You want to start with AI,
but where do you begin?
We paused our daily lives and ran our own hackathon. This site shows what came out of it and how we can help others do the same: an AI experience designed by the data experts at i-spark.
What We Offer
DIY Playbook
Free
Ideal for: Teams that need a solid starting point with AI exploration.
What you get:
- Our pragmatic hackathon playbook
- Agenda
- Checklists
Typical outcome:
Everything you need to kick off your own AI hackathon: structure, checklists and a clear agenda. You will have a solid foundation to get moving. Especially if you're testing the waters.
2-Day Sprint
Kick-off + Tool Introduction
Ideal for: Small groups wanting momentum & hands-on help
What you get:
- Everything in the DIY Playbook
- Day-1: guided kick-off & stream selection
- Day-2: live tool tutorials & rapid prototyping support
Structure we bring:
- Lightweight project plan
- Sprint board
- Role suggestions
Guided streams:
Choice of 2 streams
Typical outcome:
Working proof-of-concept
5-Day Deep Dive
Full-Week Facilitation
Ideal for: Companies ready to carve out a full innovation week
What you get:
- Everything in the 2-Day Sprint
- Five-day facilitation by our AI coaches
- Daily stand-ups, checkpoints & demo day production
Structure we bring:
- End-to-end schedule
- Stream leads
- Roles (Dev/Text/Ops)
- Progress dashboards
Guided streams:
Up to 4 parallel streams
Typical outcome:
Production-ready pilot, team up-skilled
Not sure which is right for your team? Book a quick call with an i-spark team member and we'll help you choose the best path.
AI Hackathons in Action
Our Own AI Hackathon at i-spark
June 2025 • 5 days, 3 streams, real results (and real struggles)
We didn't just create this concept we put it to the test ourselves. Our team of 10 dove into a full week of AI experimentation, discovering firsthand that "easy AI automation" is anything but easy.
The 3 Streams We Ran:
Stream 1: Website Creation
Building ai-spark.nl by non-webdesigners
- Generated content and designs using AI tools
- Mid-week pivot when initial approach didn't work
- Result: The website you're reading now
Stream 2: dbt Monitoring
Automated data pipeline monitoring
- Built webhook detection for failed dbt runs
- Created AI agents to diagnose errors and suggest fixes
- Result: Working Slack notifications with root cause analysis
Stream 3: Process Automation
Learning by doing
- Explored Make and n8n through hands-on tutorials
- Built first automation workflows from scratch
- Result: Foundation for future automations
The Unfiltered Week: What Really Happened
Day 1: Reality Check
The enthusiasm was high, but reality hit hard. Despite LinkedIn making it look easy, some struggled with n8n and AI integrations. Key learning: "Anyone can work with AI" is a myth – you need technical skills for real automation. We ended the day with more questions than answers, but solid use cases identified.
Day 2: Finding Our Feet
We introduced tutorial sprints after realizing we needed structured learning. Tested multiple LLMs (ChatGPT - OpenAI, Claude - Anthropic, Gemini - Google, Le Chat - Mistral AI) and discovered huge differences in their approaches. Security wake-up call: You can connect AI to many systems, but stay aware of all the access this creates and apply your compliance procedure accordingly.Updated our compliance policies on the spot.
Day 4: Deep Work
Heads down building. Stream 2 tested different failure types and AI models. Stream 1 pushed through website development. Stream 3 tackled increasingly complex automations. The MCP server integration proved trickier than expected. See more on Make.com.
Day 5: Making It Real
Website went live. dbt monitoring moved to production. Process automations documented and scheduled. Not everything was perfect, but everything was real and working. Our final playbook was documented in Notion.
What Actually Worked
- Having focused streams with clear deliverables
- Peer learning and shared struggles
- Tutorial sprints when we hit walls
- Time pressure forcing decisions
What We Struggled With
- Tools behaving differently than documentation claimed
- Features disappearing after updates (looking at you, Canva API)
- Lack of proper documentation for advanced integrations
- Technical complexity despite "no-code" promises
Our Biggest Learnings:
- 💡 AI won't write flawless workflows – expect multiple iterations
- 💡 You can't automate what you haven't documented
- 💡 The gap between demo and production is real
Tangible results in 5 days
10
participants
3
working solutions
100%
would do it again
"You really need to structure your process before you automate it. That reality hit us.”
Jeroen Visser
"This week, I not only learned new tools, but also saw how our role is shifting: from building models ourselves to smartly deploying and supervising AI."
Frans Melenhorst
"It’s systems, workflows, and plenty of trial and error."
Irene Oostindiër
Why a Hackathon makes Sense
Turn your organization's "we should do something with AI" into working prototypes in just one week.
Low risk, high-speed momentum
Move from zero to working prototype without risking big investments
Learn by building
People adopt AI when they build with it, not when they sit through presentations
Everyone contributes
Cross-functional teams each own a piece, together shipping one integrated solution
Freedom with focus
Our proven structure lets teams experiment while staying on track
Energy that delivers
Healthy competition and rapid wins create tangible assets that inspire what's next
Let's Talk
Turn one week into a year's worth of AI progress. Drop us a line—or grab our one-pager first—and let's see how an AI Hackathon could spark your next breakthrough.
Get in Touch
Whether you're exploring our AI Hackathon, need guidance to host one yourself, or just want to share feedback, we're here to help every step of the way.
Register for the Free AI Hackathon Playbook
Just want the essentials? Drop your email below and we'll send you our one-pager pragmatical playbook for a do-it-yourself, week-long AI Hackathon.