🚨OpenAI admits that prompt injection attacks against its Atlas AI browser "may never be fully solved," acknowledging agentic AI systems have a fundamental, unfixable security vulnerability where malicious instructions hidden in web pages or emails can hijack agent behavior—creating a new class of technical debt where security tools must become as complex as the attacks themselves.

Making money from your content shouldn’t be complicated. With Google AdSense, it isn’t.

Automatic ad placement and optimization ensure the highest-paying, most relevant ads appear on your site. And it literally takes just seconds to set up.

That’s why WikiHow, the world’s most popular how-to site, keeps it simple with Google AdSense: “All you do is drop a little code on your website and Google AdSense immediately starts working.”

The TL;DR? You focus on creating. Google AdSense handles the rest.

Start earning the easy way with AdSense.

Earn with Google AdSense

Most consulting engagements follow a familiar pattern: diagnose the problem, create a 60-page deck, recommend some process changes, collect the check, and leave.

Six months later, nothing's changed. The processes revert. The insights gather dust. And the client needs another engagement.

But a new model is emerging—one where agencies don't just advise, they install capabilities that keep working long after the consultants leave.

The shift:

Instead of delivering recommendations, agencies are building custom AI agents and skills that handle the actual work. They map how tasks flow through an organization, identify repetitive coordination work, and deploy systems that do it automatically.

Think about what happens inside most businesses: someone needs to chase down a report, another person consolidates data from three systems, someone else sends follow-up emails, and a coordinator makes sure nothing falls through the cracks.

That coordination layer—the emails, the check-ins, the status updates, the gentle nudges—often spans multiple roles. Combined salary cost? Easily $250K–$400K annually.

How it works:

An agency comes in for a few weeks (not months). They don't interview stakeholders for the sake of interviewing. They map the actual workflow:

- What reports get generated?

- What data gets pulled from where?

- Who needs to be notified when?

- What checks happen before something moves forward?

- What follow-ups happen when things stall?

Then they build it. Claude Skills that know your reporting templates. Agents that monitor systems and trigger actions. Coordination logic that used to live in someone's head, now running automatically.

The company pays once for the setup. They own the system. The agency leaves.

And here's the key: it keeps working. The skills don't quit. The agents don't take vacation. The coordination doesn't drift back to old habits.

The agency only returns when something needs tuning—a new workflow, a changed process, an expanded scope.

Why this matters now:

Traditional consulting has always had a retention problem disguised as a business model. Clients needed repeat engagements because advice without execution doesn't stick.

But AI agents and skills are installed capabilities, not recommendations. They're more like infrastructure than consulting.

This changes the economics for both sides:

For agencies: Instead of selling recurring strategy work, they're selling durable systems. The engagement is shorter, but the value delivered is measurable and permanent. And when businesses see one workflow automated successfully, they want more. It's a different kind of recurring revenue—based on expansion, not re-diagnosis.

For companies: They're not renting expertise anymore. They're buying capability that compounds. The first automation saves $300K/year. The second saves another $200K. By the third, they're wondering what else can be systematized.

The agency doesn't disappear:

Human judgment still matters. Strategic decisions, change management, edge cases, organizational politics—these aren't automatable (yet).

But here's what's shifting: agencies used to charge for thinking about the work. Now they're increasingly charging for installing systems that do the work.

The PowerPoint deck becomes the implementation. The recommendation becomes the agent. The org chart suggestion becomes the skill library.

What's already happening:

Forward-thinking agencies are already making this shift. They're hiring AI engineers alongside strategy consultants. They're building proprietary skill frameworks that can be customized per client. They're positioning around "capability installation" rather than "advisory services."

Some are even switching to success-based pricing: you pay when the automation is working, not just when we deliver it.

The clients who get this early have an advantage. While competitors are still doing quarterly strategic reviews, they're running automated coordination systems that scale without headcount.

What's next

This model will likely bifurcate consulting into two tiers:

Tier 1: High-touch, high-judgment strategic work that requires deep human expertise. Think M&A advisory, organizational transformation, executive coaching. Still expensive, still relationship-driven.

Tier 2: Capability installation. Process mapping → agent building → deployment → occasional tuning. Shorter engagements, measurable ROI, scalable delivery.

Most mid-market consulting work will shift to Tier 2. Not because it's "worse"—but because clients will prefer systems that keep working over advice they have to implement themselves.

The agencies that figure this out first will eat the ones still selling strategy decks.

BTW: The consulting industry generates about $300 billion annually in the U.S. alone. If even 20% of that shifts toward AI capability installation, we're looking at a $60 billion market for building custom business agents. That's roughly the size of the entire enterprise software market in 2010. The firms that dismissed "implementation work" as beneath them might want to reconsider.

Have an ad concept ready but don't want to deal with expensive shoots or stock footage? ScriptKit lets you generate, curate, and edit AI ads in one platform.

What ScriptKit gives you

Give ScriptKit a shot — go from concept to finished AI ad without wrangling teams or gear.

The Recap: Alphabet announced it will acquire Intersect Power for $4.75 billion in cash plus debt, buying a data center and clean energy developer to expand power generation capacity without relying on local utilities struggling to keep up with AI demand. The acquisition includes multiple gigawatts of energy and data center projects representing 10.8 gigawatts by 2028—more than 20 times the Hoover Dam's output—signaling that tech giants are now vertically integrating energy infrastructure because grid constraints have become the actual bottleneck to AI scaling, not compute or talent.

Unpacked:

Bottom line: When tech giants start acquiring energy companies at multi-billion dollar valuations, they're signaling that the constraint on AI has shifted from software to physics. Alphabet's acquisition of Intersect represents vertical integration at infrastructure scale—Google no longer trusts utilities, regulators, or market forces to deliver power fast enough, so it's building its own energy company.

The UK's National Cyber Security Centre warned that prompt injection attacks "may never be totally mitigated," advising professionals to reduce risk rather than think attacks can be "stopped"—security researcher notes that agentic browsers sit in "moderate autonomy combined with very high access" to sensitive data, meaning "for most everyday use cases, agentic browsers don't yet deliver enough value to justify their current risk profile."

Trump administration halted 6 GW of offshore wind leases again, creating a new regulatory risk vector where political decisions directly constrain AI infrastructure deployment and capital planning must account for policy volatility affecting renewable energy supply chains.

AI broke the smart home in 2025 by degrading user experience when generative AI was added to existing platforms, suggesting not all products benefit from AI integration and that AI moats are weaker than promised when implementation actually makes things worse.

Uber and Lyft will test Baidu robotaxis in London next year alongside Waymo, signaling that Chinese AI companies now compete directly with U.S. firms in Western markets and geography is no longer a moat for AI-powered services.

Apple fined $116 million by Italy over app privacy prompts, as regulators quantify and penalize the privacy tax that dominant platforms impose—signaling a shift toward forcing tech giants to internalize externalities through enforcement rather than allowing them to externalize costs onto users.

The FCC's foreign drone ban is here, embedding hardware restrictions directly into regulatory frameworks and effectively weaponizing supply chain policy against foreign tech manufacturers as geopolitical competition reshapes what technologies are permissible in domestic markets.

Pirate library ripped 86 million of the most popular songs on Spotify, revealing that at-scale data scraping is the asymmetric cost structure undercutting streaming platforms and network effects are insufficient defense against organized, ideological competitors willing to operate outside legal frameworks.

DatoCMS reached €6.5M revenue with a team of 13 after 10 years bootstrapped, demonstrating that profitable, focused B2B SaaS companies can outperform venture-backed startups on metrics that matter to founders—challenging the default assumption that VC funding is necessary for success.

Best of AI™ Team

Was this email forwarded to you? Sign up here.