Imposter Syndrome: The Return

The first few weeks of my new job

It was a bleary, hot July evening in Hong Kong. I was blasting the air conditioner, burning through Fable usage limits on the company plan, and thinking: who am I to spend these tokens when I’m not even making anything tangible?

I’d been at my new job for a few weeks, yet my imposter syndrome had already made itself at home.

I have joined Every, an AI-native company that thinks of itself as an AI test lab. We do what regular people do at work, except we try to do it with a fuckton of AI. As much as possible, everywhere, all at once.

I’m Head of Social, which means I look after our YouTube channel, LinkedIn, and X. And I was questioning myself before I even started. During my trial, our CEO Dan Shipper kindly posted about an A/B test I ran. The Twitter trolls picked it up within hours. My first thought wasn’t “haters gonna hate.” It was: oh shit, are they going to realize I’m an imposter?

I don’t come from social media. I come from journalism. Financial journalism too, which might be the most traditional media there is. But over the last year I got obsessed with producing YouTube videos, and over the past few months, obsessed with learning AI.

Am I the best YouTube producer in the world? Not a chance. The best AI practitioner? Also no. But in the overlap of those two Venn diagrams, maybe I stand a chance.

Inside Slack

Last week, Sonnet 5 dropped and our Slack lit up. Katie Parrott, writers asked everyone for takes. Kieran, our software engineer and the father of compound engineering, ran the model through his own tests. The writers put it through their writing benchmarks. People piled in with opinions.

I sat there thinking: I don’t even know what I would use Sonnet 5 for. I have a Claude Max plan. I never run out of tokens. Why would I downgrade for speed?

So I said nothing.

The White Whale

Then Fable dropped, and I immediately pointed it at my white whale: could I build a reliable social media clipping tool?

This is a recent obsession. I’ve tried to vibe code my own Descript so I can build editing workflows for my podcast. Every time, I crawled back to the existing platforms like Overlap and Riverside, because why not use a service that’s already figured out a lot of what I haven’t? Their clips were 20, 30, maybe 40% better than whatever I was duct-taping together. They’re okay, but they’re not what I’d choose if I had better options.

But the Fable livestream Every did was unclippable. A livestream isn’t a structured conversation. A question posed in minute 4 might only get answered in minute 39. Descript and Riverside can’t splice moments from across a recording into one coherent clip.

But Fable could. I spent the next few hours burning tokens, eating into extra usage limits, tinkering. V1 looked promising. V2 added subtitles. Then an outro card. By the end of the night I had four good clips from one sprawling live stream. Four shippable clips.

I was so excited I couldn’t shut up about it. I posted a sample in our #general chat. I Monologue-d my impressions into a Proof doc, wrote up why this model cleared a bar Opus never could, and even had Fable analyze why it did so well.

Minutes after I posted, I noticed an unread Slack message from Katie: “Hey! I’m in the market for cool Fable use cases.”

Everyone has an obsession

I realized it then: Every is not filled with people that are obsessed with AI in the same software vibecoding kind of way. Everyone is obsessed with AI for their own use cases.

My job is not to ship PRs to the website (though I’m tinkering). It’s not to vibe code a product (though that would be cool too). My job is to be insanely obsessed with how AI can execute my actual day-to-day content workflows, e.g. publishing YouTube videos, clipping for socials, testing what converts. I’d already made the call that horizontal clips on X bring in more subscribers and site traffic than vertical clips on Shorts, so that’s a strategy call I’m confident on. Now I have a workflow that makes that call cheap to execute.

For this obsession to flourish, four things have to stay true:

1. The company needs to be AI native.

2. The company needs to trust that I can do it.

3. I need to actually be AI native.

4. I need to trust myself.

Now, may that obsession carry me through. And may that obsession continue to help me earn my spot in the new company.

So uh, note to my, lil impostor syndrome guy… it’s time to chill out now.

---

Update log:

• 🛠️ Zach @ Substack fixed the Substack streak graphic!!

• 🥲 Started gathering advice from friends for my next chapter. I’m collecting them in a Field Notes-sized blank notebook from Pike St Press

• 🏰 Went to Disneyland with my bestie. We’ve been going to Disneyland together multiple times each year since 2018? 2017? It’s a great tradition and we hope to keep it going.

• 🤖 Built a “Steal My Workflows” page on my site to start logging workflows I have on Every: beckyisj.com/workflows

• 💕 Caught up with Karena on everything future-centric! It brings me much comfort that she’s my internet mama 🥹

• 📖 Started reading The Notebook by Roland Allen. It’s a riveting read!

My book: Grab a copy of Bite-Sized Creativity, a 45-min read about how to live a creative life on top of your 9-to-5.

Work with me: https://go.beckyisj.com/workwithme

Some links are affiliate links, meaning that I may receive a commission if you make a purchase through the links at no cost to you.