The short version
These three get compared as if they are alternatives. They are not, they solve different problems, and picking the wrong one is the most common way a first AI budget gets spent without changing anything.
01What is each one actually good at?#
Training builds shared vocabulary at scale. A hackathon builds belief and capability in a smaller group by producing working software. A pilot tests whether one specific system delivers value in production. They address awareness, capability, and evidence respectively, three different gaps.
| Training | Hackathon | Pilot | |
|---|---|---|---|
| Fixes | Nobody knows what this is | Nobody has built anything | We don't know if this works here |
| Reaches | Hundreds | Tens | One team |
| Produces | Shared vocabulary | Working internal tools | Evidence and a decision |
| Time | Hours | 1–2 days | 6–12 weeks |
| Fails when | There's no follow-on practice | There's no prep or follow-through | Success criteria weren't set upfront |
02How do you tell which problem you have?#
Ask three questions. Can people define what these systems do and don't do? Has anyone shipped something a colleague uses? Do you have evidence that a specific workflow improves? The first no you hit is the gap to spend on.
No shared vocabulary
Train first, but pair it with something hands-on within a month or it evaporates.
Vocabulary but no builds
This is the most common state, and the one a hackathon is designed for.
Builds but no proof
Pick the strongest build and run it as a pilot with defined success criteria.
03What is the right sequence?#
Short training to establish vocabulary, a hackathon four to six weeks later to convert it into capability, then a pilot on whichever build showed the most value. Running them in the opposite order is why organizations pilot things nobody internally understands well enough to maintain.
The interval between training and hackathon matters. Long enough that people have poked at the tools; short enough that the vocabulary hasn't faded. Four to six weeks is the window where the two reinforce each other.
04When is the answer none of these?#
When the constraint is leadership rather than the organization. If the people making the decisions can't yet tell a real capability from a demo, group interventions will not help, that is a coaching problem, and it is worth fixing before you spend a training budget.
Questions people ask#
- Can a hackathon replace training?
- For the forty people in the room, largely yes, the teaching block covers the essentials and they immediately apply them. For the other nine hundred people, no.
- Should the pilot come from the hackathon?
- Ideally, yes. A build that already survived a month of real use is a far better pilot candidate than something chosen from a vendor's case studies.
Related
The timely half
What I'm seeing this week, by email.
By subscribing you agree to Substack's Terms of Use, their Privacy Policy, and their Information Collection Notice.
Playbooks stay current. The newsletter tells you what changed. Free.