What you need to know

• AI’s ‘Seventh Wave’ will be the most disruptive tech shift yet, says Forrester CEO George Colony, eclipsing past revolutions like mobile, cloud and the internet. He predicts it will gut SaaS, slay traditional search, and collapse bloated software pricing models.

• Legacy vendors are masking incremental upgrades as AI breakthroughs. Colony warns CMOs to beware “AI-enhanced” products from incumbents.

• Agentic AI threatens to upend business models by shifting pricing from licences and usage to outcomes, for example call centre tools priced by resolutions, not seats. SaaS firms face massive margin pressure and potential collapse. 

• SaaS buyers should hit pause. Colony advises delaying long-term SaaS contracts, arguing the economics of seat-based software are unsustainable in an agent-led world.

• Shadow AI is coming. As in the early SaaS era, business units will bypass IT to deploy AI tools. But the risks, spanning compliance, security and trust, also return.

• Traditional search is dead; the open web is dying. The data bears him out. Similarweb data shows huge traffic declines and rising “zero-click” searches.

• Marketers must prepare for an agent-driven martech stack. Campaigns, segmentation, and optimisation functions will collapse into autonomous systems that learn and act in real time.

• He predicts a new AI-native social platform will challenge Meta. Colony sees an opportunity for agentic alternatives to today’s algorithm-heavy social giants.

• The coming cartel: model makers and the new business they spawn, not SaaS vendors. Colony predicts consolidation in LLM providers; Forrester is bearish on OpenAI’s enterprise traction despite strong models.

• Trust is the cornerstone. Agentic systems act, not just suggest. Colony insists marketers must audit and manage agents like human staff. Poorly governed AI poses strategic risk.

• Then there’s Colony’s Law: Information will double in value each year. Generative AI raises the value of information by slashing its extraction cost, shifting marketing value from content creation to interrogation and orchestration.

• Jobs won’t vanish, but will be redefined. Colony’s Iron Man metaphor: humans using AI will outperform, while those who don’t risk obsolescence. CMOs must upskill their teams now. 

• Get the full download – listen to the podcast here.

AI’s great convulsion will gut SaaS, kill traditional search, roil capital markets and VCs, overwhelm the early LLM pioneers, displace global digital giants, lead to the creation of an agentic social media network to challenge Facebook, and relegate the open web to the backwash. So says George Colony, clearly not in the mood for understatement.

The founder, Chairman and long-time CEO of Forrester Research has seen computing come in waves: the mini-computer, the PC, the internet, social, mobile, and cloud. Each shook markets, minted fortunes, and broke incumbents. But he now insists the Seventh Wave, artificial intelligence, will be the most violent of all.

“This will be the biggest technology change of my lifetime,” he says.

According to Colony, for years, AI was the tech sector’s most repetitive punchline, always the breakthrough of next year. Then OpenAI’s ChatGPT arrived in late 2022, reframing human–machine interaction with a user interface so obvious it seemed inevitable. 

Waves, he tells Mi3, are triggered not by technology alone but by shifts in how humans interact with it. The GUI gave life to the PC; the browser birthed the internet; the touchscreen made mobile inevitable. ChatGPT provided AI’s clickable moment. And as with earlier surges, the promise is as much destruction as creation. Incumbents scramble. Business models unravel. Customers acquire leverage. And the laggards are swept aside.

This time, however, the industry is larger, the pace is faster, and the level of ferment and turmoil far greater. The internet took years to reach mass adoption; AI spread globally in months – OpenAI’s ChatGPT hit 700m weekly users last month. 30 months ago virtually no one outside the Valley had ever heard of it. Even the companies that built it, Colony says, struggle to predict what comes next.

“There is a general sense of fear at all of those companies that they can be bypassed, that their positions will be destroyed by this wave.”

And fear is contagious.

“Every one of these legacy companies, they are paranoid beyond measurement. At this point, they’re all scared that they’re going to be bypassed. So fear is everywhere, and they’re all in a defensive mode.”

What a nice wooden horse

Colony has a biblical turn of phrase. “Beware the legacy vendor bearing AI gifts,” he counsels. The warning matters for CMOs and agency leaders who are already bombarded with “AI-enhanced” everything: AI copilots in office suites, AI targeting in CRMs, AI optimisation in adtech. The trick is less about innovation than disguise.

“They’re claiming to be AI, and they’ll say they’re AI, but are they truly AI? The best posture for a buyer right now, for a user, whether you’re a CIO or a CMO, is to have a high AIQ, what we call an artificial intelligence quotient. You need to be educated. You need to understand this. You need the ability to sort the sheep from the goats.”

The playbook is familiar. Incumbents buy promising start-ups, bolt on thin AI veneers, or pretend that minor features amount to breakthroughs. Such tactics may protect quarterly earnings but do not alter fundamentals. “Yes, they will offer AI,” Colony allows. “But not at the price points they have historically been charging”.

And here lies the crux: AI is a business-model killer. Enterprise software was built on licences, seats and SaaS subscriptions. Agentic systems promise to do the same tasks for a fraction of the cost, perhaps as little as a tenth, Colony says. For marketing and media firms long resigned to eye-watering contracts, that inversion of economics is revolutionary.

Yet it will not be painless. When firms like Oracle or SAP are threatened with the loss of up to nine-tenths of their revenues, they will not go gently into the night. They fight with cash piles, lobbying muscle and defensive acquisitions. The “buy, block, pretend, link” strategy, as Colony calls it, will be deployed with gusto. But in the long run, the arithmetic is unforgiving. Pricing models that made sense when seats were plentiful and software was bloated cannot survive in a world where autonomous agents replicate tasks cheaply and at scale.

SaaS on the brink

Software-as-a-Service was once the insurgent. In the early 2000s, Salesforce and Workday snuck in through marketing departments and HR divisions, bypassing wary CIOs. Today SaaS is the establishment, a trillion-dollar layer of the stack, priced by seat and padded with features most customers rarely use.

Colony predicts that this edifice is about to collapse. “If you are looking to buy a traditional SaaS platform in the next two years, wait,” he advises. “Or sign very short-term deals. Do not make big commitments.”

The reason is not merely technical. It is economic. Generative and agentic AI create an incentive to shift costs from fixed licences to variable usage, then shift them further, from usage to outcomes. Call-centre software may soon be priced by resolution, not seats. Financial systems, perhaps by reconciliations completed, not modules installed. Colony expects vendors will cling to the SaaS model in their AI early days, but prices will tank. The industry has yet to agree on what this means.

“This is totally unresolved,” Colony says. “What we are seeing now is more agent-ish than agentic.”

Capital markets, meanwhile, are overexposed. Venture portfolios are full of yesterday’s SaaS stars; public investors hold richly valued incumbents whose multiples rest on pricing power about to be destroyed. Colony’s verdict is withering: SaaS, the insurgent of one wave, may become the casualty of the next.

The one-two combo

Colony draws a sharp line between generative AI and agentic AI. Generative unleashes content, much as Gutenberg’s press multiplied pamphlets. Agentic rewrites systems, collapsing campaigns, processes and decisions into self-adjusting loops. The difference matters for marketers. Generative may churn out copy and imagery. Agentic promises to make planning, targeting and optimisation autonomous.

This is the logic of a kind of ‘univergence‘: the inexorable convergence of decision-making into a universe of machine-led systems. Today’s martech stack comprised of campaign managers, segmenters, optimisers, and attribution models, exists because humans needed tools. Tomorrow’s stack may collapse into systems of agents negotiating with one another, observed but not directed by humans.

It will take time, and the foundations are not yet ready, as Mi3 has noted. Westpac’s Group CTO David Walker flagged recently flagged that the scaffolding for billions of agents does not exist today.  He was. coincidently, speaking at a Forrester conference in Sydney where he told attendees, ““No one’s really figured out how agents work yet, not even the big companies. You can make them work, but no one actually knows how to manage them at scale globally.” 

Per Walker: “I just spent a week with Nvidia, Microsoft, OpenAI, and other companies in Silicon Valley, and even within those organisations, they don’t yet know how to manage multiple agents effectively.”

But the trajectory is one-way. When agents can test, learn and adapt in real time, human-driven campaign cycles will look as quaint as buying banner ads on Yahoo in 1998.

Shadow AI: Rogue buyers return

History rhymes. Two decades ago marketing chiefs bought SaaS licences on their corporate cards, bypassing CIOs who preferred on-premise software. It was a way of circumventing the IT Groupt’s well-earned reputation as the Department of Dr No. The result was Shadow IT. Colony predicts the sequel: Shadow AI.

“CIOs have invested their careers in legacy companies,” he says. “They do not want change”. Business lines will not wait. They will buy AI systems on their own budgets, just as they once bought SaaS. The difference is scale: AI is not a departmental tool. It touches process at the core. Shadow AI may spread faster and carry higher risks than its SaaS predecessor.

For CMOs, that creates both leverage and liability. Leverage, because they can adopt new tools without IT gatekeepers. Liability, because poorly governed AI brings security, compliance and reputational hazards. The path of least resistance may be the most dangerous.

The fall of search and the death of the web

If SaaS is wounded, Google faces something closer to amputation. “The bell has rung for Google. Indexed search is over,” Colony declares. Consumers will not “search” banks or airlines; they will converse with them. Websites become obsolete. The $300bn advertising engine that Google built on links and keywords begins to sputter.

Of course, Colony has been declaring the web dead since the 1990s. But this time, he insists, it is true. And the stats bear him out. Take media, as an example. According to Press Gazatte, 46 out of 50 of the top global media websites were down last month. Year-to-date figures are no more encouraging.

“The vast traffic declines across the board included all ten of the biggest sites in the world. The biggest drop among the top-ten sites was CNN, down 33.6 per cent to 471.6 million visits in July. Substack was the only site to see double-digit year-on-year growth in visits and was up 46.9% to 125.2 million in July.”

SimilarWeb data reveals the corrosive impact of generative search. The share of news-related Google searches ending in “zero-click” rose from 56 per cent in mid‑2024 to 69 per cent by May 2025.

And Andrew Casale, CEO of global SSP Index Exchange, recently told Mi3 that some publishers – including many of those far beyond the media sector report traffic collapses of 30–40 per cent. It hits hardest in the longtail. “I think there’s a possibility that we’ll see the web recede, and there will be a smaller number of names and titles that we know, that we can remember, that we can type in, and traffic comes directly to either by way of newsletter or [directly] typed in.”

Per Colony: The web will linger, “like AM radio, still around, but nobody listens”. For marketers, the implications are significant. Search rankings and paid clicks, once the north stars of digital spend, may vanish as metrics of relevance.

Meta, he thinks, will also struggle. Its data trove of posts and pictures gives it a training advantage. But an AI-native social network, unburdened by clunky algorithms, will certainly emerge, he predicts.

Amazon is better placed: its retail media business, cloud infrastructure and purchase history give it both data and distribution. The question is not assets but leadership. Under Jeff Bezos, Colony suggests, the answer would be obvious. Under his successor, he is much less bullish.

The new AI cartels

The next chapter is not just about the death of old monopolies but the rise of new ones. Where SaaS once created a cartel of cloud vendors, the future may bring new monopolists, although on that Colony is sceptical about whether they will emerge from the ranks of the LLMs.  Building a large language model doesn’t guarantee you a business model, he says.

“I’ll make a prediction for you. I don’t think many of these model vendors will make it to the finish line. Just because you’re able to construct a good inference engine doesn’t mean you’ll be able to offer a true enterprise system to a large bank, a large insurance company, or an automotive company.”

Indeed, he predicts half of today’s model firms will be acquired within two years. Of those left, perhaps only half again will matter.

“I think we don’t know the names of most of the big, dominating companies in the Seventh Wave.”

He says Forrester’s analysts are openly bearish on current market leader OpenAI: great models, poor enterprise traction. 

For CMOs, the risk is déjà vu: another round of dependency on platforms whose terms shift with impunity. The marketing world that chafed at Google’s dominance may soon find itself beholden to even less transparent masters.

Trust, turmoil and new risks

The keyword in Colony’s lexicon is trust. Generative systems hallucinate. Agentic systems act, and sometimes they’ll act on those hallucinations. Neither can be accepted blindly. “No agent should ever be built that we cannot trust,” he says.

He invokes the “Reaper test”: would you trust a drone armed with Hellfire missiles to decide whom to kill? The answer, from nearly all executives he asks, is no. The analogy may be extreme, but the principle applies. Agents managing marketing budgets or customer data must be audited as carefully as human staff.

Colony warns that the allure of agents can obscure their risks. Their agency changes the governance equation. “You’re going to have to manage agents as well as you manage people,” he tells Mi3.

“You’re going to have to monitor them, you’re gonna have to give them goals, and you’re gonna have to audit them as they’re moving through their day, as they’re moving through their work.”

The advice is not optional. The industry, he suggests, is already awash with hype and half-truths. “This is not a good time to be naive.”

This is not simply operational. It is societal. When prompted, Colony frets about the lawlessness of Silicon Valley executives who treat humanity as a variable. As exhibit one, Mi3 suggested Peter Thiel, Paypal co-founder and Palantir chairman, who famously struggled to answer a question by a New York Times reporter about whether humanity should endure. Not will, should.

“Uh…yep…well… I don’t know. I would…I would… there are so many questions implicit in this.”

Exhibit two: Tesla, X and xAI owner Elon Musk, who describes empathy as Western civilisation’s greatest weakness (while also frequently bemoaning how mean people are to him on his own platforms). 

Colony is blunt: executives’ character matters. “Look at the CEO. Can they be trusted as a partner? Because remember, all of this is just software. Do we trust the people building it?”

For marketers, the question is not abstract. Trust is the currency of customer experience. Deploying systems that cannot be trusted is a strategic own-goal.

Colony’s Law: Information will double in value in a year

Economists like laws, the current crop of libertarian-infused Tech Bros less so. But still, Moore’s law governed chips; Metcalfe’s law described networks. Colony has coined one of his own. According to Colony’s Law, “With generative AI, information will double in value every year.” That’s a very Colonial intervention by a man whose career is marked by direct and sometimes incendiary analysis. Colony does not just describe trends; he brands them.

“Generative is a revolution in content. It is unleashing the value of information.”

But in an age of glut, Colony’s law is also a paradox, about which he is unapologetic. “Yes, there’s a lot of information being created, but it is more valuable because we’re able to find it very, very quickly and easily.”

In the fifth wave, which saw the rise of the internet, value lay in creation. The web produced oceans of data, but humans still had to navigate them. Google prospered because it helped sort chaos into something more usable. Generative systems do more. They not only retrieve but synthesise, distil and contextualise, turning scattered shards into usable insight.

Colony offers a parable. Stranded recently on an aircraft with his departure delayed for tyre changes, he asked ChatGPT how long the fix would take. Within seconds, he had an answer, “two hours,”  and that turned out to match the pilot’s own assessment. A Google search, by contrast, would have dumped him in a swamp of manuals and forum posts. The value of information rises because its cost of extraction falls.

The analogy with Gutenberg is deliberate. The printing press multiplied words by orders of magnitude, but its true significance lay in making ideas findable, comparable and transmissible. So too with generative AI. The glut of content it produces looks like noise, yet the ability to interrogate and repurpose it means its economic value rises, not falls.

If Colony’s law holds, information is no longer a passive input but a compounding asset. Information, it transpires, wants to be useful.

Every year, as models improve and interfaces simplify, the stock of knowledge becomes more liquid, more tradable, more valuable. For marketers and media executives, the implication is sobering. Content creation will be commoditised; curation, interrogation and orchestration will be the scarce skills.

In Colony’s schema, then, information is not drowning us; it is enriching us, provided, of course, that the agents that navigate it can be trusted. That is the hinge on which his optimism rests.

Jobs lost, or enhanced?

Colony refuses to parrot Silicon Valley’s optimism that jobs will be untouched. Nor does he dip into the current dystopian narrative. Forrester estimates 7 per cent of jobs in Western economies will be displaced. Yet the larger story is substitution, not elimination. On this point at least, he is not contrarian: “You’re not going to lose your job to AI, you’re going to lose it to someone using AI”.

The metaphor he prefers for AI is Iron Man, not Terminator. Workers who put on the AI “suit” will do more, faster and better. Those who do not will fall behind. For marketing leaders, the implication is stark: upskill teams in AI or watch them become obsolete.

The sting in the tail

Colony’s verdict is unsparing. AI will gut SaaS pricing, collapse the stack, and kill traditional search. The open web, once the great commons of digital marketing, he reckons is now on life support. At best, it survives as a niche for the leftovers; at worst, it degenerates into a swamp of misinformation and trolling. He has been warning of the web’s demise for three decades, and in fairness, with social media, walled gardens, and smartphone apps that view proved prescient, certainly for marketers. At last, he believes, it has arrived. “This is the end of the web. It will be around, but as AM radio is today”.

What comes next is not merely conversational AI but a kind of univergence: the collapse of almost all regular decision-making into a universe of machine-led systems. In this universe, agents replace tools, conversations replace search, and outcomes replace seats. The upheaval will be chaotic. It always is. But in time, the turmoil may give way to harmony. Though it’s not a future, in George Colony’s telling, where all of today’s seemingly insurmountable incumbents will survive to enjoy.