Nothing works until you make it work

On shipping a CLI homepage, reclaiming a company’s DNA, and creating momentum from nothing.

There’s a particular moment in every startup’s life when the team realises, often quietly, often painfully, that no one is coming to save it.

You don’t have a famous brand to lean on.
You don’t have a giant team to execute systems on your behalf.
You don’t have unlimited datasets to soothe your uncertainty.

In Buildkite’s case, all of this is doubly true. We all need to build the culture and the systems from the ground up in order to progress this tool and this company.

Nothing works until you make it work.

Sometimes the only way to make it work is to build something that looks wildly premature, oddly simple, and strangely inevitable once it exists.

This is one of those moments.

Two weeks.
Two Design Engineers.
A CLI experiment.
A new homepage.

And a company (re)discovering, for the first time in a long time, the power of speaking its own language.

Our state of the market™️: a polite sea of sameness

Before we get to the CLI homepage, we need to address the quietly bland elephant in the room: saas branding.

I wrote about this last time, with the ‘serious approach to being unserious’ that has become my safety blanket for developing a public writing habit again.

No unserious notes here. Right to the heart of the matter.

A lot of the Design industry has above-the-folded itself into a single homogenised genre. This is especially true in software.

Then: Intangible product requiring a humanist touch? Add corporate memphis, now globally commoditised despite it’s branding origins centered on inclusion and diversity.

Now: AI tool, of any sort? Add some nostalgic pixel-art to assuage the uncertainty of the field and invoke an emotional response to a decidedly un-emotional area.

Devtools, tragically, have not escaped this evolution.

Insanely pixel-perfect conceptual illustration, soft gradients, glass treatments and headlines that read like group therapy for stressed-out PMs. Beautiful, sure. Truth to the product? No.

Max from Ragged Edge called this out in 2023 when client research presented him with the first wave of the ‘Linear effect’;

Two years on, and most devtool websites look as though they graduated from the same 12-week Figma bootcamp and have been too polite to differentiate themselves since, if not more so of late.

There are many parts to the equation. Hype cycles, trends, tooling, social media and clout and algorithms, UI and icon kits, landing page conventions…

This is criticism, yes; it’s also taxonomy.

Somewhere along the way, the industry optimised itself into a recipe for monoculture. One part competitive pressure, a dash of risk aversion, plus a small helping of fear that a bold move or message might confuse someone in procurement.

The result is… fine. Perfectly fine.

And… maybe, that’s the problem.

In a field where every brand is trying so hard to be acceptable, no one seems to be trying to be unmistakable.

The origin story we forgot to remember

Buildkite was rage-coded as an alternative to available CI tooling in 2013. It wasn’t born in an incubator, or a pitch deck.

Its earliest adopters were practitioners, working the first wave of Bay Area tech before LLMs and before company mission statements became a byline for empty promises of purpose.

They were and are builders of systems, orchestrators of pipelines, people who know the difference between a tool that works and a tool that poorly plasters over the cracks.

Over the years, as Buildkite has scaled, something subtle happened: the origins and differentiators became obscured by the gravitational pull of saas conventions.

This is how drift occurs.

Not in one catastrophic moment, but through a series of small, reasonable decisions made with the best intentions. There is no blame attached. Only information, learnings, and the power to make strategically different choices from here forward.

Reclaiming the DNA of the brand, therefore, is not a creative decision.

It’s a cultural and existential one.

Two weeks, two Design Engineers, one divisive idea

The current experiment to turn the Buildkite homepage into a command line was not conceived in a strategy deck or a prolonged workshop involving pastries and post-it notes.

It emerged because Buildkite’s Design Engineers asked themselves a deceptively simple question during the messy middle:

What would our website look like if it spoke the truth?

The gnarly part of this question is the follow-up examination of, does it currently speak the truth.

And from there, the floodgates open.

One of the ranging answers was,
it would look like a terminal.
It should look like a terminal.
An actual terminal.
On the homepage.
As the homepage.
As the site..?

Bringing a wholly product-centric view to your domain as software isn’t a new idea. The initial temptation was to go as far as possible, surface an interactive UI treatment that would sit alongside PostHog as an example of all-in.

Releasing a deep product-centric website for us, at this time, is a future step to take; that’s a strategic decision to push further when the right foundations are laid.

Buildkite-as-pipelines has built a reputation over more than a decade for standing apart in the CI field due to it’s clean, minimal interface; we are in the midst of elevating that for GA mid-2026 to be cohesive across the full platform.

With this future in mind, taking a step back to meet devs where they spend the majority of their time became a recurring itch to scratch. The concept has floated around.

There is a current VC example.

PlanetScale went full markdown, and garnered good results to boot.

What does this mean?
How far should we go?
What are the risks?
Do we understand the impact?
What could we lose?
What if it fails..?

The team had a moderately unreasonable belief in themselves, and the structural permission to make a bold move.

This is where the principle “nothing works until you make it work” earns its weight.

Brand momentum is not given to you.

It is built molecule by molecule.

So, we built it.

Leaner resources / sharper thinking

Initially pitched as a homepage overtake, then transformed into an all-in website overhaul, then brought back with compromise to retain the classic sitemap after some cross-team negotiations, the first ship date was set eight weeks out.

We brought that deadline down to two weeks, taking advantage of the company’s event presence at AWS re:Invent 2025 to test this directly amongst a spike in IRL attention.

Part of this reasoning was to stress the intent: an experiment by way of testing a hypothesis.

We had compiled a shortlist of systemic issues that could begin to be addressed by this experiment, or provide a partially closed door to then go address somewhat in the background.

Move quickly, and, perhaps, break, some, things?

This is key: those things breaking were already cracked, foundations to scale not in place, executions not clearly visible as producing benefit or impact. Kintsugi, anyone?

In an industry seemingly conditioned to believe that transformation requires armies, it feels almost impolite to point out that small teams consistently outperform large ones in moments requiring clarity.

The trick there is, leaning into the existing team as the core reference points.

Buildkite does not have:

• Brand awareness (let alone recognition or trust) to lean on
• A large Design team
• A research function
• The safety of consensus
• The illusion of certainty that data often provides

None of this was a disadvantage.

It was the removal of excuses.

Constraints have a funny way of revealing the core idea hiding under the weight of optionality.

The CLI homepage exists because the team had conviction, skill, and a blank mental page removed of bias and baggage; a rare creative canvas, providing a pathway for a bold idea to come to life.

Using AI to accelerate thinking, not replace it

AI played a role in this process. It wouldn’t be 2025 without that being an upfront acknowledgement.

But, not the role the industry likes to evangelise (at least in the algorithms I am fed across channels).

It wasn’t used to ideate an initial concept list or to generate the creative output. It didn’t design the homepage, write a metric fuck-tonne of code we couldn’t then review, develop the copy or shape the interactions.

Instead, AI acted as one of the collaborators in the middle, the mess, the hard thinking territory. The area that rarely raises its head publicly, unfortunately. The place and time where problems become hydras, ideas require research, and solutions come from connecting the dots laterally then simplifying, simplifying, simplifying.

It helped expand the exploration space, helped us interrogate our own assumptions company-wide when we gathered cross-team critique and feedback, and accelerated progress for our micro-team to execute at a very high level.

Strategic dominoes

It’s good to be clear: a CLI homepage is not a transformation.

It is the announcement that the transformation has begun.

Over the coming months, this signal will evolve into larger, entwined structural changes we implement to shift Buildkite’s perception and market position:

1. A renewed focus on culture, and community

Community does not materialise from thin air. It emerges from culture. Real culture, the kind made by people, practices and rituals. What exists internally eventually expresses itself externally. This inside-out relationship creates a continuous loop: culture shapes community; community reinforces culture; and together they define a brand experience that reflects the product at its core.

2. The first redesigned platform UI in 13 years, Design System included

Buildkite’s pipelines UI, long a differentiator thanks to the clean, minimal execution, is in the midst of evolving into an elevated platform experience driven by our Design Engineers across all facets of the product; in conjunction, a self-serve code-first Design System will allow our Engineering and Product teams to innovate, prototype and build faster than ever.

3. Reimagined UX and onboarding

We’re refining guided pathways through the platform that respect dev intuition while using friction purposefully to gauge intent and opportunities. Less guesswork, more context and immediate insight.

4. Engineering-driven publishing

We are helping drive and shape and nurture a new online home for our own practitioners, the builders behind the builders, to speak in their own voice; sharing the technical thought, experimentation, decisions and failures that shape the platform used by a crazy list of the world’s best engineering teams.

5. The next stages of the website redesign

A future where Buildkite’s market-leading UI becomes interactive in the browser, not as marketing theatre, but as a way for practitioners to experience the platform through behaviour. Keep your ears to the ground in mid-2026.

For all of the above, the CLI homepage is our first commit.

Experience as Strategy: designing from behaviour, not for decoration

I understand why there may be a reaction to this homepage as, oh nice novelty. Or, this is a gimmick. It makes complete sense. They aren’t wrong.

However, there is grey area here that we believe is worth pursuing.

Experiences are systems, not individual surfaces.

Standing from where the idea was born, seeing the bigger picture and the vision, this is our flag in the ground to signal we are moving to experience-aligned branding.

The purpose of this is to begin expressing identity through interaction rather than explanation. To find new and different ways to resonate, alongside the conventions that have become the norm.

People know when something has been designed for them versus around them. They know when a company is speaking in its native tongue, or performing in a borrowed one.

Buildkite is now doing something increasingly rare: it is starting to behave like itself.

Why this matters if everything is “fine”

We are living through an age of (saas) homogenisation.

Design is converging. It is easy to understand why. Ease, convenience, speed over quality rather than speed and quality, the vague insistence on craft and taste now as new differentiators (were they ever not?). Some of these are broader business themes, some are more Design-specific. Visual styles, kits, off-the-shelf systems, voice and tone and no personality, layouts and conventions, colours. I occasionally rant about this through our internal channels;

An unfortunate truth is that differentiation almost always translates to risk, particularly as companies mature.

Conventions are safe, they’re familiar, they’re defensible, and they offer a reassuring logic when growth goals and revenue targets (rightfully) dominate executive decision-making. It makes absolute sense.

But Buildkite has carried differentiation from the beginning: Australian-born, minimal UI, pioneering architecture.

Those roots were never decorative; they were directional. And they remain strategically relevant today.

Differentiation is not a flourish to add when convenient, or occasionally dalliance with; it is a foundational brand asset you cultivate so the end-to-end experience stays both purposeful and emotionally coherent.

The point is: these two truths do not need to oppose each other. Strategic clarity and differentiation are not mutually exclusive.

Rachel Kobetz recently wrote on this with “Leading at the edge”. Her great point is, the world is accelerating quicker than ever; playbooks are no longer enough and safety is no longer a viable long-term strategy.

Our emphasis should be long-term brand building, in order for the cult of Buildkite to become recognised; and the heroes of that cult are not us, they are the builders that ship software to over 1,000,000,000 daily users.

If you want to shift perception, you cannot do it by repeating the behaviour that created the existing perception.

You need an act.
A signal.
A break.
A first domino.

For Buildkite, the CLI homepage is that first domino.

A call for braver, smaller, more honest moves

If there’s a broader message here, beyond a homepage and a small team moving mountains, it is this:

You do not need scale to create significance.
You do not need certainty to act.
You do not need a massive brand to get attention.
You do not need unlimited data to know what’s true.

But you do need the willingness to begin.

Nothing works until you make it work.

And sometimes making it work requires doing something unusual, uncomfortable, or boldly obvious in retrospect.

This is how momentum is made.
This is how identity is claimed.
This is how a company remembers what it is, and shows the world at the same time.

The CLI homepage is not Buildkite reinventing itself.
It is Buildkite remembering itself.

And this is only the beginning.