Rebooting (something like) early Triplebyte

We're a bunch of ex-Triplebyte people. We loved it there, and we're trying to make something like it. Here's how.

Rebooting (something like) early Triplebyte

We're a bunch of ex-Triplebyte people. We loved it there, and we're trying to make something like it. Here's how.

This post is for people who were familiar with Triplebyte and want to know how Otherbranch is similar and how it differs.

If you don't know or don't care what Triplebyte was, we have a much more straightforward announcement post here.

What we loved

I loved (pre-2020) Triplebyte. I loved the later versions too, for different reasons, but the original has a special place in my heart.

I wasn’t alone in that. My co-workers loved it, too. So do hundreds of engineers who wish it could come back, even a year after the company as a whole closed down, and four years since the product most people pine for was cut.

Well, it ain’t coming back. The company doesn’t exist anymore, and its assets were sold off.

But maybe we can fork it. (In the Git sense, not that one!)

It’s not a literal fork. We’re not Triplebyte. We don’t have any of Triplebyte’s IP. We don’t have its candidate pool, company pool, candidate data, assessments IP, investors, code, or anything else. What we have is built from scratch, and no doubt we will drift further from our roots as we develop. We’ve got a whole other blog post up on what we are, independent of our inspirations.

But the people are still here. Everyone on Otherbranch's founding team worked there. The philosophy is still here. We still believe in the importance of skills-first hiring, in transparency, and in an opportunity for great people to get a foot in the door. And the need, if anything, is even more acute than it was five years ago. Job hunting in 2018 was shooting fish in a barrel for an experienced engineer; not so much in 2024.

We’re not Triplebyte. I’m a different founder than Ammon - for whom I have enormous respect - was. We’re bootstrapped instead of YC-backed. Our strategies and strengths will be different. If Triplebyte was the main branch of that idea, we are - well, the other one.

But fundamentally, the reason I started a company was that I looked back at old Triplebyte, said “that would work again, and someone should really do something like that”, realized there was no one else who was going to, and realized that I’m in the right position to.

So let me tell how how we’re trying to rebuild what we loved about old Triplebyte, how we're trying to make it last this time around, and how you can help if you want this thing to exist, too.


How we're the same; how we're different

We're rebuilding some parts of the old model and tweaking others.

Here's what should be similar if you were familiar with early-day Triplebyte:

  • [Candidates] We do first-round interviews ourselves.

  • [Candidates] Those interviews are graded without respect to your background.

  • [Candidates] Interview results can be used across all the companies in our client pool.

  • [Candidates] You get candid feedback on your results (actually we give deeper feedback than Triplebyte did).

  • [Candidates] We’re a highly technical, engineering-expertise-first organization..

  • [Companies] Every candidate has passed a pretty rigorous interview.

  • [Companies] We do the usual recruiter things (sourcing, helping contact candidates, etc).

  • [Companies] We use a transactional pricing model, 20% of first-year’s salary.

  • We de-emphasize resumes and emphasize skill assessment.

  • We work mostly with startups.

And here's what's different:

  • We don’t have a quiz before the interview as a pre-screening step.

  • We are not fully background-blind, and probably won’t be for quite some time. (Yes, we know, read on.)

  • We don’t require companies to skip candidates to onsites.

  • We don’t currently have talent managers supporting candidates through the process

So why the changes?

The removal of the quiz is simple: GPT made it too easy to cheat. It was always possible for a determined actor, but it wasn’t particularly easy (we had a couple of middlingly-sophisticated defenses in place). But GPT is really good at the type of low-context recall questions that made up the bulk of the Triplebyte quiz. Even if we had the quiz IP (and we don’t, as previously mentioned), I’m not sure it’d do anything more than select for a highly-cheating-enriched pool. Without the IP, it isn’t worth re-developing, at least not on day one.

We deemphasized (though didn't eliminate) the background-blind approach for two reasons.

One, Triplebyte was never quite as background-blind as it liked to claim (background was an explicit model input in our decisions, although typically not a major one). The broad claim of “anyone had a chance” was true, but background-blindness in a narrow sense was not. One of our founding principles for Otherbranch is to be aggressively honest, and it seems wrong on that basis to mischaracterize something so fundamental right out the gate.

And two, we need to appeal to companies in the current environment, and we need to interview people we can reasonably expect to place. That means we need to be finding candidates that companies want to hire. Down the line, a larger client base, a (hopefully) improving market for candidates, and more clients actively looking for junior candidates will mean we can cast a progressively wider net, and we do intend to transition back into something primarily background blind once we have (roughly) fifteen to twenty active clients.

Our interviews remain background-blind in the sense that they are administered and scored without reference to your background. But for the moment, we're interviewing people only if we think they have a reasonable chance of meeting a client's baseline requirements.

We cut the direct-to-onsite requirement because it was a barrier for clients and because candidates don’t care about it as much in an environment where they’re not inundated with job outreach. In practice, every company we've seriously spoken to has intended to skip several interview steps for our candidates, however (that's why they were interested in us).

And we don’t have talent managers supporting candidates because…well, we don’t have talent managers yet. We intend to have support like that in the future. We just aren't there yet.


What we are today

Here’s where that leaves us: day one, we are a recruiting firm with deep, reusable interviews and candid feedback

If you’re a reasonable match for a company (you meet their hard requirements and some of their soft ones) that’s hiring through us, we’ll do an interview with you. This interview is general and standardized and is meant to detect skills in a wide range of areas. Our interviewing philosophy is very similar.

If you do pass our interview, we refer you to clients for whom you’re a good match. These clients continue with their process (typically a few steps in). You can reuse that passed interview with our clients for some time into the future (we’ll say at least a couple years, although I think this early on we want to reserve the right to make big changes down the line if we really need to). If you don’t pass it, we’ll still give you candid feedback on how you did.


How we plan to avoid Triplebyte's mistakes

If you’re going to start a company inspired by one that failed, you’d better have a plan. And I think it’s worth taking some time to talk about how we plan to avoid Triplebyte's failures.

So, why did Triplebyte fail? I’ve got a whole other post on that, but for the purposes of this post, the short answer is “because we needed to scale and never found a way to acquire candidates that worked at scale”.

Early Triplebyte was a break-even, sometimes profitable, company even operating with the massive inefficiencies of trying to run a hypergrowth startup from a downtown Bay Area office. This was true up to a moderate scale of around 5-10M/year in revenue, the equivalent of a valuation somewhere in the high eight digits. We spent a lot to acquire candidates, but we made at least that in revenue from placements.

But Triplebyte was a venture-backed company. It needed to grow, and grow quickly - that’s the expectation, both personal and fiduciary, that you take on when you take VC money. If the product we had didn’t grow, we needed to find one that did - so we had to pivot, even if what we had was good. This was, even in hindsight, the only reasonable decision the leadership at the time could have made.

But Otherbranch does not have those constraints.

We are not venture-capital backed. I did not seek venture capital, and I don’t intend to, because I already know I'm building something that doesn't scale that way. Nor are my personal ambitions a barrier - I am ambitious, no doubt, but my ambitions are of a slightly different kind. I want to build a profitable, stable business with positive externalities. I think that’s the real challenge, and the challenge is what I’m in it for.

If we break a good thing out of the desire for hypergrowth - well, you’ll know who to blame. I won't be able to hide behind the investors I don't have.

At low to mid scale, candidate acquisition is to some extent a solved problem. Boutique recruiting shops solve it all the time. Triplebyte had good solutions to it, even in a market where talent was much more empowered than this one. That’s just a matter of execution (which to be clear is not easy, it just isn’t a thing that scares me in principle).

And as for the other reasons Triplebyte failed:

The bet on Screen not working out was a consequence of the need to scale candidate acquisition. (Or rather, having to take the bet at all was a consequence of that need.) So that doesn’t affect us, as long as we accept going in that our goal is a business, not a unicorn.

Pivoting to a thing no one wanted, same deal.

Trying to change our users is a lesson I’ve taken to heart. You’ll notice we put fewer restrictions on company processes than Triplebyte did. I don’t intend to tell companies they’re doing it wrong. I intend to tell them we’re doing it right and show them the receipts, and if our results are good, learning from that is their choice. If people are really doing it wrong, we ought to be able to get fabulously rich by doing it better - and that result would speak more to our credibility than a hundred strident blog posts. If companies ask for our advice, we’ll give it, but how they run their processes is up to them.

And while I don't think that this was actually decisive in Triplebyte's case, I do want to say a word about losing trust from the engineering community. I won't relitigate the whole thing here, but the short version is we screwed up on privacy, largely out of desperation to make the pivot work. It wasn't my decision, but if I'm going to claim goodwill towards Triplebyte, I ought to be able to answer for its errors, too.

So here's my commitment to you: if I can't run a company without selling you out, I will shut it down. Because my ambition isn't to build a company, or a successful company, my ambition is to build a company that actually does good in the world. My standard for this when it comes to data is that we should ask for consent directly wherever reasonably possible, and where we cannot, we should (a) make our best judgement about what you'd say if we asked you, and (b) round away from what is in our own interest in cases of ambiguity to avoid motivated reasoning. You give us data for an intended purpose, not for us to exploit to our heart's content until the end of time - we can make reasonable inferences, but we cannot substitute our judgement for yours.

Otherbranch will have our own problems. Bootstrapping is hard. We don’t get a baked-in set of initial customers. The market is rough. I’m an inexperienced founder. This is by no means a home run. But we have, at least, learned from the mistakes we saw in the past.


What you can do to help

You’ve read this far, so I’m guessing you’re reasonably invested in what we’re trying to do.

If you want this to exist, the two biggest things you can do are sign up right now and introduce us to people who are hiring. We know how to do assessments. We have a better process. But building out a network is hard, and if we fail, the failure to build a network is likely going to be why. It’s also the part that you can most help with.

If you know people who are hiring, send them to this page. It’s also worth noting that, while I do not expect this to be our primary business, we’re open to helping companies interview their own candidate pools ~at cost and helping people they don’t hire find jobs elsewhere.

If you’re looking for a job yourself, or you're open to one, you can sign up here. It should take very little time, and we won’t ask you to interview unless we think we have a job we can get you if you do well - think of it as just banking a potential interview later on. We’re especially interested if you are (a) someone who has been around the startup world for a while or (b) someone who is really good at coding, really wants to work for a startup, and just needs someone to give them a chance. (If you’re not sure whether you’re good at coding, try this exercise as a benchmark - "really good" would be something like "halfway through step 4" or better.)

If you have questions, feedback, or other stuff you want to tell us about, you can email us at feedback@otherbranch.com

Thanks for reading, and hopefully we’ll have more for you soon.

Rachel

Founder/CEO

I'm the founder here at Otherbranch. I used to be the head of product at Triplebyte (YC S15), and my deeper background is in mathematics and education. I love the challenge of business, and I think if you see problems in the world you've got a responsibility to work on the ones you can. I want to see business done openly and honestly, without pragmatism serving as an excuse for greed.

Rachel

Founder/CEO

I'm the founder here at Otherbranch. I used to be the head of product at Triplebyte (YC S15), and my deeper background is in mathematics and education. I love the challenge of business, and I think if you see problems in the world you've got a responsibility to work on the ones you can. I want to see business done openly and honestly, without pragmatism serving as an excuse for greed.

Improve your onsite pass rate.
Skip the grind of tech screens.
Hire better and faster.

For employers

Improve your onsite pass rate.
Skip the grind of tech screens.
Hire better and faster.

For employers

Improve your onsite pass rate.
Skip the grind of tech screens.
Hire better and faster.

For employers

Set yourself apart from the crowd.
Show your hands-on skills first.

Get candid interview feedback.

For engineers

Set yourself apart from the crowd.
Show your hands-on skills first.

Get candid interview feedback.

For engineers

Set yourself apart from the crowd.
Show your hands-on skills first.

Get candid interview feedback.

For engineers