Challenges, hackathons and promptathons can start useful AI practice, but an event does not establish lasting adoption. The stronger test comes afterward: someone uses the work again on an ordinary task, checks the result and has support to maintain it. Decide who will receive promising work and how they will assess it before you promise that a successful event will change daily work.
This guide is for adoption leaders and managers choosing an event and its follow-through. A hackathon brings people together to build a prototype around a problem. A promptathon can focus on using and refining prompts without requiring a software build. Either can have a legitimate learning goal. Be clear about whether you are commissioning exploration, a project or a change in regular work.
Decide what the event is meant to produce
An event can make room to try something with colleagues who rarely work together. Participants may discover a useful approach, learn where AI struggles or find someone who understands a difficult part of their task. Those are worthwhile outcomes to examine. They are different from a maintained workflow.
Choose the promise you will evaluate:
| Event goal | Evidence to inspect | What you can reasonably report |
|---|---|---|
Learning and exploration | What participants can explain or attempt afterward | A specific learning outcome, with remaining gaps |
Developing a project | A working prototype and a funded next development step | A project is continuing |
Changing regular work | Later task outputs, checks, repeat use and maintenance | A defined practice is being used over the observed period |
Keep attendance and submissions as event records. They tell you who took part and what was presented. A winning demonstration may deserve recognition, but it cannot tell you whether the next shift can use the result. The recognition guide helps separate thanking people for a contribution from turning activity into a contest.
If the goal is learning, do not require every team to invent a permanent project to make the event look successful. Ask what participants learned, how they checked an output and where that learning could be used next.
Look beyond the demonstration
Research on hackathon continuation gives reasons to plan beyond the event. In a 2023 study by Zanele Ratsoga and Mpho Primus, 20 purposefully selected interviews concerned hackathons held from 2018 to 2020 at one company operating in Africa. Participants described funding, suitable implementation skills and integration into company operations as important to continuation. One team merged overlapping work with another team; another stopped when a similar project had already been completed. These are qualitative accounts of corporate projects, not a measured effect of AI hackathons on employee adoption.
A 2020 archival study by Alexander Nolte, Irene-Angelica Chounta and James Herbsleb also distinguishes early project continuation from longer activity. Its Devpost and linked-repository evidence concerns development activity, not whether employees regularly use a workflow. Work can also move elsewhere, and participants can learn without continuing the original project. Choose a measure that matches your event's purpose.
There is encouraging recent employer evidence, with a useful limit. In Kiara Graham's July 2026 account of Workleap's Claude-a-thon, she describes preparation and coaching, then projects in recurring use within 30 days, including a Monday sales-pipeline audit. Workleap also introduced continuing team expectations. Graham, who works in talent management and organizational development at Workleap, explicitly says adoption was already growing and the event cannot receive all the credit. This is an employer's early follow-up account, rather than independent proof of enduring adoption.
Use that distinction in your own reporting. “The team used the workflow on subsequent weekly tasks” is more informative than “the hackathon drove transformation.” Say what happened, over what period and what else changed.
Arrange a receiving owner before the event
A prototype needs somewhere to go when the event team returns to its usual responsibilities. The receiving owner should understand the task, have authority to prioritize the next step and be able to decide whether the work belongs in an existing process.
Before the event, agree these arrangements for work that might continue:
- The task and recipient. Which recurring job would change, and who relies on its output?
- The next attempt. When will the team try the approach outside the demonstration, using approved access and appropriate inputs?
- The checking responsibility. Who can judge whether the output meets the task's requirements?
- The capacity decision. Who can provide time, necessary skills and any funding for follow-through?
- The maintenance home. Where will the current instructions, sources and known limitations live, and who updates them?
These can be provisional arrangements for an exploratory event. You need a route for evaluating a promising idea, not a commitment to deploy every submission.
Avoid leaving the winning team to negotiate everything afterward. If continuing work requires an integration team, bring that team into the decision. A participant who can demonstrate a tool may have neither the time nor the authority to maintain a service others depend on.
Test the work on the next ordinary task
Suppose Sam Taylor dispatches camera equipment for an audiovisual rental company. Dispatch means arranging which equipment goes out for each booking. During a promptathon, Sam and technician Jamie Lee develop an AI-assisted draft of an availability reply from a prepared booking list and equipment record.
At the event, the record shows a complete camera kit available. On the next ordinary dispatch, Jamie opens a returned case and finds a battery missing. The booking list alone cannot establish that the complete kit is ready to go.
Sam needs the approved current booking information and Jamie's equipment check before confirming availability. If the draft promises the full kit, Sam must correct it and resolve the missing item with the person responsible. A convincing event reply is not enough.
Review the complete task, including the work around the prompt:
- Did Sam have the current information when the reply was needed?
- Did the draft identify a question that needed Jamie's physical check?
- What correction or extra checking did the reply require?
- Could another dispatcher find the instructions and use the approach with support?
Save the checked approach with its sources and limitations. Do not save the demonstration as a current availability record. If the process depends on Jamie telling Sam about a returned kit, make that handoff part of the workflow.
Look again when the task meaningfully repeats. A routine can appear useful while its original builder supplies unrecorded help. Ask the next person what they needed, then inspect the finished output and the effort to check it. The workshop follow-up guide helps distinguish missing opportunity or access from a skill that needs further practice.
Make room for participation and follow-through
An event consumes working time, and its promising projects can consume more. Agree what ordinary work will move, who covers essential service and how participation will be interpreted. Otherwise, an invitation to learn may feel like another deadline.
A March 2026 Reddit user discussing a company hackathon described struggling with their usual workload and worrying that either a poor showing or nonparticipation could damage their reputation. They wanted to observe and learn. In a contrasting reply, another user who described managing a team said only some colleagues participated and that declining did not harm their perception of those employees. The anonymous discussion concerns a general company hackathon; it does not establish either employer's policy or the effects of AI events.
The accounts make a practical question visible: do employees know what the invitation means? Explain whether participation is optional, what observing involves and how managers will handle competing commitments. Follow through on those terms.
Include implementation time in the capacity discussion. A team can enjoy an event and still be unable to maintain its output alongside normal work. Use the manager's learning-time guide to make the priority decision explicit. If the problem is an existing task that needs supported practice, time with a knowledgeable colleague may be a better next step than a new competition.
Decide which work should continue
Arrange the review with the receiving owner before the event ends. Bring the prototype, what participants learned and the unresolved work needed for a real attempt. An award is one piece of context, not an implementation decision.
From event output to a continuation decision
Describe the intended work
Name the recurring task, the people receiving its output and what the prototype can currently do.
Check the next real attempt
Use approved inputs and access, inspect the result and record corrections, dependencies and checking effort.
Review the support required
The receiving owner compares ongoing value with maintenance, skills, capacity and related work already underway.
Continue, merge or stop
Record the decision and owner. Keep useful learning even when the original project will not continue.
Continue when the task is useful and the support can be provided. Merge when an existing team or process is a better home. Stop when the approach does not meet the need or its costs cannot be justified. Record the reason and retain any useful learning, so the next event does not rediscover the same dependency.
Your next step is to take one proposed event output to its prospective receiving owner. Agree the ordinary task you will inspect afterward and the capacity needed to attempt it. The people and change guide connects that decision with the wider support for adoption.
Questions about AI hackathons and promptathons
How should we measure adoption after an AI hackathon?
Measure later use on the task the event was meant to improve. Inspect outputs, corrections, checking effort and whether people can repeat the workflow with its agreed support. Keep attendance, submissions and ongoing project development as separate records. State the observation period and other changes, such as new access or training, that could affect use. Begin with the event goal and evidence comparison, then schedule a check when the task next occurs.
Can a promptathon help people who do not code?
A promptathon can give people a supported setting to try AI on a task without building software. Choose approved materials, a manageable output and someone who understands how to check it. For a rental dispatcher, that might mean drafting an availability reply and comparing it with bookings and a physical equipment check. Follow the event with another appropriate attempt before treating the approach as a regular practice. The rental example shows why the surrounding work matters.
Should every winning project continue?
No. A hackathon award recognizes the event submission under its judging criteria. Continuing the project requires a separate decision about useful work, required support and an appropriate owner. A winning idea might fit an existing project better, or reveal a dependency that makes continuation impractical. Review the continuation decision with the receiving owner and preserve what the team learned even if the original project stops.
When should we check whether the event changed daily work?
Start at the next meaningful opportunity to use the approach, then review later repetitions appropriate to the task. A weekly sales audit and an occasional equipment booking have different rhythms. Agree the checks before the event, include maintenance and support, and report the period actually observed. One successful attempt establishes less than repeated checked use. Use the receiving-owner arrangements to make the follow-up concrete.
When is ordinary practice a better choice than another event?
Choose supported ordinary practice when the team already has a relevant workflow but needs time, access or help to use it well. Another challenge will not resolve an unmade priority decision by itself. Identify the obstacle with the employee and work owner, provide the support they can authorize and inspect the next task. The capacity discussion helps decide whether an event has a useful job to do.
