What an AI hackathon costs

01What are you actually paying for?#

Three things, and the facilitation on the day is the smallest of them. You are paying for prep, the interviews and the curated problem list, for the facilitation itself, and for the written debrief that decides what survives. A vendor who quotes only for the day is quoting for the least valuable third.

02Which variables move the price?#

Group size, prep depth, one day versus two, on-site versus remote, and whether follow-on support is included to harden the surviving builds. Group size matters less than most people expect; prep depth and follow-on support matter more.

• Participants: eight and forty are different days, but not proportionally different prices, the prep scales with the number of functions, not the number of bodies.
• Number of functions represented: each one adds interviews and a different problem shape.
• One day vs. two: the second day is usually hardening, not more building.
• On-site: travel, and a materially better day.
• Follow-on: a few weeks of support to get two or three builds into real use.

03What's the cost nobody budgets?#

Participant time. Forty people for a day is the largest line item in the engagement by a wide margin, and it is invisible because it never appears on an invoice. This is the actual argument for spending properly on prep, the prep is what protects that investment.

A badly prepared hackathon does not cost less than a well prepared one. It costs the same headcount-day and returns nothing.

04How do you build the business case?#

Compare against the recurring manual work in the room, not against a training budget. Take the three or four workflows most likely to be automated, estimate the hours they consume per month, and apply a conservative survival rate, a third of what gets demoed still running after a month.

That framing also sets the right expectation internally. You are not buying an event. You are buying the removal of some specific recurring work, plus a capability that makes the next removal cheaper.

Questions people ask#

Is it cheaper to run it internally?

In cash, yes. In practice, internally-run days most often fail on the prep, because the person organizing already has a full-time job and the friction interviews are the first thing to get cut.

Do you charge per participant?

No. A fixed fee scoped to group size, prep depth, and format is cleaner for everyone and avoids the perverse incentive to invite fewer people.

What about tool and API costs?

Small. A day of forty people building on commercial AI tooling is typically a rounding error next to the time cost, but get the accounts provisioned in advance regardless.