Dubai Municipality recently opened what it's describing as the world's first AI-powered park design challenge - a pilot inviting architects, urban planners, landscape designers, researchers, startups and AI specialists to reimagine how public parks get designed, using AI as a working part of the process rather than a gimmick layered on top of it. On the surface, this reads as a story about urban planning. Look at how the challenge is actually structured, though, and it's something more useful to anyone running an AI programme inside an enterprise: a live, public example of disciplined AI governance, applied outside the usual banking-and-retail case studies we normally reach for.
I want to walk through what the challenge is actually asking participants to do, and then make the case for why the shape of it - not the park design outcome - is worth studying if you're responsible for AI adoption inside a GCC business.
What the Challenge Is Actually Asking For
Strip away the "park design" framing and the challenge is asking participants to demonstrate something specific: that an integrated set of AI tools can meaningfully support an entire design journey - from analysing a site and understanding the people who'll use it, through concept generation and testing different scenarios, all the way to a workable, buildable proposal - while the humans running the project keep authority over the final call.
A few structural details stand out:
It spans the entire lifecycle, not one step. Participants aren't asked to use AI for a single task - generate a rendering, say, or crunch some site data. They're asked to show AI woven through analysis, ideation, scenario testing, optimisation and visualisation as a connected process.
Final decisions stay human-led, by design. Despite the heavy AI integration required, the brief is explicit that design decisions remain with the people, not the tooling. AI augments judgement; it doesn't replace it.
It's explicitly multidisciplinary. The eligibility criteria bring together designers, architects, technologists, researchers and AI specialists - deliberately, not as an afterthought - and teams are required to name a single accountable lead.
Evaluation isn't just "did you use AI." Submissions are judged on the strength of the AI-integrated approach, yes, but equally on spatial feasibility, inclusivity, sustainability, and - critically - how well the entrants translate their data analysis into a practical, buildable outcome. A technically impressive AI process that doesn't produce something feasible and human-centred doesn't score well.
It's built to produce prototypes, not just concepts. The stated goal is scalable ideas that can actually be tested as real, physical prototype parks - not a portfolio of slide decks.
That's a genuinely well-designed brief. And if you swap "park" for almost any enterprise process, it reads like a checklist for how AI initiatives should be scoped and judged inside a business.
Why This Is a Better AI Governance Model Than Most Enterprises Actually Use
I've written before about the most common way enterprise AI programmes go wrong: starting from the technology instead of the business process, skipping the discipline of scoring use cases on feasibility and value, and treating a compelling demo as though it were a business case on its own. What's striking about this challenge is that its structure quietly avoids every one of those failure modes - not because it was designed as a corporate AI governance framework, but because good judgement about AI adoption looks the same whether you're designing a park or a claims process.
AI across the full journey, not a bolt-on. The most common mistake I see in enterprise AI pilots is treating AI as a point solution - a chatbot bolted onto a support queue, a model bolted onto an existing dashboard - rather than as something woven through an entire workflow. The park challenge's requirement that AI support the whole design journey, from analysis through to visualisation, is exactly the standard enterprise AI initiatives should be held to. A use case that only touches one narrow step of a process rarely produces the value a broader, end-to-end integration does.
Humans stay accountable for the outcome. This is the single most important detail in the entire brief, and it's the one enterprises get wrong most often. AI can meaningfully accelerate analysis, generate options, and surface patterns a human might miss - but the accountability for the final decision has to sit with a person, especially wherever the outcome affects customers, employees, the public, or a regulator's judgement of your business. A park is a public, physical, human-facing space; final design authority staying human isn't a limitation on the AI, it's the correct governance model. The same logic applies directly to credit decisions, hiring processes, medical triage, and any other enterprise AI use case with real consequences for real people.
Multidisciplinary teams, not siloed data science. The challenge doesn't just permit collaboration between designers and technologists - it structurally requires it, with a single accountable lead. Enterprises that run AI initiatives out of an isolated data science function, disconnected from the business owners who actually understand the process being changed, consistently produce technically interesting work that never gets adopted. The organisations that get this right build cross-functional teams from day one, exactly as this brief mandates.
Multi-dimensional evaluation, not "did you use AI." Judging entries on feasibility, inclusivity, sustainability and the translation from data to practical outcome - not simply on how much AI was used - is precisely the scoring discipline I'd want any enterprise applying to its own use-case pipeline. "We used AI" is not a success metric. "This is feasible, valuable, and works for the people it affects" is.
Prototypes over concepts. Requiring that winning entries be testable as real prototypes, rather than rewarding the most polished concept deck, mirrors the proof-of-concept-first discipline that separates AI programmes which reach production from the ones that stall as permanent pilots.
What This Means If You're Running AI Adoption Inside a GCC Business
You don't need to be designing parks to take something practical from this. A few things worth taking directly into how you structure your own AI initiatives:
Map the whole process before choosing where AI touches it. Don't scope a use case around a single AI tool. Scope it around a business process, end to end, and then identify every point along that process where AI can meaningfully contribute - analysis, option generation, testing, optimisation, communication of the outcome.
Keep a named human accountable for every AI-assisted decision that matters. Not as a compliance afterthought, but built into how the initiative is structured from the outset - the same way this challenge requires a named team lead and preserves human design authority throughout.
Build the team around the process, not around the AI. Pair the people who understand the business process with the people who understand the AI tooling, from the very first workshop - not as a handoff between two departments that rarely talk to each other.
Judge initiatives on more than "we deployed AI." Score use cases - and pilot outcomes - against feasibility, inclusivity of the people affected, sustainability of the approach, and whether the data-driven insight actually translates into something practical. A dashboard nobody acts on isn't a win.
Demand a testable prototype before a full rollout. If a use case can't be proven against a real, working prototype - not just a proposal - it isn't ready for production investment yet.
A Genuinely Useful Signal for the Region
Beyond the governance lessons, initiatives like this one signal something worth paying attention to in its own right: GCC public institutions are moving from talking about AI adoption to structuring real, judged, outcome-based programmes around it - with meaningful incentives (a AED200,000 prize pool across three placements, open to professionals, researchers, startups and AI specialists alike, with entries due by 25 August 2026) attached to producing something that actually works, not just something that sounds innovative. That's the same bar enterprise AI programmes across the region should be holding themselves to.
If you're building out an AI strategy for your organisation and want a second opinion on whether your current initiatives would survive this kind of scrutiny - full-journey integration, human accountability, multidisciplinary ownership, and a credible path from concept to working prototype - that's exactly the conversation our AI Consulting & Strategy practice has with clients. Get in touch, and let's pressure-test what you're building before you scale it.