Poetic has an incredible opportunity, and perfect timing. The rest of the industry has spent the last few years warming up all the big fish, but there’s no one to deliver: startups are unreliable, SIs are idiots, and (neo)labs too busy trying to make their big models even bigger. Markie’s pushed Poetic to cruising altitude with impressive new logo growth, and more importantly, dramatic expansion. Revenue per headcount is my favorite metric, and we’re killing it on that front! The team and people surrounding the company are truly fantastic; I don’t have much else to say here because the opportunity is so obvious.
Labs are focusing more and more on enterprise applications, buying and becoming their own SIs and spinning up forward deployed teams. OpenAI’s headcount has 2xed this calendar year, and it’s largely not in EPD. Poetic needs to thread the needle of benefiting strongly from the tailwinds of stronger models, while still being specialized enough in a sufficiently annoying set of use cases that the labs don’t go for it directly. Ideally, we should be giving them a strong incentive to partner vs compete (on this note, we should be doubling down on our investor relationship with OpenAI).
As always, scaling the key players at the company is now the biggest challenge. After sitting in for a couple days, there’s a lot of (exciting!) work to do on this front.
Eng is the heart of the company, and our team has a worrying penchant for over-engineering. SandDB is a particularly good example, but far from the only one1. Sikan is absolutely brilliant, but likes to go off in his own world and build, rendering a rather laissez-faire approach with the other engineers. Coupled with at least one team member’s ability to run remarkably wild with runaway decisions, the current state of the product is an extreme over-investment in core the engine, and a worrying underinvestment in the long tail of tools that the FDEs actually need to do their jobs2. Low-touch management is ideal with such a small team, but the team must be composed of individuals who benefit from that style; this is worth revisiting in the coming months. Some small nudges here with Sikan could make a world of difference: Anubhav has already begun this process.
FDEs are, by some measure, the most important people at the company at the moment. They are what will make or break revenue and growth goals over the next year, and scaling them must be the top priority. Having grown so fast, the state of their tooling (or lack thereof) and rigor leaves something to be desired, but it’s crucial we don’t simply write this off as “it’s like that everywhere”; the rest of everywhere is screwed in a way we must not be when building the Last Piece of Software™. I watched the Eng team make decisions based on what can generously be described as guesses at the information the FDEs should already know, and the disconnect is growing. The people building the product should have easy access to an index of all FDE knowledge; instead we have FDEs spinning up personal Render accounts to run cron jobs that the Eng team finds out about weeks later. Kent is working towards rectifying this, but needs to feel more empowered. As is the recurring theme, all that’s needed is some small nudges in the right direction, some small bridges between teams.
And then of course we have our dear CEO, whose job I very much do not envy. Markie is learning the tough lesson every CEO does as the team scales, though hopefully we can make it a little easier for her than it is for most first-timers. Her “product” is now the company as a whole: Markie needs to build Poetic so Poetic can build the Last Piece of Software™. She’s made leaps and bounds of progress on this front, especially through the trial-by-fire of figuring out which leaders and deputies to keep vs let go, but still has some habits to fix. I watched 1/4 of the Eng team potentially lose a week of time in response to a single offhand message Markie sent Kent about switching over Anthropic spend. As Sikan needs to work on facilitating pushback in response to his opinions, so too does Markie need to work on empowering people to say “no”. Serializing these requests through trusted deputies can also go a long way in the interim.
Don’t get me wrong: the above makes me pumped! Lots to fix means lots of fun work to do, and it’s easy to sit on the sidelines and critique; I’m excited to get my hands dirty and help in every way I can.
Table stakes is that I’m able to jump into the engineering team and be another Sikan. It might take me a hot sec to get up to speed with everything he’s built, but I’m confident I can get there in a way the rest of the team haven’t yet managed to. Sikan and I see eye-to-eye on a lot of what he is building, and I’ve spent the last two years doing in-depth research on the next set of primitives he happens to need. Our success pairing on the VM snapshotting architecture is a great sign.
I’ll be able to play a similar role to Anubhav and Kent as well, in starting to nudge things in the right direction culture-wise; hopefully the three of us can provide Sikan with the bandwidth and support he needs to level up as the company grows. Alternatively, if he decides he’d rather hire someone to take over all the managing, I’m pretty damn good at findings VPEs!
There’s some low-hanging hiring I can jump straight in and help with, such as an ops/sec eng hire to get the compliance and security work out of the slow path. Sikan and I have additionally talked about beginning to distribute the infra work amongst the team, building towards full-stack ICs rather than siloed specialists when it comes to “real” product work.
My background, coupled with my unique relationships with Markie, Sikan, and Leigh Marie, puts me in an excellent spot to help Markie achieve more leverage across the org, from nudges and bridges with the FDEs to board meeting prep. I don’t want her job, but I’m excited to help her do it the best she can.
I’m also perfectly happy to not help with this stuff. I think I would be underused as purely an engineer, but if we think that’s what the company most needs, you won’t see me complaining. It’s worth explicitly deciding what Markie wants my role to be. I’m averse to bullshit titles, but in the event we do want to include the “everything else”, it is well worth being intentional about how we represent the role to the rest of the company. We need to walk the line between giving me the leverage to help Markie be a force multiplier, and not confusing people when I start showing up at meetings. The Andrew Qu / “Office of the CEO” model is a good one to emulate.
I too am not without risks. To enumerate a few:
125 bps, standard vesting
$450k annual salary
Option, but not obligation, to sell up to 25% of my advisor equity at any future round, to either the company or an existing investor
Poetic acquires Aneta for $13
Start date +2 weeks out