Running hackathons, with Carter Rabasa from Langflow
A hackathon can put more than a hundred motivated developers inside your onboarding flow for a weekend — for perhaps $5,000 to $25,000 — but Carter Rabasa says it only works when serving the community is the real objective.
Carter Rabasa leads developer relations at Langflow and has organized hackathons across San Francisco, New York and London. Jack met him at Hacking Agents in London and wanted to understand both why it worked and when a DevTools company should try the format itself.
We cover product feedback, community relationships, timing, costs, venues, promotion, local markets and the details that decide whether a weekend feels vibrant or empty.
The Q&A below is experimental — extracted from the episode transcript with AI. If you spot a mistake, let us know at jack@scalingdevtools.com.
Why run an in-person hackathon when content can reach many more developers?
Awareness alone is not a good reason; cheaper channels can provide it. DevRel should serve developers, and a hackathon can combine awareness with product feedback, education and authentic community engagement. Developers get to leave their normal work, learn new tools and play creatively. The company gets to sit beside motivated users while they try to build something real.
Why is the product feedback at a hackathon unusually valuable?
The attendees are motivated, experienced and inclined to solve things themselves. They have already been given the best docs and tutorials, so when they finally ask for help, the friction is worth listening to. You see them use not just the code you shipped but the whole product — website, signup, onboarding, docs and tutorials — for many hours in a real project. That is much richer than an expensive, artificial one-hour research panel.
What do you gain by partnering with an existing developer community?
The event creates a much deeper connection than an email introduction. Good community organizers understand both the tooling landscape and the needs of their members, so their feedback is exceptionally useful. Working with AI Tinkerers in London helped the event serve an established community and began a relationship we can keep building rather than treating attendees as a list of leads.
At what product stage are hackathons most useful?
There is a broad middle where they work, with small 'too early' and 'too late' edges. Do not use a hackathon as QA when you already know the product is unreliable or the docs are poor; you will spend money teaching people not to trust it. At the other extreme, a product as mature and universally known as S3 has less to learn. A simple test is whether you would launch the product on Hacker News or Product Hunt. If you would, it is probably ready for a low-cost hackathon experiment.
What does a good hackathon need?
Start with a venue that has reliable Wi-Fi, plenty of power and — if it is summer — air conditioning. Bring in partners with interesting tools developers can combine like Lego. Help solo attendees form complementary teams. Most importantly, make a promotional plan that reaches the right local communities, because nobody wants to attend an empty bar, concert or hackathon. The desired emotion is that everyone leaves thinking it was fun and wanting the next one.
How did you promote Hacking Agents, and what did it cost?
Most promotion was organic: emailing existing customers and working with communities such as AI Tinkerers, with a small amount of paid advertising. A venue capped at around 150 ended up with roughly 130 to 135 people. A hackathon might cost about $5,000 to $25,000 depending on whether you can get a free venue; catering is usually the next major cost, then prizes and optional production. That is roughly an order of magnitude below many large conference sponsorships.
Is professional photo and video coverage worth paying for?
Not for every first event, but it became worthwhile for our third because we plan to run more. A developer deciding whether to give up a weekend needs to know what the experience will feel like, and real footage helps them opt in or opt out. It also signals that the event exists and is cared for in a world where even conference speaker lists can be fictitious.
Why focus field marketing on particular cities?
We are trying to reach professional engineers actively building AI products or prototypes, and activity clusters in certain places. San Francisco, New York and London were our first three because each has a strong concentration of AI engineering. With a four-person distributed team and limited money, every trip competes with scalable work such as content, tutorials and product collaboration, so we choose the markets offering the most leverage.
What should a company consider before taking an event to another country?
Meet developers where they are, including in language and local customs. In Japan, for example, translated documentation alone is not enough for field work; you need people fluent in the language and sensitive to how the event format itself may need to change. First ask whether good local events already exist and could be supported. If there is a genuine unmet need, act, then write a one-page retrospective afterward covering what worked, what failed and what you learned.
Links & references
- Carter Rabasa's X ↗
- Carter Rabasa's LinkedIn ↗
- Cascadia AI Hackathon ↗
- Cascadia AI Hackathon on Luma ↗
- AI Tinkerers ↗
- Bolt Virtual Hackathon ↗
This episode is brought to you by WorkOS.