I launched Anansi yesterday! 🕸️ 🎉
Anansi has been a labor of love for the last 4.5 months. So many of you have offered to help, given me feedback, and signed up to be its first users. Thank you, thank you. I couldn’t do this without you. 🙏
It feels scary to be launching publicly when the product is so raw. There are a million things I want to make better. But I don't know what I don't know, and I think that getting Anansi out there is the best way to make sure I'm not spending my time in the wrong places.
From here, I have to move my attention to something I’m less experienced with: Marketing and Sales. I need to find my audience, and I still don’t know exactly who I’m targeting. That’s where being scrappy and experimenting are going to come in handy, and I’ll need to step out of my comfort zone and start promoting. Will Anansi resonate more with parents who want to read with their kids? Or will it be better for independent readers? Will my core audience be adults who want to share the choose-your-own-adventure stories of their youth with their kids, aunts and uncles who want to gift a novelty experience to their nieces and nephews, or parents who care about quality edutainment? Or is it none of these yet, since the product isn’t ready yet?
These questions keep me up at night, but I also realized I couldn’t answer these questions on my own and on a budget. The best thing for it was to launch… so I did.
Not by accident, I also pushed Anansi live on the day Grasshopper1 (a learn-to-code platform to support underrepresented folks in tech) was officially sunsetted by Google. I’d had June 15th on my calendar, wanting to celebrate the 2M+ users who spent over an hour learning to code, but then I was like, “Anansi is almost ready to go… no time like the present!” There’s always a reason to delay, so having a meaningful launch milestone felt right, even if it was a bit of a dash to the finish line.
Through all of this, Sarah, my awesome full-stack engineer, was a tremendous partner. Since my last update, we built a lot of things to get Anansi to a level of polish worthy of a public launch. She stayed up late(r), pushed for quality improvements, and made Anansi magical2. I’m super bummed that, for burn rate reasons, I’ll need to wind down working with her for now. If you know of anyone looking to hire an amazing freelance developer, I couldn’t recommend her more highly.
And with that, I’ll wind down the raw thoughts and share a round-up of asks, surprises, and info diet. I’d love to hear from you: advice, questions, or connections that might be helpful.
Asks
Share Anansi with anyone you think would love it - Share the tweet and/or the website with any potential user. I’d love to hear from them.
Is anyone willing to spend 30 minutes talking to me about user acquisition and marketing experimentation? I’m going to search far and wide to find the users who get what I’m trying to do with Anansi. Any pointers are appreciated.
Does anyone know any educators who are willing to give me feedback? Getting educator endorsement will really help, and I think the first step is getting educators’ thoughts + iterating on their feedback.
Things that surprised me
Friends&Family launch - At my last update, my plan was to have a friends and family launch, then iterate a little until I was ready to share with y’all. A week after my last newsletter, I shared a link with immediate family, thinking that I could just hand-wave my way through the rough edges… and I was met with a lot of justifiable confusion. I made some fixes, then sent out the invite to 5 more people, and then got hit with the same feedback from everyone: latency was way too high for the experience to be enjoyable. So Sarah and I bit the bullet: One reader3 reminded me that OpenAI has a “streaming” API, allowing me to update the story as the text was completed… So we did it. We completely rewrote all the content serving logic. Then once it was in, Anansi felt good enough to push out. So I broke my own plan, and hopefully, it doesn’t bite me 🙂
GPT speed-ups - Last week, OpenAI sped up its gpt-3.5-turbo model by (IMHO) about 2x faster. And then things started to break for me. Turns out I’d baked in some logic that actually relied on OpenAI being slow! It’s fun being on the cutting edge, but it does result in some funny bugs. Here’s a plug for Helicone, a useful proxy service that has helped me out a lot in debugging + rate limiting.
Google 403 on embedded browsers - Our unexpected launch bug was realizing that Google Authentication doesn’t like being in an in-app mobile browser. So a number of y’all flagged this for us, and Sarah got Facebook auth + a user-agent detection message up and running within a couple of hours. Phew!
Onboarding matters - I’ve spent so much time making our core story engine work that I forgot how our Daily Story might be the user’s first experience with our stories. I’ve gotten reports of a few grammar errors, and I’ll be diving in this weekend to improve. Moving forward, I think I need to keep first-run experience top of mind (but it’s so easy to get blind to it).
Info Diet
Resources for building AI products
Educational stuff
Synthesis launched a Tutor (access can be yours for only $300+/year!)
Non-topical good stuff
I’m revisiting Contagious as I think about marketing; it’s got some great stories that are good thought starters
After a lot of encouragement, I read Tomorrow and Tomorrow and Tomorrow. It’s not my usual genre, but I gotta hand it to the author on a remarkably well-executed book that kept me engaged despite my preferences. Extra kudos for how beautifully she captured the mixed-race experience.
Since Google shut down Grasshopper, I only have this Wikipedia link I can share... but also: I made a thing that has a Wikipedia article! Bucketlist ✅
I asked Sarah to "make the text and images animate in magically," and she just ran with it and knocked it out of the park.
Thank you! I read all comments and really appreciate the folks who push me out of my comfort zone.