At the product retreat, we used GPT-3 to understand why customers cancel their paid Todoist accounts.

Here’s a small write-up of how this works and how you could utilize this to get valuable insights into improving your products. 🤖 It’s critical to note that you won’t be able to feed data to GPT-3 because there is a token limit (e.g., 4096 tokens for Davinci). You also won’t be able to fine-tune the custom model because fine-tuning only works for prompt+completion optimization.

Some examples of @OpenAI davinci-002 versus davinci-003 using Todoist's Suggest Tasks from AI.

It's getting wild 🔥

Lose weight: Learn Spanish:

Async communication is a simple concept: you send a message without expecting an immediate response. The surprising thing is the side-effects of async if you make it the default communication.

Here are some core learnings we've done at Doist in the last 8 years 🧵 🤝 Default to Trust

You don't know when or if people are working. You also need to trust that your teammates will deliver on time. The only way you'll make this work is in a super high-trust environment.

One of the core aspects of making async work is being great at handoff.

Handoff could mean sharing a product spec with an engineer or sharing decisions between the leadership team.

🧵 with core tips on how to become better at it. Follow the Inverted Pyramid

Start with the most important piece of information in any handoff, so readers can get the main point whether they read the whole document. nngroup.com/articles/inver…

Over 1 billion knowledge workers across the globe are tasked with some of society's most supercritical work, yet their time and precious mental energy are wasted on busywork. The status quo work environment is ineffective, toxic, and a massive problem for humanity.

🧵 Some stats Knowledge workers spend up to 50% of their time in meetings. That's 130 workdays per year. The yearly cost of unproductive meetings is estimated to $399bn in the US and $58bn in the UK (and who knows how much globally).

blog.otter.ai/meeting-statis…

6 years into our journey with Twist, most companies would’ve quit. Radically changing the way people work is a hard battle. But we’re not afraid.

Today, we’re doubling down on our bet against status-quo and launching a new Twist.

🧵 A thread about our biggest gamble yet: 1/ The backstory: As a fully remote team, @doist used Slack in 2015. It was fun, but the honeymoon with real-time chat didn't last. We spanned 10 time zones and needed an async tool to work. So we built one, quit Slack cold turkey, and never looked back. blog.doist.com/betting-agains…

🧵 Crypto and blockchain might revolutionize not only money and value but also corporations. Here are some thoughts on how Corporations 2.0 (Corps 2.0) might look like.

0/6 1/ The Dutch showed in the 17th century that you could build a corporation that could be more powerful than many nations at that time. It was a revolutionary concept.

See Dutch East India Company en.wikipedia.org/wiki/Dutch_Eas…

0/ 🧵 The tools shape us, and we shape them. The incentives for tech products in both the personal and work environment are awful. Products are not optimized towards your well-being but for corporate gain and growth. 1/ The Social Dilemma explores the personal environment and the dangerous human impact of social networking. Our work environment isn't a lot better. netflix.com/us-en/title/81…

0/ 🧵 At Doist, we have for 10+ years competed against Google, Microsoft, and Apple in the hyper-competitive market of todo apps. Some thoughts follow on how you can compete against trillion-dollar Goliaths.

I'll start with examples, and end with core principles. 1/ Some examples of companies that have successfully done it:
📹 Zoom built a $50bn+ company in a market where the Goliaths have operated for years.
🎧 Spotify built a $50bn+ company and a market leader in music as they competed head-on against Apple and Google.

0/ 🧵 How can a small team get work done in the best possible way? At Doist, we have been at this for many years, and we have learned things the hard way. Here are some of the core lessons 👇 1/ 🔮 The team needs to have a clear vision of what they need to build with a well-defined problem and a solution space — before they start working on it.

Over the last many years, Basecamp was a significant source of inspiration, but their book on product design is a hard read. Some thoughts follow where I disagree with their approach. Their book is located here: basecamp.com/shapeup 1/ A lot of interesting projects take longer than a six-week cycle, so it's strange to try to limit everything to fit an artificial time limit. We also work in cycles, but we accept that some ambitious projects can take multiple cycles to complete.

Level, a similar service to Twist, is shutting down. A thread follows with some thoughts on starting a product these days. derrickreimer.com/essays/2019/05… The easiest path is to solve a problem you have. This way, you don’t have to second guess if you are solving a real problem, and you don’t have to do user research. We started Twist to solve problems we had with distractions of Slack, I built Todoist to manage my work and life.

0/ 💡 Some core things we've learned about retreats over the years. Our learning is based on 5 company-wide retreats around the world. This is a response to

https://twitter.com/andreasklinger/status/1126571386892312578

1/ 💸 Retreats are expensive, most of ours have been $3000+ per employee. This cost is easily worth it, especially given that we don't spend anything on office leases.

0/ Some thoughts on metrics, and what we’ve learned over the years 📈 1/ When I started Todoist in 2007 I used to do SQL queries to figure out things. There were no dashboards. In hindsight, this might have been good because I just cared about building a great product and not about the numbers.

0/ 🔀 Some thoughts on how to build and promote an asynchronous culture inside your team. Doing this can help you produce better work while having the ability to disconnect and have a better life quality. 1/ 👏 Evaluate people differently, e.g., based on their output and results and not how responsive they are or if they are online all the time.

0/ Remote work isn't exceptional as companies that are spread around multiple offices have done it for the last many years. The special sauce is communicating asynchronously as the default 💡 Here's a thread about why. 1/ 🌎 Communication between timezones becomes easy, which means that you can hire from anywhere in the world and truly build a stellar and diverse team.

💡 1/ Improving decision making, leverage or impact of work is much harder than just working 24x7. Most of the advice today is only on working hard and all the time. We are focusing on the wrong things. Here are some thoughts on this matter. 2/ Before the information age, there wasn’t much difference between people as it’s physically limiting how many more things you can make than another person, even if you were more skilled or worked all the time.