As I head to #JPM2026 I'm reminded of the principle that "how you do anything is how you do everything" cuts through the noise like nothing else. Small oversights (skipping rigorous testing, glossing over ethical edges, or slapping jargon on old ideas) aren't isolated; they reveal deeper patterns of negligence that undermine entire innovations. Yet the ecosystem rewards the "what" (the shiny new concept) while the "how" (practical execution, integration, discipline, sustainability) quietly evaporates. This gap fuels failures, erodes trust, and keeps us trapped in endless cycles of overpromise and under-delivery.
The latest symptom? "Context Graphs," crowned AI's trillion-dollar opportunity in a late-2025 Foundation Capital piece. The thesis is seductive: AI agents need a living record of decision traces (the "why" behind actions, exceptions, precedents, cross-system context) to move beyond static knowledge graphs or ontologies into true autonomy. Startups, they argue, can capture this dynamic layer in-path, turning tribal knowledge into queryable precedent that compounds over time. It's framed as the missing link in enterprise data stacks, a new system of record for the AI era.
Jessica Talisman nailed the critique when she called it "wet water": graphs are already about context; ontologies already structure it. The term risks becoming another rhetorical tautology, rebranding decades of work in information architecture, process mining, and formalized knowledge capture. But here's my spin: it's not just that it's old wine in new bottles. It's that the hype skips the how entirely.
Nobody wants to do the boring work. Taxonomists and information architects have preached formalized process knowledge for years, encoding decision logic, handling contradictions, building cultural commitment to ongoing curation. The tools weren't always there for seamless, context-aware applications; natural language processing was clunky. Now, with better LLMs and embeddings, we finally have the means to deliver on those old promises: robust information architecture paired with process engineering. These disciplines are needed more than ever. They are the how. But the buzz around Context Graphs pretends this is a brand-new thing, not the disciplined application of timeless principles with upgraded tools.
Shiny objects dazzle in controlled demos. How many apps look like pure magic in promo reels, only to disappoint in the real world? I downloaded Rocket Money after seeing the ads: effortless, one-tap subscription magic. In reality? A tedious quagmire of manually configuring accounts, chasing down hidden charges, and endless verification loops, nothing like the seamless illusion the ads sold.
That's where the HOW fails hardcore: people are led to believe the magic is real and effortless. It's the same as watching Criss Angel float above the Luxor in Vegas: jaw-dropping, seemingly impossible. Then imagine being told, "Okay, now YOU do it." Suddenly you're staring at the harnesses, the years of practice, the invisible rigging, and the sheer discipline required. Tech executives chase the demo glow, ignoring the human grind, governance, and phased rigor needed to make any of it work at scale. The allure hides the truth: real execution is rarely magical; it's disciplined, boring, and unforgiving.
This disconnect starts at the top, often board level. I've heard leaders say outright, "I don't care about the how." Digging into details is seen as slow, while "fast fast fast" is the only virtue that seems to be important. Jeff Bezos talked about the SEAL Team mantra: "Slow is smooth, and smooth is fast." Yet in practice, I've seen operations far from smooth: an insatiable hunger for velocity that pretends people are machines. Many claim to be "humanist," but expect machine-like speed from everyone while glossing over very important details, especially what problem are we actually trying to solve.
This thirst for fast pairs dangerously with the myth that tech companies thrive on generalists disrupting verticals. Big tech sometimes styles itself as a "well-funded startup," but reality bites: true disruption comes from founders who've lived the pain for years or hire deep domain experts as decision-makers. At Amazon/AWS, "working backwards" from the customer was a powerful ritual. But I hate to say that in the last couple of years, AI has let it drift into a mess of AI slop, its an illusion that we can spec and document our way into understanding deep customer pains without ever talking to them. This creates a chasm: between people who speak to customers daily and those who think AI-generated docs shortcut empathy and rigor. In agentic AI, where context capture demands vertical depth, this mindset accelerates the hype-execution gap.
This is not new, the pattern repeats across eras. "Digital transformation" promised modernization but often meant bolting software onto legacy chaos without process overhauls. Volkswagen's Cariad unit chased a unified platform across brands, sidelining phased development and risk governance for ambitious announcements. Cue delays and turnover. Generative AI gets deployed broadly without grasping data quality, ethical training, or integration realities, yielding hallucinations and bias amplification. "Agile" devolves into hybrid messes that neither iterate nor incorporate feedback. Blockchain/NFTs promised decentralization and ownership but ignored scalability, energy costs, and provenance basics, crashing after the euphoria.
Even "no-code" platforms downplay security assessments and data management, fostering shadow IT. Context Graphs fit the same mold: touted as revolutionary for decision traces and semantics, yet often just rebranding knowledge graphs without the hard work of formalized capture, contradiction-handling, and cultural commitment.
Enterprises lack the infrastructure, culture, and discipline for any of this. Leadership incentives reward announcements over execution; fast-culture punishes the slow rigor that builds credible systems. Integrity in small acts (validating claims, prototyping rigorously, documenting processes, talking to real customers) is how you build moats in an age of commoditized models.
Context Graphs aren't inherently bad. The need for better decision lineage is real, and better tools make the old work more feasible. But if we repeat the pattern (hype the "what," vanish the "how"), we'll get another cycle of overpromise, underdelivery, wasted resources, and eroded trust.
In tech, the "how" isn't optional. It defines outcomes. Time to stop chasing buzzwords and start rewarding the discipline that actually delivers.
About the author
Bo Lora is a pragmatic futurist and strategic advisor who seamlessly blends user experience expertise with cutting-edge web technology. With over 30 years of experience at industry leaders like Amazon, Dell, Travelocity, AT&T, BearingPoint (KPMG), Motorola, and USAA, Bo has established himself as a trusted expert in driving innovation and delivering exceptional user experiences across large-scale products and applications.
As a strategic leader, Bo guides cross-functional teams toward cohesive solutions that balance user needs with operational excellence. His vision extends beyond individual projects, shaping industry best practices through his active participation in the LinkedIn community and mentorship of UX designers, researchers, product managers, and engineers.
Bo embraces AI not as a replacement for human ingenuity but as an "exoskeleton" that empowers professionals to unlock productivity levels never seen before. His pragmatic approach seeks to harmonize human interaction with intelligent systems, removing inefficiencies while maintaining a user-centric focus. Now operating independently as a strategic advisor in healthcare and life sciences, Bo focuses on transforming clinical trials through advisory services, thought leadership (including co-authoring a book and preparing a TED talk), and innovative solutions to break down data silos and enhance interoperability. Bo's unique ability to identify opportunities, simplify complex challenges, and foster consensus empowers organizations to stay ahead of the curve, continuously evolving their digital offerings to meet the ever-changing demands of the modern landscape.
#ExecutionOverHype #ContextGraphs #TechLeadership #TheHowMatters #BuzzwordFatigue