The death of SEO – how agentic search changes everything you knew about search

The traditional landscape of Search Engine Optimization (SEO) is undergoing a seismic shift with the advent of agentic search. This new paradigm, powered by autonomous AI agents, is redefining how users interact with information online, rendering many established SEO practices obsolete.


Understanding Agentic Search

Agentic search refers to AI-driven systems capable of performing complex tasks autonomously, such as retrieving information, synthesizing data, and executing actions without human intervention. Unlike traditional search engines that provide a list of links based on keyword matching, agentic search delivers direct, conversational answers tailored to user intent. For instance, Google’s recent introduction of “AI Mode” exemplifies this shift, offering users synthesized responses and follow-up questions instead of standard search results.


The Disruption of Traditional SEO

The rise of agentic search challenges the core principles of SEO:

  • Keyword Optimization: Traditional SEO relies heavily on keyword placement to rank content. Agentic search, however, focuses on understanding user intent and context, diminishing the importance of exact keyword matches.
  • Link Building: Backlinks have been a cornerstone of SEO, signaling content authority. In agentic search, the AI’s ability to evaluate content quality and relevance reduces the reliance on backlinks as a primary ranking factor.
  • Click-Through Rates (CTR): SEO strategies often aim to improve CTR by optimizing meta descriptions and titles. With agentic search providing direct answers, users are less likely to click through to websites, leading to a decline in organic traffic.

Emergence of New Optimization Strategies

In response to these changes, new optimization methodologies are emerging:

  • Answer Engine Optimization (AEO): AEO focuses on structuring content to be easily interpreted and utilized by AI agents, ensuring that information is presented in a way that aligns with how these systems process and deliver answers.
  • Generative Engine Optimization (GEO): GEO involves optimizing content for generative AI systems, aiming to increase the likelihood of being cited or included in AI-generated responses.
  • Artificial Intelligence Optimization (AIO): AIO emphasizes enhancing content structure and clarity to improve its retrievability and interpretation by AI systems, focusing on semantic relevance and contextual authority.

Preparing for the Agentic Future

To adapt to the agentic search era, content creators and marketers should:

  • Focus on User Intent: Develop content that addresses specific user needs and questions, moving beyond keyword-centric approaches.
  • Enhance Content Clarity: Ensure that information is presented clearly and logically, facilitating AI comprehension and retrieval.
  • Implement Structured Data: Utilize schema markup and other structured data techniques to provide context and improve content discoverability by AI agents.
  • Monitor AI Trends: Stay informed about advancements in AI and agentic search technologies to adjust strategies proactively.

Conclusion

Agentic search represents a fundamental transformation in how information is accessed and consumed online. As AI agents become the primary intermediaries between users and content, traditional SEO practices must evolve to meet the demands of this new landscape. Embracing strategies like AEO, GEO, and AIO will be crucial for maintaining visibility and relevance in an increasingly AI-driven digital world.

The Rise of the AI Agent Economy

For over a decade, the phrase “There’s an app for that” encapsulated the convenience and ubiquity of mobile applications. Today, a transformative shift is underway, heralding a new era where “There’s an MCP server for that” becomes the norm. This evolution signifies the transition from the traditional app economy to an emerging AI agent economy, fundamentally altering how we interact with digital services.


Understanding the Model Context Protocol (MCP)

Introduced by Anthropic in late 2024, the Model Context Protocol (MCP) is an open standard designed to facilitate seamless communication between AI models and external tools, data sources, and services. By standardizing these interactions, MCP enables AI agents to perform complex tasks autonomously, accessing and manipulating data across various platforms without the need for bespoke integrations.

Think of MCP as the “USB-C of AI applications” or “the new HTTP”—a universal connector that allows AI agents to interface with diverse systems efficiently. This standardization reduces the complexity of integrating AI into existing workflows, promoting scalability and interoperability.


The Shift from Apps to AI Agents

Traditional applications require users to navigate interfaces, input data, and interpret results manually. In contrast, AI agents equipped with MCP can autonomously perform tasks such as scheduling meetings, managing emails, or analyzing financial data by interacting directly with relevant services.

This paradigm shift is not just theoretical. Companies like Microsoft are actively developing platforms to support AI agents, aiming to transform their ecosystems into “agent factories” where businesses can build and deploy AI agents tailored to their specific needs.


Implications for the App Economy

The rise of AI agents and MCP servers poses significant implications for the traditional app economy:

  • Reduced Need for Standalone Apps: As AI agents can perform tasks across multiple platforms, the necessity for individual apps diminishes.
  • Shift in Monetization Models: Revenue streams may transition from app sales to service-based models, where access to MCP servers or enhanced AI capabilities becomes the primary commodity.
  • Increased Emphasis on Data and Services: The value may shift towards the quality and accessibility of data and services that AI agents can utilize, rather than the apps themselves.

This transition suggests a potential decline in the traditional app economy, giving rise to an AI agent-driven market where services are consumed through intelligent intermediaries.


Challenges and Considerations

While the AI agent economy offers numerous advantages, it also introduces challenges:

  • Security and Privacy: Ensuring secure interactions between AI agents and MCP servers is paramount, especially when handling sensitive data; this is something vendors are actively working on.
  • Standardization and Compatibility: Widespread adoption of MCP requires consensus on standards to ensure compatibility across diverse systems and platforms.
  • Ethical Considerations: The delegation of tasks to AI agents raises questions about accountability, decision-making, and the potential for unintended consequences.

Addressing these challenges is crucial to realizing the full potential of the AI agent economy.


Conclusion

The evolution from “There’s an app for that” to “There’s an MCP server for that” marks a significant transformation in our digital landscape. As AI agents become more capable and integrated through protocols like MCP, we are entering an era where intelligent systems handle tasks on our behalf, streamlining processes and redefining user experiences.

This shift presents both opportunities and challenges, necessitating thoughtful consideration and proactive measures to harness the benefits while mitigating risks. The AI agent economy is not just a futuristic concept—it is an emerging reality reshaping how we interact with technology.

In an Age of Grifts, I Only Care About Tech Built with Heart

In today’s tech world, it’s easy to be dazzled by trends, venture capital headlines, and the next AI-powered shortcut to easy money. Hype cycles move faster than ever. Yesterday’s breakthrough is today’s boilerplate. But in the middle of this rapid-fire innovation — or, let’s be honest, imitation — one principle keeps me grounded:

I only care about tech people who build with heart.

That doesn’t mean sentimentality. It means intention. It means designing with empathy, coding with conviction, and solving problems that matter, not just those that trend. Heart shows up when a developer chooses ethical boundaries over gray-zone growth hacks. Heart shows up when an engineer stays late, not for a KPI, but because someone’s accessibility feature has to work. Heart shows up in open source contributions done for the love of a better ecosystem, not a better résumé.

Too often, the spotlight lands on those who chase virality — the drop-shippers of code, the AI prompt bandits, the startup founders more fluent in pitch decks than their own tech stack. But while flash can get you followers, it’s depth that earns lasting respect.

And depth takes time. It takes heart.

In a world spinning on speed, the real rebellion is to care — care deeply about users, about craft, about consequence. It’s not just about what we build, but why we build it, and how we treat people along the way.

So here’s to the creators, the maintainers, the quiet builders. The ones who still feel something when they deploy. The ones who choose substance over spectacle, and put humanity at the center of their code.

We see you.

And you’re the reason tech still has a soul.

You’re in the Room for a Reason

There’s a moment we all experience at some point in our careers — sitting in a meeting or stepping onto a stage or into a boardroom — when we pause and wonder, “Should I speak up?” or worse, “Do I belong here?”

Let’s set the record straight: you are in the room for a reason.

Whether you earned your seat through expertise, grit, collaboration, curiosity, or leadership — or all of the above — you’re not there by accident. You bring something that others can’t: your perspective, your story, your lived experience.

And here’s the kicker: it’s not enough to simply be present. Presence without participation is a missed opportunity — not just for you, but for everyone else in the room.

Trust Yourself

Too often, people stay quiet because they second-guess their instincts. But most of the time, your gut is rooted in your knowledge, your preparation, and your unique vantage point. Trust it. Confidence doesn’t mean having all the answers; it means believing that your questions, observations, or input are worth voicing.

Speak with Purpose

You don’t need to dominate the room to make an impact. A well-timed insight, a thoughtful question, or even a moment of active listening can shift a conversation. When you contribute, aim to add value — not just airtime. Think: How can I help this team move forward? What friction point can I clarify? What idea can I build on?

Silence Can Be Expensive

Holding back can cost you influence, relationships, and even opportunities. It can signal disengagement or self-doubt, even when you’re thinking deeply and strategically. Don’t let fear of imperfection rob you of your voice. Progress over perfection. Impact over ego.

You Belong — and Others Need You to Show Up

By being in that room, you already passed the test. Now you’re part of the decision-making, the shaping, the building. When you participate with authenticity and courage, you don’t just elevate your career — you make the entire room better.


So the next time you hesitate, remember this:

You’re in the room for a reason.

Because rooms don’t need echoes. They need voices. Let yours be heard.

When Tech Influencers Hallucinate Harder Than the AI

In the discourse around artificial intelligence, one phrase dominates criticism: AI hallucination — when models generate output that is plausible-sounding but factually incorrect. These errors are framed as the Achilles’ heel of AI systems, especially in high-stakes environments like law, medicine, and finance.

But while AI models hallucinating is a real issue, it is not the most dangerous one.

The bigger problem? Tech influencers hallucinating about what AI can actually do.


When Imagination Outpaces Engineering

In a rush to stay relevant and ride the hype wave, too many tech pundits, influencers, and even respected leaders in the space are promoting exaggerated — sometimes outright fictional — capabilities of AI systems. Claims like:

  • “This AI will replace 90% of coders in 5 years.”
  • “You can build a billion-dollar startup with just prompts and ChatGPT.”
  • “AI already understands your feelings better than your therapist.”

These are not just optimistic projections — they are hallucinations in their own right.


Why This Is More Dangerous

  1. Misplaced Trust
    When influencers inflate AI capabilities, non-experts overestimate what these systems can reliably do. This leads to dangerous applications, over-reliance in critical processes, and blind trust where skepticism is due.
  2. Policy Panic
    Exaggerated narratives fuel government fears, prompting knee-jerk regulations or bans based on science fiction, not science. This hinders responsible innovation while ignoring the real harms: data misuse, labor exploitation, and surveillance creep.
  3. Startup Bubble Thinking
    Founders chase AI silver bullets instead of solving real problems. Investments go into flashy demos rather than long-term viability. The result? Burnout, disillusionment, and another dot-com-bubble-like burst.
  4. Ethics Theater
    With the spotlight on far-off existential risks (e.g., “AI might kill us all”), we lose focus on tangible issues today — such as bias in training data, environmental cost of compute, and AI-generated misinformation.

What Can Be Done?

  • Ground Expectations
    Influencers need to take responsibility for the narratives they shape. Not everything needs to be a revolution — sometimes incremental progress is the real headline.
  • Show, Don’t Just Say
    Claims should be backed with demos, datasets, reproducible benchmarks, and peer-reviewed validations. If it sounds magical but lacks evidence, it’s likely marketing, not machine learning.
  • Prioritize Realism Over Virality
    Responsible thought leadership means helping the public understand how AI works and what it can’t do yet. That’s not boring — that’s sustainable trust-building.

Final Thought

AI models may hallucinate text, code, or citations. But it’s the human hallucinations — the overpromises, the hype, the seductive sci-fi visions — that may do the most lasting damage. The real alignment problem might not be between humans and machines, but between influencers and reality.

Let’s stop dreaming for AI and start understanding it — so we can shape a future that’s powerful and practical.

Day 2 of apidays NYC: A Deep Dive into FINOS

Day 2 of apidays NYC was all about FINOS for me. I spent the entire day in Room Emery (except for a few quick bathroom breaks), immersed in a full lineup of talks and demos centered on APIs, open source, and finance tech.

We kicked off with Olivier Poupeney, FINOS Field CTO, setting the tone for the day. He promised a packed schedule covering cloud, AI, fake trading apps, communication buses, and more — and he delivered. The speaker lineup was stellar, featuring Nicholas Kolba, Daniel Paes, Tom Healey, Daniel Schwartz, Diego Mastroianni, Xiao-Yang Liu (Yanglet), Leigh Capili, Mia Gougisha, Rob Moffat, Luca Borella, and many more.

We started by exploring the real legend — Legend — presented by Daniel Paes. Alongside it, he introduced the FINOS Common Domain Model (CDM) through an IFRS17 insurance regulation compliance use case. But we needed more CDM — and Tom Healey delivered. He expanded on how CDM intersects with Web3, DLT, and blockchain. (Who had that on their bingo card?)

Daniel Schwartz followed with practical insights on integrating CDM into your API calls — a perfect fit for apidays.

Then came the best session of the day — mine! I presented TraderX, the open source, cloud-native, but entirely fake trading app. I dove into the why’s, how’s, whatnots, and explored its extensibility, flexibility, and — of course — APIs. I had to skip some great talks on FDC3, AI Governance, and Cloud Control to answer all the amazing questions folks had about TraderX at the FINOS booth.

The day continued with Diego Mastroianni’s session on transforming with GenAI. He opened with a powerful question: “Is there anyone here who hasn’t used GenAI yet?” He teased an upcoming FINOS GenAI Orchestration platform — name TBD at a future conference.

My final session of the day was Xiao-Yang Liu’s incredibly technical and detailed walkthrough of AI in finance — from A2A to MCP, from FinGPT to “relearning.” I was so impressed I invited him to share more at an upcoming Zenith call!

apidays NYC Day 1 – Summary

This week I had the pleasure of participating in the apidays NYC conference, and what a first day it had (besides it raining…)!

I spent a good portion of my time at the FINOS booth, connecting with attendees, answering questions, and sharing insights about open source in finance. I also enjoyed a lively booth crawl, exploring the vibrant community of API enthusiasts and innovators.

The day kicked off with opening remarks from apidays CTO Baptiste Parravicini , who reflected on the French roots of the conference and emphasized the importance of engaging with your fellow attendees. He also highlighted apidays’ active efforts to close the gender gap, promote federated events (like Openfinity, AuthCon, Green IO, and more), and expand globally—now spanning 14 locations with 80+ events, 100+ sponsors, 100,000+ attendees, 60,000+ companies, 3,000+ speakers, and over 300,000 community members. A memorable moment: “At apidays, the only place you are allowed to use SOAP is in the RESTrooms.”

Next up was Gartner’s Mark O’Neill , who delivered a keynote on the surprising resurgence of XML—pointing out that modern AIs, including GPT-4.1, tend to handle XML more effectively than JSON. (Highly recommend checking out the GPT-4.1 prompting guide: https://cookbook.openai.com/examples/gpt4-1_prompting_guide.) Mark also introduced the new “Beyoncé Rule”: if you like it, you should put an MCP server around it.

One of the most nostalgic moments was seeing Kin Lane, one of the original apidays presenters, who came equipped with the same binder he used 13 years ago! He took us through the evolution of APIs from Salesforce’s early XML APIs to today, where 83% of internet traffic is API-based. His takeaway: while we say “design-first,” for enterprise-scale governance, it’s often more “code-first.”

I then attended Ryan Day ’s talk on building machine learning APIs using FastAPI and ONNX, and got a glimpse into his new book: https://a.co/d/a2f8rx3. Definitely worth a read for anyone working on ML integrations.

A highlight of the day came from X’s API team, with Evan Dolgow and Christopher S.J. Park showcasing astonishing usage metrics—trillions of API calls and impressions that truly highlight the scale of their platform.

I also caught an insightful session on democratizing open banking titled Stop Calling Community Banks the Long Tail, which ties directly into TraderX and BankerX—topics I’ll be covering during my talk on Day 2.

To close out the day, I joined the Capital One session on API Drift Detection and Data Protection. It sparked ideas on improving versioning and governance across API lifecycles.

TBC with the second day summary.

Churn Starts at the Top: Why Leadership, Not Just CS, Owns Customer Retention

When customers leave, it’s tempting to point fingers—usually toward the Customer / Member Success team. After all, their title suggests they’re responsible for the “success” of the customer relationship. But this mindset oversimplifies a complex reality and risks undermining the real levers of retention. The truth is: customer churn is not a Customer Success problem—it’s a company problem.

The Myth of the Siloed CS Team

Customer Success often gets treated like a damage control department. When churn metrics rise, companies look to CS to plug the leak, offer discounts, or chase down feedback. This reactive approach assumes churn is caused by poor communication or lack of support—but in reality, churn often stems from issues that originate upstream:

  • A misaligned product that doesn’t solve the real customer pain point.
  • Overpromising during sales that leads to unmet expectations.
  • A confusing onboarding experience owned by multiple departments.
  • Lack of continuous value delivery due to stalled innovation.

Blaming CS for churn is like blaming the flight attendant for a plane crash—they may be customer-facing, but they’re not flying the plane.

Churn as a Lagging Indicator

Churn is the final stage in a series of unaddressed problems. By the time a customer churns, they’ve likely:

  • Been frustrated by unmet needs or buggy features.
  • Felt ignored during critical phases of adoption.
  • Tried to self-serve their way through opaque documentation.
  • Lost trust in the brand due to inconsistent experiences.

This journey involves product teams, engineering, marketing, sales, and executive strategy—not just Customer Success. Retention is holistic; it reflects the entire lifecycle of a customer’s experience.

Cross-Functional Ownership Is the Answer

To reduce churn meaningfully, organizations must stop treating CS as a lone department and start embracing company-wide accountability. That includes:

  • Product teams building with empathy and user feedback in mind.
  • Sales teams setting accurate expectations.
  • Marketing targeting the right audience, not just the largest.
  • Executives aligning incentives around long-term value, not just quarterly revenue.

Customer Success can be the voice of the customer—but without buy-in from the rest of the organization, they’re shouting into the wind.

Make Churn Prevention a Strategic Priority

Fixing churn requires more than playbooks and QBRs. It demands a shift in culture where every team is measured not just on acquisition, but on retention. Start by:

  • Including churn metrics in cross-team KPIs.
  • Mapping the full customer journey and identifying friction points.
  • Creating closed feedback loops between CS, product, and engineering.
  • Empowering CS with authority, not just responsibility.

Final Thought

If your company is losing customers, don’t just look at the Customer Success team. Look in the mirror. Churn is a company-wide reflection of how consistently you deliver on your promises. Solving it requires unified action, shared goals, and a collective commitment to putting the customer at the center—not just at the end of the funnel.

Where to Find Me This May

May is packed, and I’m thrilled to be part of some incredible events across open source, AI, quantum, and the future of developer platforms. If you’re around any of these, let’s connect!

May 15Catch me at apidays.global, where I’ll be diving into TraderX, the open-source project redefining financial data transparency and interoperability.

May 19 – Join me in New York City for the Microsoft Build 2025 Watch Party, a day filled with product launches, developer energy, and community vibes. If you’re building on .NET, Azure, or just want to geek out about copilots, this one’s for you.

May 28 – A double feature:

  • At the Global Morgan Stanley Technology Expo, I’ll be presenting on Quantum Technologies in Finance—think Qubits, QKD, and portfolio simulations beyond classical limits.
  • Later that day, I’m speaking at our Tech Talk Series on GenAI in Extended Reality—where AI agents meet spatial computing.

May 30 – Wrapping the month at Morgan Stanley’s internal Microsoft Conference, with none other than David Fowler and Mark Russinovich joining as guests. From high-performance .NET Aspire to Azure-scale innovation, it’s a dream lineup.

If you’re attending any of these, come say hi—or shoot me a message ahead of time!

(and a little June peek – on June 17th, thanks to Finra and the World Economic Forum, I will be part of roundtable focusing on XR in Finance together with some other Guild members!)

Turning Resilience into a Competitive Edge in an Uncertain World

In today’s world of volatile markets, climate shocks, cyber threats, and geopolitical shifts, uncertainty is the only constant. Yet amidst all this unpredictability, one insight shines with clarity: companies that deliberately focus on building enterprise resilience are not just surviving — they’re thriving.

This isn’t about bracing for impact anymore. It’s about making resilience a strategic advantage.

Resilience as a Capability, Not Just a Reaction

Traditionally, resilience was seen as the ability to bounce back after disruption — an after-the-fact response. But leading enterprises are flipping the script. They’re embedding resilience into their very capabilities: their technology infrastructure, their talent strategies, their supplier networks, and their governance models.

Take cloud-native architectures as an example. By adopting scalable, fault-tolerant systems, companies can maintain continuity even in the face of outages or demand spikes. Or consider talent mobility: organizations that invest in cross-training and internal career pathways recover faster from workforce disruptions.

These are capabilities, not contingency plans.

From Insurance Policy to Innovation Engine

When resilience becomes proactive, it shifts from being a safety net to an engine of competitive differentiation.

Resilient companies spot risks sooner. They respond faster. They recover with fewer costs. But more importantly, they adapt and evolve more quickly than the competition. They use data not just to predict failure, but to reimagine possibilities.

This agility attracts customers who value reliability, partners who seek long-term stability, and talent who want to build the future rather than fear it.

Resilience Isn’t Just Risk Management — It’s Strategy

Enterprise resilience doesn’t have to be a defensive posture. It can be a growth posture. When companies build with resilience in mind — modular supply chains, AI-enhanced threat detection, decentralized decision-making — they are not just future-proofing. They are positioning themselves to lead in the future.

It’s a welcome bit of certainty: in an uncertain world, resilience isn’t just about staying afloat. It’s about catching the next wave — and riding it ahead of the pack.

Because those who invest in resilience today, are building the certainty — and the competitive edge — of tomorrow.