Reflecting on such a long time makes things clearer in a way, and it's very clear to me (and not just me, I'm sure) that I have not been treating Yakk as a business of any kind. In fact, I stopped referring to Yakk as a startup for that reason over a year ago, and started to describe my ambition and time spent with it merely as a project.

But Yakk is not dead, and seeing what Duolingo have done with their Bots feature today has been a real kick in the ass. It's important to me that Yakk actually helps bring people a step closer to learning a language they want to speak fluently some day.

Unsurprisingly, Duolingo have done a great job of making their Bots feature feel like a joy to use. And certainly, a good part of it overlaps with what before was a fairly unique idea with Yakk ā€“ that a language app could emulate structured conversation and use it as a method in its own right for learning a language.

By later this afternoon, I was left feeling a little deflated. But I was also clearer on a few things too. Yes, Duolingo is moving in on this kind of approach. Yes, Yakk stands no chance at all of helping anybody if I don't find a way to get it under their thumbs. And, yes, I do still want this.

So where does that leave Yakk?

After a few hours of gathering my thoughts and doing some research, here's where I am:

1: The glass is half full ā€“ Duolingo know what they're doing, and have done for some time now. The fact they're pursuing conversational features is a strong signal that conversation-based approaches are valuable, and they're now going to widen that perception among a great many people.

2: I spent 45 minutes with their new Bots feature, and despite wanting to talk about my work in German, instead I was being introduced to Tina, whose horse enjoys eating bread. It'd be unfair to leave out Robert, too, whose shield-toad likes to eat carrots.

While I was technically conversing in form, I still wasn't doing anything differently than if I had been in a random lesson practising the same strange, context-unaware phrases that remain the core idea behind Duolingo. The bot is just a veneer for the same content, and the same lack of a planned, focused learning approach.

From the very outset, the approach I have always wanted to take with Yakk is to have very well-designed curated content, created by people who have learned languages before and know their way through the recurring themes of that problem.

Yakk and Duolingo Bots sit at opposite ends of the table when you consider the fundamental differences in where content and context comes from.

3: It's clear that Duolingo Bots is just a feature ā€“ a means to consume existing content in a more personalised way (the bots use AI to shape the way they interact with each person over time).

Yakk takes a different position by putting the conversation method front and centre, and uses it to deliver contextual, structured learning material that's dependent on your goals. Someone who needs to learn how to order beer on their next Spanish holiday should not take the same path as someone aiming to reach CEFR C2 in Spanish. Duolingo simply doesn't know what your goal is.

Since the conversation itself is the learning tool with Yakk, you can pause to learn about particular conversation elements you may be curious or confused about right from within the conversation itself. Yakk is designed around the conversation being key, it's not merely a feature on top of some other underlying approach.

And when I begin considering future roadmap features for vocab practice taken from completed conversations (so you practise only the words you actually need), and social karma/discussion features to encourage users to help each other, the gap widens even further from where Yakk is aiming and what Duolingo Bots is today.

Today has helped me reflect and realise what I need to do. The important thing is that I still believe in the core idea behind Yakk, and have no intention of stopping. The more I think about it, the more it becomes clear that while Duolingo Bots is an ace addition to their primary app, it's still a distinctly different kind of experience than what I'm working on.

Today's events have made me realise that, more than ever, I want Yakk to succeed, and I certainly have the energy to keep going and find a way to make that happen.

Thanks Duolingo, I really needed that.