NOW THIS IS SCRUM & AI - NEW ICON EDITION

Zoe Reinhardt Managing Director
3 min read in AI
(742 words)
published

How we keep to the rules and still move faster than everyone else!

Our philosophy at New Icon is people, process and technology and the Scrum Guide underpins our process and plays a major part of our approach to people. We want to break down how we use a guide from the 90’s relevant in a post-AI World, track by track. 

Each track takes a Scrum concept and translates it into the real delivery behaviours we apply as an AI-era systems integrator.

Track 01 |  Transparency beats intelligence

Scrum Value: Transparency

AI can produce answers, but delivery only works when the prompts, assumptions and outputs are transparent, auditable and understood by the entire Scrum Team.

Track 02 | The AI product owner

Scrum Role: Product owner | New Icon designers

We use AI to analyse requirements and surface edge cases early, then help draft clearer acceptance criteria so nothing relies on assumption. Our backlogs become shaped and intentional rather than guessed.

Humans still decide value, AI just expands the possibility.

Track 03 | Backlogs Are hypotheses

Scrum Artifact: Product backlog | New Icon artifact: Design sprint 

We create outcome-based items instead of feature requests so we’re solving problems, not prescribing solutions. We’ll define problem statements with you and AI helps expand scenarios and uncover missed cases, turning uncertainty into small, testable slices. Because most backlogs aren’t plans, they’re assumptions waiting to be proven wrong.

Track 04 | Sprint planning without guessing

Scrum Event: Sprint planning

We use AI to surface complexity and integration risk before we commit to anything, so surprises happen early not late.

We then break work down properly, often live in our collaborative “priority bingo” sprint planning sessions with clients and pairing humans and AI so everyone actually understands what “Done” will be.

Estimation becomes a risk conversation, and planning is about aligning confidence together, not pretending we can predict time or have access to Bernard’s watch! 

Track 05 | Developers, not typists

Scrum Role: Developers

Our engineers aren’t typists, they’re system thinkers. AI can write code, but developers design how the whole thing actually works, and the faster code comes out the faster unstable and unscalable architecture shows up.That’s why engineering maturity matters more now than ever, AI amplifies capability, but it also amplifies mistakes.

Track 06 | Daily Scrum: The 24-Hour risk radar

Scrum Event: Daily Scrum

Daily Scrum’s are still our best friend and we’ve decided to keep them human only. At the same time everyday, we use the same 15 minutes to connect, inspire self-management and resolve any impediments together. 

If you’re lucky and catch us on filter Friday’s, you will snap us as a mixture of pirates, pets and mysterious moustaches. We still stick to our 15 minute slot and cover what happened yesterday, what’s our plan for today and most importantly, are there any blockers?

Track 07 | Definition of done in the AI era

Scrum Artifact: Definition of Done

AI helps us move faster, we can all use AI to re-write our user stories and definitions of done, but “done” definitely doesn’t mean it compiles. Security, observability and maintainability are part of the work, not extras, and any AI-generated code means the Definition of Done has to be stricter. 

Velocity will wannabe your lover but it’s gotta get with our DoD’s.

Track 08 | Reviews: Showing reality, not slides

Scrum Event: Sprint review

Sprint Reviews give everyone a clear, practical view of progress using real behaviour. We demonstrate real behaviour and often run AI-generated scenarios live so stakeholders can see outcomes, not just UI. It stops expensive imagination, we validate a concept and what works before we scale it.

Track 09 | Retrospectives: Continuous adaptation engine

Scrum Event: Retrospective

In retrospectives we use AI to spot patterns across incidents so we’re not relying on memory or opinion. For example, it might highlight that most late defects came from integration assumptions rather than coding errors, which tells us the problem is discovery, not testing. Humans then solve the causes, not the symptoms. The process evolves every sprint, continuous improvement is something we run as a system, not something we do when we feel like it.

Track 10 | Scrum isn’t slowing us down, it’s letting us go faster safely with AI

AI massively increases output, but without control it just increases risk at the same speed. Scrum gives us the guardrails so change lands safely, not expensively,  less Top Gear chaos, more well-planned Grand Designs.

Together they let us move quickly and still stay predictable.

 


Zoe Reinhardt Managing Director at New Icon

Join the newsletter

Subscribe to get our best content. No spam, ever. Unsubscribe at any time.

Get in touch

Send us a message for more information about how we can help you