When Intelligence Enters the Org Chart

AI Is Changing the Company, Not Just the Campaign

Before the main story, a few headlines worth your attention.

• OpenAI’s newly released GPT-5.5 shows meaningful gains in professional work, especially for coding, research, data analysis, documents, and tool use. It performs strongly across major benchmarks, including GDPval, OSWorld-Verified, and Tau2-bench Telecom. GPT-5.5 Instant reduced hallucinated claims by 52.5% versus GPT-5.3 Instant on high-stakes prompts, suggesting stronger reliability for complex work. More at OpenAI.

• Middle managers are being asked to drive AI usage through dashboards, incentives, and practical workflow ideas, even as some companies reduce traditional management layers. The story highlights a leadership paradox: managers may be at risk from AI, but they are also essential to scaling it across the workforce. Business Insider has the story.

• Anthropic launched Claude Design is a new research-preview product that lets people create polished visual work, including prototypes, slides, one-pagers, landing pages, and campaign assets, by collaborating with Claude. It can ingest brand systems, codebases, documents, and websites, then help teams refine designs through comments, edits, sliders, and conversation. More at Anthropic.

• OpenAI is making ChatGPT ads easier to buy by launching a self-service ads manager in beta for U.S. advertisers, along with new partners including Pacvue, Kargo, and StackAdapt. Advertisers can now buy on a cost-per-click basis, signaling ChatGPT’s shift from experimental ad inventory to a more scalable performance-marketing channel. Adweek reports on the developments.

• Cohosted the AI-Verse at POSSIBLE recently, where I interviewed Michael Lacorazza, CMO of U.S. Bank; Gail Horwood, CMO of Novartis, U.S.; Vineet Mehra, CMO of Chime; and others on the real AI transformation taking place inside corporations. I presented on Memory in AI as the next competitive advantage. Also this month, further AI strategy consulting work across enterprises, more AI training for companies, and two speaking engagements at Fortune 500 companies.

---

We’re Thinking About AI the Wrong Way

For the last two years, most business conversations about AI have been trapped in the wrong metaphor. We keep calling it a tool: a productivity tool, a creative tool, a research tool, a coding tool, a customer service tool, a cheaper, faster, always-on tool that can help people do more with less. That framing is not wrong exactly. It is just too small.

A tool waits to be used. A tool does not reason across systems, remember past decisions, coordinate workflows, simulate audiences, generate strategy, develop ideas, write code, negotiate trade-offs, or increasingly act on behalf of the organization. AI does. That is why the more useful starting point may be the one Dario Amodei, the Anthropic Co-Founder & CEO has been pushing us toward. AI is not simply another technology wave. It is something that may test who we are as a species, because we are being handed a level of power our institutions, companies, and social systems may not yet be mature enough to wield well.

That may sound dramatic in a marketing context, but it is not. Even inside marketing, one of the most visible functions now being reshaped by AI, the deeper question is already surfacing:

Are we using AI to make old systems faster, or are we redesigning marketing around a new kind of intelligence?

Right now, most companies are still doing the former. They are adding AI to workflows built for email, meetings, PowerPoint, agencies, approvals, campaign calendars, and quarterly and annual planning cycles. In other words, they are bolting artificial intelligence onto operating models that were never designed for it.

Coinbase is one of the more recent companies to make that shift explicit. Its ambition to become AI-native points to a different operating model, one where intelligence sits closer to the center of the business and humans play more deliberate roles in direction-setting, judgment, oversight, and accountability. The harder implication is that this kind of redesign may also mean fewer employees as more work is reorganized around AI. That is the part many leaders still do not want to say out loud. AI is not simply being added to the organization. It is beginning to change the organization. And that means the questions we ask need to change too.

1. “Human in the Loop” Is Not Enough

The tool metaphor leads to tool-level thinking. Leaders ask which platform to buy, which use cases to prioritize, which prompts to teach, and which workflows can be made more efficient. Those are useful questions, but they are transitional ones. The more important questions are about authority, judgment, accountability, and the design of the work itself.

This is where one of the most comforting phrases in enterprise AI starts to fall apart. “Human in the loop” sounds responsible. It signals that people are still involved. It reassures legal, compliance, HR, management teams, and boards that the machine is not running wild. But as an operating model, it is too simplistic. A human can be in the loop and still not be meaningfully in control. A human can review something without understanding how it was produced. A human can approve something because the workflow requires approval, not because they have exercised real judgment.

The better question is not whether a human is in the loop. It is what role the human is playing. Sometimes the human should be driving: defining the problem, setting the direction, owning the decision, and carrying the accountability. Sometimes the human should be orchestrating: coordinating AI systems, data sources, internal teams, external partners, approval flows, and business objectives. Sometimes the human should be augmenting: bringing taste, context, judgment, ethics, brand understanding, and institutional memory to make the AI’s output better. There are also moments when the human is conducting rather than directly doing the work. They are not playing every instrument, but they are shaping the tempo, sequencing the work, managing dependencies, and deciding when the whole system is ready for the next movement. And there will be moments when the human is supporting, where the AI does the first pass, the heavy synthesis, the routine analysis, or the repetitive production, while the human verifies, redirects, or intervenes only where needed.

These are not semantic distinctions. They are organizational design choices. A company that does not define them explicitly will default to confusion. Some teams will over-control the work and slow everything down. Others will over-delegate to AI and create risk. Many will end up in the messy middle, where everyone says “human in the loop” but no one knows who is actually responsible for the decision. The messy middle is where many AI efforts stall: too much delegation to feel safe, too much control to create leverage, and not enough clarity to produce meaningful results.

2. AI Will Change Work Before Jobs Go Away

The same fuzziness shows up in the jobs conversation. The optimistic version says AI will create more jobs than it destroys (AI boomers). The pessimistic version says AI will eliminate entire categories of white-collar work (AI doomers). Both are too confident. The more honest answer is that we do not know yet.

OpenAI’s recent AI Jobs Transition Framework is useful precisely because it moves away from the simplistic question of whether a job is “exposed” to AI. It looks across more than 900 occupations covering 99.7% of U.S. employment and tries to understand where labor market pressure may emerge by combining technical exposure, human necessity, demand elasticity, and observed ChatGPT usage. That is the right level of nuance because “AI can technically do part of a job” is not the same as “the job disappears.”

Some work still depends on trust, judgment, accountability, human relationships, physical presence, regulatory interpretation, or deeply contextual expertise. Some work may become cheaper, which can increase demand. Some roles may shrink. Some may become more valuable. Some may be redesigned so thoroughly that the old job title becomes almost meaningless. AI changes tasks before it changes jobs. It changes workflows before it changes org charts. It changes decision rights before it changes titles.

That is why leaders should be careful about making sweeping workforce claims too early. The more useful work right now is not predicting the exact number of jobs that will disappear. It is mapping how tasks, decisions, approvals, workflows, and accountability are shifting between people and machines. The companies that understand this will not simply use AI to produce more of the same. They will rethink what the work is, who owns which decisions, and where human judgment creates the most value.

This is especially important for marketers, but the lesson applies well beyond marketing. We have seen a technology wave before. Digital marketing rewired channels, targeting, measurement, media buying, customer journeys, and the economics of attention. It changed a lot. But it did not change the fundamental nature of cognition inside the company. AI does. AI is not just another channel shift. It is not search all over again. It is not social all over again. It is not programmatic all over again. Those were changes in distribution, data, and media architecture. AI is a change in how organizations think, create, coordinate, remember, and act.

3. Regulated Industries May Not Be Behind

There is another myth taking hold, especially in marketing circles, that regulated industries will inevitably move too slowly and get left behind. Healthcare, financial services, insurance, pharmaceuticals, and other trust-sensitive sectors are often described as too cautious, too governed, too approval-heavy, and too constrained to fully benefit from AI. Maybe some will be. But that is not the whole story.

At POSSIBLE’s AI-Verse, when I interviewed Gail Horwood, Chief Marketing Officer of Novartis US, who grew up in digital marketing as I did, she made a point that stayed with me. In regulated industries, you often have to go slow to go fast. Not because the ambition is lower, but because the foundation matters more. In a regulated industry, AI adoption cannot be treated as a clever demo followed by a press release. The data is too sensitive. The customer trust is too valuable. The ethical boundaries are too important. The legal and compliance implications are too real.

Going slower at the beginning is not the same as falling behind. It can be the precondition for moving faster later. The companies that do the hard foundational work now, from data readiness to governance, workflow redesign, approval pathways, risk classification, content standards, escalation rules, and cross-functional operating models, may be the ones that can scale AI with more confidence later. That work rarely looks exciting. But it may be what separates organizations that experiment with AI from organizations that actually transform with it. It is also why Anthropic’s launch of AI agents for financial services matters. The move reflects a broader belief that regulated industries, far from being left behind, may become some of the most important arenas for enterprise AI.

4. Boring Work Makes Breakthroughs Possible

This is the part of AI transformation that rarely gets enough attention. Truthfully, it is where my team and I spend most of our time. Everyone wants the breakthrough use case: the agentic workflow, the AI-generated campaign, the synthetic audience, the self-optimizing media plan, the personalized customer experience, the dashboard that talks back, the prototype that appears in seconds. But the breakthrough usually depends on work that looks far less glamorous: metadata, taxonomies, data hygiene, content rights, brand standards, legal parameters, institutional memory, workflow maps, decision logs, governance rules, evaluation rubrics, feedback loops, and source-of-truth systems.

None of this feels like the future. It feels like plumbing. But in the AI era, the plumbing is the strategy as well. Without that foundation, AI becomes a clever intern with no context, no standards, no memory, and no accountability. It can produce impressive outputs and still make the organization dumber. It can move fast and still move in the wrong direction. With the right foundation, something else becomes possible. AI can help teams see across disconnected systems, surface patterns buried in old work, connect strategy to execution, make institutional memory usable, prototype scenarios before resources are committed, and compress learning cycles.

That is when AI starts to become more than a productivity layer. It becomes an operating layer. This is also why AI creates such a strange tension for marketers. AI reflects history. It is trained on what already exists including the language, campaigns, tropes, assumptions, best practices, biases, benchmarks, and artifacts of the past. Marketers, however, are supposed to imagine the future. If every brand asks the same models the same questions using the same public data, the same category conventions, and the same generic prompts, the output will converge into AI-slop. The work may look polished. It may even test well. But it will increasingly feel like a better-formatted version of what already happened. That is not leadership. That is nostalgia with a faster engine.

Where does this leave leaders?

The marketer’s job, and increasingly every leader’s job, is not simply to use the machine. It is to prevent the machine from pulling the company toward the average. It is to bring the human imagination, cultural read, strategic courage, and business judgment that the model cannot generate on its own because the future has not happened yet. It is also to think strategically and thoughtfully about the exact role humans should play in an intelligence system, how to build for those new roles, and how to find new ways to measure effectiveness and performance.

This is why the “tool” metaphor is so limiting. A tool helps you do what you already intended to do. AI forces you to ask whether the thing you intended to do still makes sense. The next era of marketing will not be defined by who has the best prompts or the most tools. It will be defined by who redesigns the work, clarifies the role of human judgment, builds the foundations, and understands that intelligence is becoming an operating layer of the firm.

The wrong question is what AI can do for marketing.

The better question is what kind of marketing organization becomes possible when intelligence is no longer scarce and sits at the heart of the function, with people surrounding it to provide direction, imagination, judgment, and accountability.

Where I’ve been

AI in marketing has moved beyond the prediction phase. We are now in the proof phase. The AI Verse, jointly produced by AI Trailblazers and POSSIBLE, was built around one idea: Show, Don’t Tell. Across three days of programming, we explored how AI is already reshaping strategy, insights, creative, media, operations, and the broader marketing function.

Highlights included conversations on AI creative factories inside marketing with Vineet Mehra, CMO of Chime, and Toygar Bazarkaya, CCO of RECE; the Answer Engine Economy with Hilary Batsel, VP, Marketing at LinkedIn, Pete Blackshaw, CEO of BrandRank.AI, and Kirthiga Reddy, CEO of OptimizeGEO; agentic AI with Amit Shah, CEO of Instalily AI; the future of the marketing function with Lynn Teo, CMO & IAA board member and AI agents in action with Padma Hari, Chief Digital Officer of Nestlé Purina NA, Vivek Vaidya, CTO of Kana, and Aaron Fetters, CEO of Transparent Partners.

We also explored advertising in the age of AI with Puru Patnekar, VP Marketing at Charter Communications, David A. Steinberg, CEO of Zeta Global, and Bob Lord, President of Horizon Media; function transformation with Don McGuire, EVP and CMO of Qualcomm, interviewed by Greg Kahn, CEO of GK Digital Ventures; regulated-industry innovation with Gail Horwood, CMO of Novartis, U.S.; intelligence inside AI with Andrew Swinand, CEO of ITG; and synthetic personas at scale with Michael Lacorazza, CMO of U.S. Bank. The through line was clear: AI is no longer a side experiment. It is becoming a core operating layer for modern marketing.

What I’m reading

• The AI Jobs Transition Framework (OpenAI)

• Introducing Claude Design by Anthropic Labs (Anthropic)

• This Is How We Get Moral A.I. Companies (NY Times)

• Wall Street’s plan to build the ‘McKinsey of AI’ (Business Insider)

• TikTok Welcoming Third-Party Agents for Ads (Adweek)

What I’ve written lately

• The End of the Brief as We Know It (April 2026)

• A Public Company Goes All In on AI (April 2026)

• AI Is Rewriting Who Decides (March 2026)

• Fighting Cognitive Surrender (March 2026)

• Who Remembers Wins (February 2026)

  ---

Shiv Singh is the CEO of Savvy Matters, which helps business teams translate AI disruption into practical business and marketing strategies, organizational design, executive-ready roadmaps, and bespoke education programs. He is also the Co-Founder of AI Trailblazers, a vibrant community uniting marketers, technologists, entrepreneurs, and venture capitalists at the forefront of AI.

A former two-time Chief Marketing & Customer Experience Officer and author of Marketing with AI for Dummies (4th print run, translated into five languages), Shiv built his career at LendingTree, Visa, PepsiCo, and The Expedia Group, and serves as a public-company board member of a Fortune 300 company and private investor.