The Last Reorg Ever

The org chart is where companies actually commit to what they value. Everything else is talk. Deel knows this. They created a Ghostbuster role and put one in each function, reporting to its C-level. Two lines from the job req: “AI-native. You help every team rethink how they work, every single week,” and “Ghostbusters find broken things and fix them. Own the outcome end to end.” It’s an interesting role. The bigger move is the declaration on the org chart: this matters, this is the way forward.

Two years ago, GPT-4o launched. Suddenly, product, design, and engineering all had multimodal AI at their desks. The lines between disciplines blurred. The tools became a common language. But the org chart stayed the same. The SDLC stayed the same. The roadmap stayed the same. And a lot of things broke. Sprint rituals built for when design was the bottleneck. Design reviews that took longer than the work. Handoff protocols that assumed clean roles. A decade of cemented process running into a world where execution went cheap.

That gap, new tools, old chart, old process, is what every product org is wrestling with. It’s a compelling moment for design. And yet it’s the one discipline that hasn’t grown. Why?

Lenny Rachitsky put words to what a lot of leaders were noticing:

“Unlike PM and eng, which started growing in 2024, design didn’t… because AI is allowing engineers to move so quickly, there’s less opportunity — and less desire — to involve the traditional design process.”

Unpacked, the reason is structural. AI hit engineering first. Code got cheap to write, and value shifted to what hits the glass, not what sits in Figma. The functions with formal authority over shipping absorbed the productivity dividend. Design, which had the least chart authority to begin with, didn’t have a way to claim it.

None of this means traditional design process is dead. Discovery, research, system design, craft judgment still create value. What’s broken is the authority around them. Design does the work. The chart never gave design a veto.

The Ghostbuster role isn’t a new invention. It’s a Tiger Team or Special Projects role made permanent on the chart. The lineage is good news, not bad. Those patterns work because they let people go fast on hard things without distraction. The difference is the structural commitment. Tiger Teams were temporary. Ghostbusters live on the chart.

This is good news for experienced designers. Of the three disciplines that could claim this slot, design’s training maps cleanest to the work. We revel in what comes before: discovery, the user’s actual journey, the real problem hiding behind the stated one, the details that make an experience useful and delightful. That work wanders. Scope expands at times. Other times the simplest fix presents itself. Tangents emerge. All of it in service of solving the right problem. Finding what’s broken and building the fix is the job experienced designers have always wanted. Stripe’s Walk the Store, Airbnb’s UX-fix culture. The brands we love have operationalized this.

Human + AI working together is going to be the default. The companies that pull ahead will be the ones that prepare their orgs for the new way of working with intention. And roles on the org chart like Ghostbuster should become more common. If you’re a design leader, here are three places to start.

AI wants structure, rules, and data. For design, that means a mature design system: tokens, components, content patterns, codified, documented, ingestible. Without it, AI hallucinates your brand. With it, AI propagates your decisions at scale. The design leader’s job is to be in the system, not above it.

“AI Everywhere” doesn’t work as a delegation announcement. The design leader has to be hands-on with the tools: sitting in working sessions, pairing on prompt patterns, running design crits that include “show me your prompts” alongside “show me your mocks.” If the design leader isn’t fluent, the team won’t be either. Skill compounds where the leader does the work.

When everyone is building on top of AI, technical execution stops being the differentiator. Brand clarity is. The design leader’s job is to codify what your brand sounds like, looks like, and feels like, in a form AI can read. Voice guidelines. Editorial principles. The “no” list. Without that, AI defaults to generic. With it, your product sounds like you at scale. (See Martin Eriksson’s decision stack for the product layer; design needs its own.)

“Everyone is a builder” was a nice 2024 slide. In 2026, we have to operationalize it.

The pre-AI division of labor was tooling-based. PMs wrote PRDs in Notion. Designers made mocks in Figma. Engineers wrote code in their IDE. Each had a different deliverable, a different stack, a different handoff. The lines were clear because the tools enforced them.

AI collapsed the tooling moats. PMs, designers, and engineers all work in the same handful of tools now: Claude, ChatGPT, Codex. Same prompts. Same outputs. Same chat window. The deliverables blur. The handoffs blur. “Everyone is a builder” stops being a slide and becomes a real organizational design problem: when everyone can prompt their way to a working prototype, who decides what gets built, what ships, and what holds it all together?

The 2026 division of labor isn’t about tools or deliverables. It’s about what each discipline is responsible for and accountable for. We don’t have to reinvent the wheel; no new titles are required, no net new responsibilities. We just need to reinforce the areas of responsibility and distribute the trust.

The lanes themselves aren’t new. Some version of who-owns-what has been textbook for twenty years. What changed isn’t the lanes. It’s the enforcement. When tools drew the lines, accountability followed the artifact. Now everyone’s in the same tools, doing similar work. The lanes have to come from the chart, not the tools.

Product owns the case for building. Customer pain. Market timing. Business outcomes. The “why this, why now, why us” argument. PM has the authority to prioritize. And the authority to say no to building something that doesn’t earn its slot.

Design owns the case for shipping. Craft. Brand integrity. Recognizability. Editorial discipline. The “this is good enough to carry our name” argument. Design has the authority to delay launch. And the authority to say no to shipping something that drags the brand backward.

Engineering owns the case for sustaining. Feasibility. Architecture. Performance. Reliability. The “this won’t break next quarter” argument. Eng has the authority to push back on scope. And the authority to say no to architectural debt the team can’t afford.

Explicit lanes. What each discipline is accountable for and what they’re not. Real veto. Each discipline gets to say no to launch, not just feedback to ignore. Three vetoes, not two-and-a-suggestion. Making any of them structural, not personality-dependent, is the work.

This is the chart that holds. Built around who decides, not what they use, it doesn’t have to be redrawn every time the tools shift.

The biggest structural change is at the top. PM, Design, and Eng all report directly to the CPO. Not VP of Design reporting to a CDO. Not VP of Engineering reporting to a CTO. Three disciplines, one apex.

Design leaders may balk at this. Reporting to CPO sounds like a demotion from the CDO seat we’ve spent twenty years trying to get. Worth saying: the CDO seat can absolutely work. Deel kept theirs and made it stronger by housing a Ghostbuster under it. The argument here isn’t to replace CDO. It’s to put a single role above the discipline leads, where cross-discipline veto conflicts get resolved. Call it CPO, call it CPTO, call it whatever your org uses. The upgrade isn’t the title. It’s the veto.

Ed Biden talked a year ago about a CPTO role that’s both CPO and CTO. This could work if you can find someone qualified in both areas and a cultural fit for your org.

Lanes and vetoes only work if there’s a single role where conflicts get resolved. When two disciplines disagree, someone has to call it, and that call has to be made on product strategy, not on inter-VP politics. The CPO is the only role with the authority and context to make that call without rerouting. Anywhere else, the veto becomes a turf war.

This requires a real CPO. Not a “Head of Product who’s actually a senior PM.” Not a CEO doubling as the product owner. Someone who can articulate the product strategy clearly enough that the three disciplines underneath can run without escalating every decision. Get the CPO right, and the rest of the chart works.

Lanes, vetoes, a single apex. These are the bigger chart commitments. The Ghostbuster role works without them — Deel proved that. But it works more durably with them. The bigger chart commitments are what let the smaller ones compound, not just exist.

Back to Deel. The job req again: AI-native. Find broken things. Fix them. Own the outcome end-to-end. Read that and tell me it doesn’t describe a great product designer. Designers spend their careers obsessed with where users came from, what the real problem is, and what’s actually broken in the flow. They’re quickly becoming AI-native by trade. Prompting models all day to render, prototype, and analyze. They can build the fix, not just specify it. They speak product. They speak (enough) engineering. They speak brand. They own the outcome because the outcome is the experience.

I may be biased. But properly empowered, with explicit lanes and a real veto, designers would make superb Ghostbusters. Quite possibly the role’s most natural fit in any product org.

Deel named the role and gave it real seats across the chart. The rest of us have some pre-work to do and an org chart to review.

Good news: it doesn’t have to keep happening. Tools will keep changing. The chart won’t have to. Get it right once, and the team gets to spend the next decade building instead of restructuring. This is the last reorg ever.