Skyvern? What’s that?

Skyvern is an AI Agent that helps companies automate browser-based workflows using LLMs and Computer vision. We provide companies with a simple API that lets them automate their boring back-office workflows with natural language.

How did the launch go?

We spent a few hours in the #1 spot on hackernews and the rest of the day on the front page. Over the day, we received 420 upvotes and 138 comments, blowing our wildest expectations.

How did we expect it to go?

This wasn’t our first time launching our product on Hackernews. Last summer, we were working on a different product called Wyvern, which we open-sourced and launched on Hackernews. The launch fell extremely flat, garnering 10 upvotes, and never making it out of the “newest” filter on YC Show.

Based on this experience, a wildly successful launch would feature 500+ stars, 50+ upvotes, and 5 strangers running Skyvern.

What did we do to prep?

Hackernews has a comprehensive set of suggestions when drafting a ShowHN post here: https://news.ycombinator.com/yli.html. We highly recommend everyone thinking about launching on hackernews check it out.

We had 4 major areas we spent most of our time on ahead of the launch:

1. We wanted to ensure people could try some version of the product to see it in action. We created Demo videos, a quick start guide to run things locally, and added some pre-loaded samples (that actually work) so they can experience the product
2. Our ShowHN post had to be interesting for the Hackernews audience. We tried our best to avoid marketing/sales speak, and instead focused on technical decisions + customer motivations, and tried to include lots of examples of how companies would use the product and why they might find it valuable
3. We wanted Skyvern’s README to include enough context to independently capture people’s attention, as many readers only have enough time to look at a readme before moving on
4. We set up a Discord Server so that anyone with any questions/concerns could come and get more synchronous feedback from us

How did the hackernews community roast us?

What we expected to be a roast ended up being one very warm embrace with 3 key takeaways:

1. Technical suggestions: Hackernews’ audience got fixated on the operating costs of Skyvern, and came up with many suggestions on strategies we could take to lower the costs. These ranged from doing code generation to output browser automation code (that you can later execute for free), or doing some level of prompt / action caching (that you can later execute for free).
2. Cross-platform issues: We had optimized our repo set-up to MacOS devices, and people using Windows or Linux struggled to get up and running with our quickstart. To our surprise, though, people in the discord started helping each other out, and we even got some contributions to help improve our setup script.
3. Demo before discovery: Some readers tried our open-source product out, and continued to book demo calls to access Skyvern’s private cloud beta. We worried that open-sourcing would turn some customers away, but instead, it piqued their interest and made them want our cloud version more!

Aftermath of the launch

We had seen several companies do successful Hackernews launches, but we didn’t understand how much of an impact it could have.

Some quick stats:

• ⭐ 3000 Github Stars
• 🤝 71 Inbound meetings booked via my public calendar
• 💌 39 Inbound emails to join our Skyvern Cloud waitlist
• 📈 1298 workflows executed by open-source users

A surprising ripple effect?

One thing we didn’t expect was the third-party coverage that comes after being at the top of hacker news. Our launch was featured in over 30 different AI groups/newsletters, causing a “ripple” effect on our adoption.

A few notable ones were:

1. We got featured on GitHub trending
2. TL;DR AI, TL;DR WebDev, AI News Sunday, and Alphasignal all covered us in their weekly newsletter
3. World of AI did a youtube video showcasing Skyvern in action
4. GeekyGadgets featured Skyvern for web automation tasks
5. A korean pytorch forum did some coverage on Skyvern

This caused us to continue reaching more people well after we were off the front page of Hackernews.

And the resulting impact on my calendar:

A special thank you to dang, lucasfcosta, matiasweber for giving us feedback ahead of our launch!