Knowledge · AI Basics

AI Hackathon · In-House AI Hackathon

Internal AI Hackathon

📂AI Basics⏱3 min read📊Practical🔗7 related terms

What Is an AI Hackathon?

In one line
An internal event where employees form teams and spend one to three days focused on solving a problem from their own department with generative AI, then present and judge the results. The word hackathon combines hacking and marathon.
Bring this topic in-house📚 Related Course💬 Ask About Training
⚡ Summary in 3 lines
  • The people with the problems and the people who know the tools, at one table
  • Judge on real-world impact and scalability, not polish
  • Without a committed rollout budget, it ends as a one-off event

What an in-house AI hackathon changes is where AI training ends. Ordinary training ends in the classroom, but a hackathon ends only when people bring 'this task from our department' and build something that actually runs. The core is putting the business owners who have the problem and the people who know the tools at the same table, a combination that training rarely creates. Participants leave with the experience of having 'built it,' not just 'tried it,' and the company is left with a list of small use cases it can scale right away next year.

Formats fall into three types. The no-code type builds results with chatbots, automation tools, and agent builders without coding. The hybrid type pairs the IT department with business teams. The technical type involves developers only. If the training team is hosting, start with the no-code or hybrid type and keep non-developer roles at half of participants or more for the greatest impact. A day and a half to two days is a safe length. Beyond three days, team leaders won't release people because of the gap in day-to-day work. Rather than leaving topics open, group them in advance into two or three tracks such as 'report-writing time,' 'customer inquiry handling,' and 'repetitive data cleanup' so judging is possible. If you don't agree with the security team in advance on whether internal data may be used and, if so, to what extent, the event stalls on the morning of the day.

Mentors and judging criteria are what organizers most often overlook. Unless one mentor per team checks in every two to three hours to unblock them, half of the teams spend the afternoon just tweaking prompts. Judging should weight real-world relevance, reproducibility, and scalability to other departments over technical polish, so that teams deliver usable results rather than demos built for the stage. Prizes don’t need to be large, but the budget and an owner for actually rolling out the winning projects must be committed before the event. Without that commitment, sign-ups drop the following year.

The most common failure in practice is treating the hackathon as a one-off event. Results that won applause on demo day often end up in a folder nobody opens two months later, and from then on employees remember the hackathon as 'the day we were given more work.' Another mistake is picking the winning team for the flashiest technology. If teams with developers always win, non-developer roles won't show up the next year. Holding it in December overlaps with the performance-review and HR season and lowers participation, so schedule it for October–November or February–March, and put two or three of the results on next year's training plan and AI project list so the event leads to lasting change.

Courses that cover this term · AI & DX
Recommended courses · Recommended instructors · Quick quote
📚AX Workshop: Start Business Transformation with AI Agents4 hours · In-person corporate training · John KimSee course →📚Building a Framework So AI Doesn’t Make Mistakes: Harness Engineering for Non-Developers4 hours · In-person corporate training · John KimSee course →
Instructor John KimJohn KimInstructor John Kim, author of B2B Marketing Design and CEO of Nanum ConsultingInstructor Seolyeong ParkSeolyeong ParkGenerative AI and AX training specialist with 17 years of hands-on experience and 200+ lectures, helping non-technical staff put AI to work right away
🧮 Mini quote · Based on instructor fees
About KRW 1.2 million (≈ US$900)2-hour talk · up to 20 people · Korean
Get a quote for these requirements →
💡 In plain English
It’s like skipping the cooking class where you only learn recipes and instead cooking a full meal in one day from what’s in your own fridge.
Why it matters in practice
Lectures alone don’t produce people who have actually applied AI to their own work. A hackathon is one of the few formats that yields use cases from the training budget.
Origin and sources

The term hackathon came from software developer communities in the late 1990s, combining hacking and marathon, and referred to an event where people gather for a short, intensive stretch of development. It later spread to startups and IT companies as a way to find new products. As generative AI made it possible to build results without coding, it became an internal training and innovation event that includes non-developer roles.

An example

Consider a hypothetical case. Company A, a food manufacturer with 1,200 employees, finished a company-wide AI introductory training in the first half of the year, but three months later usage had fallen to around 20%. The training team decided to hold a two-day internal AI hackathon in November using the remaining KRW 9 million (≈ US$6,730) of its KRW 30 million (≈ US$22,400) training budget.

  1. Ms. Lee from the training team grouped the themes into three tracks: 'report writing,' 'quality data organization,' and 'supplier inquiry handling.' She formed 14 mixed teams, each with 3 business-side members plus 1 IT member or AI champion.
  2. The head of security stated flatly, "Not a single line of customer data may be uploaded," and Manager Lee prepared three pseudonymized sample datasets and distributed them the day before the event.
  3. In the afternoon judging on day two, teams were scored on technical completeness (20 points), real-world relevance of the problem (30), scalability to other factories (30), and reproducibility (20). A no-code alert bot for abnormal quality readings, built by the quality team, took first place.
  4. At the awards ceremony, the CEO promised a KRW 20 million (≈ US$15,000) budget to scale up the top three results in Q1 next year and assigned one IT staff member to each team.
  5. Three months later, the quality team’s alert bot was in use at 2 factories, weekly company-wide AI usage had risen from 22% to 41%, and sign-ups for next year’s hackathon grew from 14 teams to 23.
Related terms
AI Use Case Discovery · Task IdentificationBreaking a department’s real work into individual tasks, finding the ones suited to AI, and ranking them by impact and difficulty to settle on a list of projects.AI Champion · Internal AI AmbassadorA business-side employee who, after AI training ends, builds use cases within their own department, answers colleagues’ questions, and drives adoption, or the program that selects and supports such people.AI PoC · PilotA PoC (Proof of Concept) is a small experiment to confirm that AI works for a given task. A pilot is a trial run in which some departments use it in real work to decide whether to expand company-wide.WorkshopA participatory group activity in which attendees discuss, build, and produce results together rather than just listening as in a lecture. Companies use it widely for training, problem solving, and team engagement.AI Use Case Contest · Internal Best Practice AwardsA program where employees submit results from applying AI to their own work in a set format, which are judged and awarded, and the selected cases are shared company-wide. If a hackathon is a day for building, a contest is a place to collect what people have already done.Internal AI Usage GuidelinesAn internal policy document setting out what employees must follow when using generative AI at work, such as information that must not be entered, responsibility for verification, and labeling of outputs.Design ThinkingAn innovation method that starts from empathy with users, reframes the problem, and solves it through rapid prototypes and experiments.
Training linked to this term
📚AX Workshop: Start Business Transformation with AI AgentsNanum Consulting corporate training · On-site or onlineSee the course →
Need in-house training on this topic?
Get a curriculum and quote tailored to your needs right away.
Ask About Training
📚 See related courses🗂 All programs📄 Free resources🔎 Glossary home
Last updated 17 Oct 2026 · Reviewed by John Kim

FAQ

Yes. Design it as a no-code format where participants build results without coding, such as making a chatbot, connecting automation tools, or using an agent builder. Still, each team needs one person familiar with the tools or an external mentor, or the day goes nowhere.
One and a half to two days is most common. One day is too short to produce real results, and with three days or more, team leaders won’t release people because of the gap in day-to-day work. If you collect problem definitions as pre-work, two days is enough.
Agree on the scope with the security team in advance. Exclude customer and personal information, and if needed, have the organizers prepare and distribute pseudonymized sample data beforehand. Which AI services data may be uploaded to is governed by your internal AI usage guidelines.
Costs such as instructor and mentor fees, venue, and materials are usually treated as training expenses. Prizes are often classified as rewards under company policy, so it is best to split the line items with the finance team in advance. Whether the program meets the requirements for a reimbursable course should be checked separately.