Why Domain Expertise Matters in Technology

The deeper I delve into AI, the more clearly I see that the relative values of different skillsets are being rebalanced. This shift has particularly large implications for career transitioners and students entering the data field... Recently, I posted about how the ability to build emotional ties and trust with stakeholders will be the most critical skill of the AI-era: https://lnkd.in/efMV6tdi Similarly, I believe the value of #domainexpertise (DE) will continue to grow, as the value of "technical stack" skills declines (as #AI increasingly assumes those duties). THE COMPONENTS OF DOMAIN EXPERTISE 🔸Factual Knowledge - the terminology, definitions, and data relevant to a domain 🔸Conceptual Knowledge - the theories, models, and structures that explain how things work within the domain 🔸Procedural Knowledge - how to perform domain-specific tasks, techniques and processes 🔸Strategic (aka Metacognitive) Knowledge - how to apply these components to solve problems and make decisions 🔸Tacit Knowledge - the implicit understanding, skills, insights, intuition, etc necessary for expert performance 🔸Contextual Knowledge - the industry-specific factors, regulatory environment, market dynamics, and cultural factors that define the full context in which the domain operates 🔸 Domain-Specific Data - the data sources and metrics essential for analysis and decision-making 🔸Problem Framing - the questions and factors to consider when tackling domain-specific challenges 🔸Interpretation - the ability to translate domain analyses into actionable insights 🔸 Continuous Learning - the discipline and adaptability to keep pace w/ new domain developments, trends, and best practices WHY IS DOMAIN EXPERTISE SO CRITICAL TO AI? There are two primary ways to improve AI models - improve the underlying models themselves or train them on better data. It is DE that generates this higher quality training data. I've been working since GPT4O was released on a custom Power BI GPT that is vastly outperforming both the base 4O model, and every GPT in the GPT store that I've tested it against. This is because mine captures IMO the top 15 books related to Power BI (6,000+ total pages), as well as datasets and data models, courses, articles/blogs, video and audio transcripts, images, thousands of code solutions, etc. - fully leveraging years of experience as a CCO and trainer in this domain. In the ultra-competive business world, where every org will have access to the same base models, the advantages afforded by a superior model trained on better data will be enormous, and those who have the DE to provide that edge - in health care, finance, law, construction, logistics, IT security, public policy, etc. will be in extraordinary demand. This is why IMO #career transitioners with DE from a different sector are entering at a perfect time, and why students should orient their studies to obtaning data skills in the context of building DE in a second area.

Everyone says prompting will be worthless in 2 years. While they wait for AI to "get smart enough" to not need prompts, companies encoding domain expertise are building $100M moats that compound daily. Here's why prompting + human expertise is the most defensible IP you're not building: 1. "AI will understand context eventually" — Sure, and your competitors will eat your lunch while you wait GPT-6 won't magically know why your enterprise deals die at stage 3. Or which contract clauses trigger 90-day payment delays. Or why your ICP ghosts after the CFO meeting. Your experts know. Their prompts encode it. That's unreplicable. 2. The data doesn't lie: Domain expertise beats technical perfection Stanford HAI's 2023 study proved it: Domain expert prompts outperform engineer-written ones by 29% on accuracy, 42% on relevance. McKinsey's 2023 report confirmed: "expertise in the problem domain is often a better predictor of AI effectiveness than technical depth alone." 3. "Prompting is just syntax" — Tell that to the 65% failure rate Cognilytica's 2023 survey found 65% of enterprise AI failures stem from domain misalignment, not bad models. A 10/10 prompt engineer writes beautiful JSON that misses the point. A 7/10 prompter with 10/10 domain expertise. hits the mark every time. 4. Your prompt library is your company's collective brain, weaponized One prompt: Your revenue strategist's "tactics, messaging archetypes, and objection-handling insights no AI engineer could infer." Another: Your legal ops leader's "risk tolerance logic and redline priorities." Not instructions. Encoded wisdom. 5. They copy features. They can't copy decades of experience Prompts aren't commands—they're "encoded representations of mental models, decision criteria, heuristics, and playbooks reflecting years, if not decades, of lived experience." Your competitor can steal your UI, they can steal a prompt and reverse engineer it, but they can't steal how your best people think and build more. 6. "Everyone will have AI" — Exactly. Human Excellence becomes MORE valuable, not less. When AI makes "OK" results universal, domain-specific mastery becomes the only differentiator. "The differentiation will shift from prompt form to prompt intent, and intent will always be a human product." Generic competence is commoditized. Expertise is amplified. 7. The Ironman principle proves it: Human + AI > Either alone "It is not Jarvis (AI) nor Tony Stark (human) alone that yields outsized results. It is the integration. AI without the right human insight is blind; the human without AI is slow. Together, they produce superhuman output." Right now, someone at your competitor is dismissing prompting as "not unique." Right now, smart companies are encoding decades of expertise into strategic IP. In 18 months, one will be desperately playing catch-up. The moat isn't the model. It's the human. And minds, encoded at scale, are unstoppable.

In a world where the cost of execution (coding, design, etc) is quickly dropping towards zero - what is only going to increase in value is a founder's strong expertise in a particular niche. There is a record number of startups in the world today. For literally any idea, you’ll find tens of contenders pitching mostly the same narratives. You have hundreds of AI SDRs and AI ad managers, and thousands of AI design and marketing tools targeting the same very broad audiences. So how do you break through? What really matters - and what’s only going to help you stand out even more in today’s world - is your obsession with a particular niche or a very narrow customer segment. This is what AI can’t do right now (or at least, not yet). It’s your depth of understanding that becomes the moat. When you’ve spent years in a specific industry or solving a specific problem, you start to see the invisible things: the language your customers actually use, the friction points no one talks about, the workflows that look simple on the surface but are filled with edge cases. AI tools can replicate execution, but they can’t replicate insight. They don’t have intuition built from hundreds of conversations, failed experiments, or deep pattern recognition from years in the trenches. That’s why domain obsession isn’t just a “nice to have” anymore — it’s your core strategic advantage. In a sea of generic AI-powered startups, the founders who truly know their people - and build relentlessly for them - are the ones who will win. And ironically, the more AI commoditizes execution - the more valuable your very human, very hard-earned expertise becomes 🙌

Probably one of the best papers written about the impact of AI on product development, scientific discovery, engineers and scientists to date. 🔁 The paper highlights the dual nature of AI’s impact—boosting overall innovation while introducing challenges related to skill utilization and work satisfaction. 🦾 Increased Productivity: AI-assisted researchers discovered 44% more materials, leading to a 39% increase in patent filings and a 17% rise in new product prototypes. These AI-generated materials showed enhanced novelty and contributed to significant innovations. 🧑🏫 Disparate Impacts: The tool disproportionately benefited the most skilled scientists, doubling their productivity while having minimal impact on lower-performing peers. This exacerbated performance inequality, showcasing the complementarity between AI and human expertise. 🤖 Shift in Research Tasks: AI automated 57% of idea-generation tasks, allowing scientists to focus more on evaluating and testing AI-suggested materials. Top researchers effectively leveraged their expertise to prioritize the best AI outputs, while others struggled with false positives. 😞 Impact on Job Satisfaction: Despite productivity gains, 82% of scientists reported lower job satisfaction, citing reduced creativity and underutilized skills as significant concerns. This underscores the complexity of integrating AI into scientific work. 🚀 Broader Implications: The study's findings imply that AI can significantly accelerate R&D in sectors like materials science, emphasizing the value of human judgment in the AI-assisted research process. It suggests that domain knowledge remains crucial for maximizing AI’s potential.

DeepSeek R1's performance and cost numbers have the industry's attention, but its approach to model building might be the bigger breakthrough. Yes, it's a gift to app developers who get powerful new open-source models to build on. And yes, major labs will use these efficiency innovations to build even bigger models. But there's a deeper story here: DeepSeek shows how domain expertise might matter more than raw compute in building specialized AI models. Their approach - using verifiable data and efficient reward functions - creates a new path for teams with deep domain knowledge to build their own models at a fraction of the usual cost. Consider that DeepSeek emerged from a hedge fund, where the reward function is crystal clear. This hints at how other domain experts might leverage similar techniques - whether in finance, medicine, law, or industrial operations. We're entering an era where smart training may matter more than raw compute. The future of AI may not just belong to those with the biggest GPU clusters, but to those who can most effectively combine domain expertise with clever training techniques. I explore this shift in depth here: https://lnkd.in/gKtcx6X8

🎯 Finding the Right Fit About eight years ago, I remember a colleague who was recruited into a project simply because he had a specific programming skill that no one else in the organization possessed. However, the project ignored their domain expertise, reducing his role to just technical execution. Instead of accepting that narrow function, they followed the connection between their domain knowledge and ML, eventually building a unique program that fully leveraged their interdisciplinary strengths. Today, that program stands out as one of a kind. This story came to mind during a recent mentoring conversation. One of my mentees asked why they were struggling to hear back from industry roles despite strong technical skills. Through our discussion, I arrived at three key insights: 1️⃣ The value is in the combination – It is possible to market oneself purely based on programming or ML skills, but much like a gold diadem, you are worth far more than just the material itself. A diadem can be melted down and sold as raw gold, but its full value comes from its craftsmanship. Similarly, your strength lies in the integration of ML, programming, and domain expertise—not just in any single skill. Finding a role that recognizes this unique combination requires a more strategic approach but ultimately leads to greater impact and fulfillment. 2️⃣ Focus on vision, not just past experience – Too often, we define ourselves only by what we have done, rather than what have we done in pursuance of the certain North Star. The key is framing your experience in the context of your future vision—showing how each skill, project, and challenge has prepared you for what comes next. Employers are not just looking for a list of past accomplishments; they want to see how you think about the future and how your expertise will contribute to it. 3️⃣ Yes, it takes longer to find the right fit—but it’s worth it – If you bring an integrated skill set, it is naturally more difficult to find a role that fully values all of your strengths. Many hiring processes are optimized for specialists rather than interdisciplinary thinkers. But when you do find the right match, it will be a far more fulfilling role—one that allows you to make a unique impact. It may take longer, but the result is a position where your full expertise is not just utilized but essential. Curious to hear others' thoughts—have you ever faced a similar challenge in aligning your interdisciplinary skill set with career opportunities? 🚀

1b+ people just got a “Deploy” button. Lovable, v0 by Vercel, and the LLM dev stack collapsed 'idea to an app' into a single prompt. Historic. These 1b+ people, are they “developers”? Yes and no. If “developer” means “can get an app into the world,” the answer is increasingly yes. If it means “can reason about what the AI produced and why,” that’s where the definition bites. Knowing why password storage must be Argon2id/bcrypt with a per-user 16–32-byte random salt, a server-side pepper in KMS/HSM, memory/time costs tuned (e.g., m ≥ 64 MB, t ≥ 3), and constant-time compares isn’t trivia—it can be the line between a cool demo and credential-stuffing, which might lead to account takeover and drained balances. If the scaffold uses raw SHA-256 or a global salt, you’ve built a rainbow-table buffet. Now, why do you need to know what that means? Why does anyone need to going forward? Aren't we just going to "AI it"? The new pattern I keep seeing: people expect the path from idea to production to be instant. That’s human. The Internet trained us to want the right answer, right now and for free. But if you can’t crack open the GitHub repo your AI tool generated and judge the contents, you’ve built a black box you don’t understand. The fix isn’t gatekeeping; it’s fundamentals. Not just software basics, but domain basics: particle physics, derivatives and hedging, healthcare workflow, whatever world your product lives in. You need to be able to evaluate the outcome and how it got there. Software understanding and domain expertise. This is why experience compounds. People who’ve shipped products for a decade bring scar tissue and systems thinking. That perspective prevents some mistakes, and just as importantly, gets the idea to market in the right shape. I’m so bullish on Lovable and similar class tools and on flattening the pyramid. Accessibility is good. But execution-hungry orgs still run on fundamentals plus judgment. I'm seeing brilliant software developers and product experts get incredible demand for their expertise. I'm also seeing folks reinvent themselves in ways that are remarkable. Paradigm shifts present such opportunities at all levels. We didn’t replace software engineering. We widened access and at the same time, raised the bar for quality products. Dan Jouko Anton Patricia Tomi Lourdes Sandeep Stephane Jeremy Joe Hanna Dr. Michael Ciprian Andy Markku Ted Valto Aki Paul PS. Here's a fun view of San Francisco (almost) and the GG bridge. Probably not one for the tourism posters however :D

75% of B2B buyers now expect sellers to know their industry. Buyers are sharper. They’re coming in with research already done. They’ve talked to peers. They’ve asked ChatGPT. Your rep has about five minutes to prove they belong in the room. And if they don’t speak the language—technical, vertical, or strategic—they’re out. This is especially brutal in complex markets: cyber, data infrastructure, fintech. If your team can’t go toe-to-toe with a CISO or a CIO, the deal is dead on arrival. Too many orgs rely on “smart sellers who’ll figure it out.” That’s not enough anymore. Here’s what the best sales orgs are doing: - Hiring sellers with real domain expertise - Investing early in deep product and buyer training - Pairing reps with experts when needed—but pushing for fluency fast Today, every buyer is a power user. Every decision-maker is busy. If your rep doesn’t show up with context, insight, and credibility—there’s no second call. Domain expertise isn’t a nice-to-have. It’s the price of entry.

As API costs plummet and AI capabilities expand, we're witnessing a shift in the software landscape. The competitive advantage is moving from pure technical expertise to deep domain knowledge. Subject matter experts who intimately understand their industry's challenges can now build and deploy solutions that once required large engineering teams. We’re moving from grow at all costs to grow at small costs. This democratization of software development means specialists in energy, construction, manufacturing, or any complex field can directly translate their expertise into scalable solutions. They understand the nuanced relationships between stakeholders, the hidden friction in processes, and the real pain points that need solving. I see a lot of focus on vertical SaaS, but it’s more than that. Think of it as "expertise as a service"—where industry veterans can finally bridge the gap between their deep domain knowledge and modern technical capabilities. The barriers to entry aren't just lowering; they're fundamentally changing. The next wave of innovation will emerge from experts who've spent years in the trenches of their industries, now empowered with tools to build faster than ever before. #ai #energytransition #saas

I just wrapped a community session for my friends at the AI Exchange, sharing how I use voice to power AI automations - it was a blast! This group never fails to amaze me, and I always love connecting with folks in our group. During the session, I shared a behind-the-scenes look at a LinkedIn automation I built. I demonstrated how I talk to a GPT I created in ChatGPT, it sends the transcript to Zapier, and through a series of steps, it crafts a LinkedIn post from that transcript and emails me 3 drafts (one written by GPT-4o, one written by Claude, and one written collaboratively by both of them). To demo the automation live, I asked if anyone had an idea for a LinkedIn post. Kevin Hardin, one of the community members, jumped in and spoke directly to my GPT to share his thoughts. Below is what we got back from the collaborative run (with only 90 seconds of edits - mostly so I could make it fit with this intro also) - shared with Kevin's permission: ----- AI isn't replacing us; it's enhancing what we can do. But to truly take advantage of its power, you need to know your field. Think about it:  • Domain expertise + AI = Faster, better results in your field  • No expertise + AI = Potential disaster I call this the "AI Illusion" — when you think AI has given you a shortcut in an area you're unfamiliar with, but you lack the expertise to judge the quality of the output. For instance, if I used AI to draft a patent or grant, I might think, "This looks great!" But if I showed it to a patent expert or grant writer, they'd probably spot major flaws I missed because I don't have the expertise to evaluate it properly. The takeaway? AI is extremely powerful, but it's not a substitute for real knowledge. You need enough domain expertise to recognize quality output and use AI effectively. Many people worry AI will replace them, but the more we use it, the clearer it becomes: AI amplifies existing skills. If you have domain expertise, AI helps you do what you already know—just faster and better. The problem arises when you use AI in areas where you lack expertise. This is where the "AI Illusion" kicks in. You might think the output looks good, but without the right knowledge, you can't tell good from bad. The bottom line is that AI works best as a tool to enhance what you already know, not as a replacement for expertise. You need a solid foundation in your field to evaluate and use AI-generated results effectively. What do you think? Have you experienced the "AI Illusion" in your work? I'd love to hear your thoughts!