[{"data":1,"prerenderedAt":590},["ShallowReactive",2],{"post-\u002Fhow-llms-identify-experts":3},{"page":4,"translations":437,"nav":445,"related":570,"random":581},{"id":5,"title":6,"body":7,"categories":406,"category":409,"changeHistory":409,"date":410,"description":411,"disclosures":412,"draft":415,"extension":416,"firstLiveAt":409,"image":417,"imageAlt":418,"kind":419,"lang":420,"meta":421,"navigation":422,"omitGermanLocalizationDisclosure":415,"path":423,"publishedAt":409,"readingTime":424,"rights":409,"seo":425,"seoTitle":426,"slug":427,"sources":409,"stem":427,"tags":428,"translationKey":427,"type":435,"updated":409,"__hash__":436},"posts\u002Fhow-llms-identify-experts.md","What Makes an LLM Recommend Someone as an Expert?",{"type":8,"value":9,"toc":392},"minimark",[10,14,17,20,23,26,31,34,37,40,43,46,49,53,56,59,62,65,68,72,75,98,101,105,108,129,132,135,139,142,145,148,151,155,158,161,164,167,170,174,177,200,203,206,210,213,216,219,222,225,228,232,235,238,241,244,247,250,254],[11,12,13],"p",{},"Imagine a company looking for an adviser in the emerging field of AI search. An executive asks an LLM, “Who are the leading experts on this subject in Turkey?” The model returns three names, short explanations, and a handful of sources.",[11,15,16],{},"That list may be a useful place to begin. It becomes risky when it is treated as a ranking of expertise without asking what the model actually measured.",[11,18,19],{},"The model has not worked on these people’s projects. It may never have seen how they tested their claims, what failed, or what happened after their recommendations were put into practice. What it can see is mostly the public record: articles, links, citations, conference programmes, podcasts, company pages, academic publications, and search results.",[11,21,22],{},"Sometimes that record represents real expertise remarkably well. Sometimes it tells us mainly who is most visible.",[11,24,25],{},"An LLM recommendation needs to be read with that distinction in mind. Did the system assess expertise, or did it assess the online traces that expertise may leave behind?",[27,28,30],"h2",{"id":29},"what-ai-sees-public-traces-on-the-web","What AI sees: public traces on the web",[11,32,33],{},"Consider a digital marketing agency that turns a new AI search concept into a service category within a few weeks. It publishes a stream of blog posts, glossary entries, webinars, and LinkedIn posts. As it explains the concept, it also names its own frameworks.",[11,35,36],{},"None of this is inherently suspicious. The agency may genuinely have developed early expertise. In a new technology category, expecting twenty years of experience would make little sense. A large body of content may also be the natural result of research and field knowledge being shared openly.",[11,38,39],{},"The same public picture can appear when a few assumptions have not been tested but are rewritten repeatedly. Within months, the agency and its executives occupy a large share of the search results. Other sites begin using the same concepts.",[11,41,42],{},"From the outside, two different realities may look similar. The agency may have become one of the field’s leading experts, or it may have created one of the largest online footprints in the field.",[11,44,45],{},"Expertise is also specific to context. Someone may understand the technical architecture deeply but have limited experience applying it commercially. A method that works for an agency in one sector may not have been tested in another. Product knowledge that was correct last year may already be out of date.",[11,47,48],{},"Content can be evidence of expertise. Content is not expertise itself. The distinction lies less in how much someone publishes than in how the claims were formed, tested, and corrected over time.",[27,50,52],{"id":51},"ai-answers-can-draw-on-information-in-different-ways","AI answers can draw on information in different ways",[11,54,55],{},"An LLM may answer from patterns learned during training, without searching the web. In other cases, it searches the web at the time of the question. Visibility can matter in both situations, but not through the same mechanism.",[11,57,58],{},"Without web search, the model relies on associations it has encountered before. When a person’s name repeatedly appears alongside a particular concept across different documents, that association may be easier to recall. Recall does not mean the model has assessed whether the person tested the claim or produced good results.",[11,60,61],{},"With web search, a different sequence of decisions comes into play. The system must first find pages, then judge which appear relevant and credible, then decide which ones to use in the answer. Choosing which sources to display as citations is a separate decision.",[11,63,64],{},"Search engines and AI products use more than content volume. Relevance, links, freshness, originality, source reputation, and signs of spam can all matter. Reproducing the same material a hundred times is therefore not a dependable shortcut.",[11,66,67],{},"We do not know the exact weight of these factors. The systems do not all work alike, and they change over time. A study of repetition in training data cannot tell us which pages ChatGPT Search will rank today. Google’s organic search signals do not reveal which person–topic associations an LLM learned during training.",[27,69,71],{"id":70},"what-we-know-what-we-can-infer-and-what-remains-speculation","What we know, what we can infer, and what remains speculation",[11,73,74],{},"There is no published formula showing how an AI system concludes that a person is an expert. It helps to separate three levels of evidence.",[76,77,78,86,92],"ul",{},[79,80,81,85],"li",{},[82,83,84],"strong",{},"Known:"," Search and AI search systems assess the pages they can access using factors such as relevance, quality, reputation, originality, and freshness. What a model encounters during training also affects what it may later recall. Appearing as a cited source does not guarantee that the source is correct.",[79,87,88,91],{},[82,89,90],{},"Reasonable inference:"," Independent third-party references, a publication record, inspectable work, and a consistent association between a person and a subject can give the system more public evidence to work with. How much weight any system gives these signals when recommending people has not been disclosed.",[79,93,94,97],{},[82,95,96],{},"SEO\u002FGEO speculation:"," There is no strong general evidence that any single signal—such as the number of LinkedIn posts, conference appearances, author schema, a Wikipedia entry, or social-media followers—directly raises an “AI expertise score.”",[11,99,100],{},"None of these signals is worthless. Links, publications, conference appearances, and a visible record of work can all be traces of genuine expertise. The problem is that the system cannot always see the experience that produced them.",[27,102,104],{"id":103},"how-visibility-can-become-authority","How visibility can become authority",[11,106,107],{},"A recurring pattern appears in fast-moving fields. It does not need to be deliberate.",[109,110,111,114,117,120,123,126],"ol",{},[79,112,113],{},"A new concept emerges while the evidence is still limited.",[79,115,116],{},"Advisers, agencies, and software companies begin explaining it and packaging it as a service.",[79,118,119],{},"Early assumptions are repeated across blogs, social posts, and webinar summaries.",[79,121,122],{},"The connection to the original source weakens, and the same claim begins to look like independent knowledge on different sites.",[79,124,125],{},"Search and AI search systems select sources from this expanding content surface.",[79,127,128],{},"Generated answers return the same narrative to circulation with new wording and new citations.",[11,130,131],{},"The cycle can break at any point. Search engines may suppress copied or low-value pages. Strong primary sources may rank above the rest. A user may open the sources and inspect the evidence. A model may retrieve different material when the question changes.",[11,133,134],{},"For that reason, publishing at scale does not automatically create authority. The narrower concern is that when visibility, repetition, and citation reinforce one another, an untested claim can begin to look more established than it is.",[27,136,138],{"id":137},"twenty-sources-can-trace-back-to-one-claim","Twenty sources can trace back to one claim",[11,140,141],{},"When an AI system finds twenty pages making the same point, it may not be looking at twenty independent observations. Every page could descend from one report, press release, conference presentation, or blog post.",[11,143,144],{},"If the link to the original source disappears during rewriting, the common origin becomes hard to see. AI-generated summaries can repeat the claim in different words. Search results may eventually suggest a broad consensus even though the evidence still comes from the same original source.",[11,146,147],{},"This problem predates generative AI. Research on citation networks has shown how an initial hypothesis can come to be presented as fact through selective citation and unattributed repetition. People, too, tend to find a familiar statement more plausible simply because they have encountered it before.",[11,149,150],{},"“According to multiple sources” therefore deserves scrutiny. A large number of URLs is not the same as a large number of independent observations. Unless an AI system traces sources back to their common origin, it may struggle to distinguish repetition from corroboration.",[27,152,154],{"id":153},"the-economics-of-content-production-have-changed","The economics of content production have changed",[11,156,157],{},"Producing hundreds of blog posts, glossary entries, FAQs, webinar summaries, and social posts once required a substantial editorial budget. Generative AI has sharply reduced at least the cost of producing competent prose.",[11,159,160],{},"That does not make content production a bad practice. Experts can use the same tools to explain their research more clearly, organise their notes, and reach different readers. More high-quality information can be published as a result.",[11,162,163],{},"Tested expertise still requires time with real cases, failed attempts, measurement, and reputational risk. If the cost of producing content that looks like evidence of expertise falls faster than those costs, the economic gap widens.",[11,165,166],{},"Search engines are developing filters for scaled, copied, and low-value content. Those measures make volume alone less useful. They will not be perfect. Distinguishing original-looking but shallow material from genuine first-hand knowledge is not an easy classification problem.",[11,168,169],{},"A more careful conclusion is that some visibility signals are becoming cheaper to produce. The cost of tested expertise may not fall at the same pace. The wider that gap becomes, the more important it is to look beyond content volume.",[27,171,173],{"id":172},"harder-questions-for-evaluating-expertise","Harder questions for evaluating expertise",[11,175,176],{},"Visibility should not be ignored when assessing a person or organisation. Public work is still one of the best forms of evidence we have. But it needs to be examined with harder questions than how much was published.",[76,178,179,182,185,188,191,194,197],{},[79,180,181],{},"Is there original data, a test record, a product outcome, or a real case behind the claims?",[79,183,184],{},"Does the person separate what they know, what they observed, and what they are inferring?",[79,186,187],{},"Is it clear what evidence would prove a claim wrong?",[79,189,190],{},"Do they explain what failed and why the approach changed?",[79,192,193],{},"Do they visibly correct an earlier claim when new evidence arrives?",[79,195,196],{},"Are the references truly independent, or are they derivatives of the same source?",[79,198,199],{},"Is the result specific enough to show the context in which it applies?",[11,201,202],{},"Real experts usually know not only what should be done, but also what did not work and why. Explaining the mechanism of failure usually requires closer contact with the problem than observing it from a distance.",[11,204,205],{},"These criteria should not automatically favour older, established names. A newcomer can develop real expertise quickly in a new field. Seniority is not the deciding factor. What matters is how the claim was tested and how visible the supporting evidence is.",[27,207,209],{"id":208},"what-ai-systems-could-do-better","What AI systems could do better",[11,211,212],{},"A perfect expertise score is unlikely to be realistic. Better controls than URL counts and broad domain reputation are still possible.",[11,214,215],{},"Systems could group sources by their original evidence instead of simply counting URLs. If five domains have rewritten the same press release, they should not be presented as five separate confirmations.",[11,217,218],{},"A claim should, where possible, lead back to the source that first made or directly measured it. A provider’s own documentation can explain how a product is intended to work. Assessing whether it works in practice requires independent research or an inspectable case. Users should be able to see which passage in a source supports each material claim in an AI answer.",[11,220,221],{},"Time and correction history could also be part of the assessment. How did earlier predictions turn out? Were errors corrected? Did the person continue making the same claim after the evidence changed?",[11,223,224],{},"There is a trade-off. If past reputation and citation counts carry too much weight, already visible people become even more visible. New or underrepresented experts may be pushed aside because their online footprint is not yet large enough.",[11,226,227],{},"A better system would show the origin, independence, and limits of the evidence instead of compressing everything into one authority score. When it is uncertain, it should be able to say so plainly: this person is highly visible, but independent evidence of their expertise is limited.",[27,229,231],{"id":230},"how-to-read-an-ai-recommendation","How to read an AI recommendation",[11,233,234],{},"The person recommended by an LLM may be an excellent expert. The model may also have found exactly the right evidence. Visibility and expertise are not opposites. Good work often develops a public record naturally.",[11,236,237],{},"Even so, an AI system does not measure expertise itself. It draws on associations learned earlier, pages it can access, sources selected by ranking systems, and citations chosen for the answer. Each can be a useful proxy for expertise. Each also has its own failure modes.",[11,239,240],{},"An AI recommendation is therefore better read as a discovery signal than as a decision. The next step is to open the sources, trace their common origins, look for independent evidence, and examine past performance.",[11,242,243],{},"Two questions remain.",[11,245,246],{},"When an AI recommends someone as an expert, has it measured their expertise or the size of the trace they have left online?",[11,248,249],{},"And if creating a sufficiently large appearance of expertise becomes cheaper than developing real expertise, how will that distinction be preserved?",[27,251,253],{"id":252},"further-reading","Further reading",[76,255,256,270,279,288,297,306,315,324,333,342,351,360,383],{},[79,257,258,269],{},[259,260,268],"a",{"className":261,"href":263,"rel":264,"target":267},[262],"dofollow","https:\u002F\u002Facademic.oup.com\u002Fct\u002Farticle\u002F35\u002F1\u002F37\u002F7876430",[265,266],"nofollow","noopener","_blank","Epistemic authority in the digital public sphere",": Van der Linden and colleagues distinguish between possessing knowledge, being perceived as an authority, and presenting oneself as one.",[79,271,272,278],{},[259,273,277],{"className":274,"href":275,"rel":276,"target":267},[262],"https:\u002F\u002Fproceedings.mlr.press\u002Fv202\u002Fkandpal23a.html",[265,266],"Large Language Models Struggle to Learn Long-Tail Knowledge",": Kandpal and colleagues show that models recall information more readily when it has a broader presence in training data. The study does not measure expert recommendations.",[79,280,281,287],{},[259,282,286],{"className":283,"href":284,"rel":285,"target":267},[262],"https:\u002F\u002Faclanthology.org\u002F2022.acl-long.577\u002F",[265,266],"Deduplicating Training Data Makes Language Models Better",": Lee and colleagues show that removing duplicate training data can reduce the generation of memorised text.",[79,289,290,296],{},[259,291,295],{"className":292,"href":293,"rel":294,"target":267},[262],"https:\u002F\u002Faclanthology.org\u002F2026.findings-acl.526\u002F",[265,266],"Characterizing Web Search in the Age of Generative AI",": Kirsten and colleagues find that generative search systems differ in how they use sources and how stable their results are.",[79,298,299,305],{},[259,300,304],{"className":301,"href":302,"rel":303,"target":267},[262],"https:\u002F\u002Faclanthology.org\u002F2023.findings-emnlp.467\u002F",[265,266],"Evaluating Verifiability in Generative Search Engines",": Liu, Zhang, and Liang examine why displaying citations does not mean every claim is supported.",[79,307,308,314],{},[259,309,313],{"className":310,"href":311,"rel":312,"target":267},[262],"https:\u002F\u002Faclanthology.org\u002F2026.eacl-long.115\u002F",[265,266],"Assessing Web Search Credibility and Response Groundedness in Chat Assistants",": Vykopal and colleagues show that AI assistants can differ considerably in the credibility of their sources.",[79,316,317,323],{},[259,318,322],{"className":319,"href":320,"rel":321,"target":267},[262],"https:\u002F\u002Fwww.bmj.com\u002Fcontent\u002F339\u002Fbmj.b2680.abstract",[265,266],"How citation distortions create unfounded authority",": Greenberg traces how one claim can acquire the appearance of strong consensus through citation and repetition.",[79,325,326,332],{},[259,327,331],{"className":328,"href":329,"rel":330,"target":267},[262],"https:\u002F\u002Fpubmed.ncbi.nlm.nih.gov\u002F20023210\u002F",[265,266],"The truth about the truth",": Dechêne and colleagues review how repetition can influence people’s judgments of truth.",[79,334,335,341],{},[259,336,340],{"className":337,"href":338,"rel":339,"target":267},[262],"https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41586-023-06883-y",[265,266],"Online searches to evaluate misinformation can increase its perceived veracity",": Aslett and colleagues show that encountering low-quality supporting results can increase belief in some false claims.",[79,343,344,350],{},[259,345,349],{"className":346,"href":347,"rel":348,"target":267},[262],"https:\u002F\u002Fdoi.org\u002F10.1126\u002Fscience.adh2586",[265,266],"Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence",": Noy and Zhang measure reduced completion time and higher output quality in specific professional writing tasks.",[79,352,353,359],{},[259,354,358],{"className":355,"href":356,"rel":357,"target":267},[262],"https:\u002F\u002Facademic.oup.com\u002Fbook\u002F7809\u002Fchapter-abstract\u002F152991558",[265,266],"Experts: Which Ones Should You Trust?",": Alvin Goldman considers argument quality, independent agreement, interests, and track record as evidence for evaluating experts.",[79,361,362,368,369,375,376,382],{},[259,363,367],{"className":364,"href":365,"rel":366,"target":267},[262],"https:\u002F\u002Fdevelopers.google.com\u002Fsearch\u002Fdocs\u002Fappearance\u002Fai-features",[265,266],"Google’s explanation of AI features",", ",[259,370,374],{"className":371,"href":372,"rel":373,"target":267},[262],"https:\u002F\u002Fsupport.microsoft.com\u002Fen-us\u002Fbing\u002Fhow-bing-delivers-search-results",[265,266],"how Bing delivers search results",", and ",[259,377,381],{"className":378,"href":379,"rel":380,"target":267},[262],"https:\u002F\u002Fhelp.openai.com\u002Fen\u002Farticles\u002F9237897-chatgpt-search",[265,266],"the ChatGPT Search overview",": Provider documentation is a useful starting point for separating disclosed mechanisms from industry speculation.",[79,384,385,391],{},[259,386,390],{"className":387,"href":388,"rel":389,"target":267},[262],"https:\u002F\u002Fdevelopers.google.com\u002Fsearch\u002Fdocs\u002Fessentials\u002Fspam-policies#scaled-content",[265,266],"Google’s policy on scaled content abuse",": Google explains why producing many pages without adding user value can be treated as spam regardless of the tools used.",{"title":393,"searchDepth":394,"depth":394,"links":395},"",2,[396,397,398,399,400,401,402,403,404,405],{"id":29,"depth":394,"text":30},{"id":51,"depth":394,"text":52},{"id":70,"depth":394,"text":71},{"id":103,"depth":394,"text":104},{"id":137,"depth":394,"text":138},{"id":153,"depth":394,"text":154},{"id":172,"depth":394,"text":173},{"id":208,"depth":394,"text":209},{"id":230,"depth":394,"text":231},{"id":252,"depth":394,"text":253},[407,408],"ai","business",null,"2026-09-04","LLMs and AI search systems do not measure expertise directly. They infer it from visible online evidence. This article examines the gap between visibility, repetition, and tested expertise.",{"aiUse":413,"aiNote":414},"ai-assisted","The question, scope, and editorial direction of this article were defined by Evren Bal. AI-assisted tools supported academic and technical source research, counterargument review, draft development, source classification, and English adaptation.",false,"md","\u002Fimages\u002Fhero\u002Fllm-expertise-public-signals.avif","A blue inspection frame surrounds a profile and public content signals, leaving test records, a broken prototype, and a performance gauge outside the frame.","Analysis","en",{},true,"\u002Fhow-llms-identify-experts",12,{"title":6,"description":411},"Can LLMs Distinguish Expertise From Online Visibility?","how-llms-identify-experts",[429,430,431,432,433,434],"llm","ai-search","seo","geo","epistemic-authority","content-strategy","post","ZJWpjwsJOnqYvMftoFl8aD7MsyfiDYHSE0sx0-MZXsc",{"en":438,"tr":439,"de":442},{"path":423,"title":6},{"path":440,"title":441},"\u002Ftr\u002Fllm-bir-uzmani-nasil-tanir","LLM bir konunun uzmanını neye göre önerir?",{"path":443,"title":444},"\u002Fde\u002Fwen-empfiehlt-ein-llm-als-experten","Wen empfiehlt ein LLM als Experten – und warum?",{"prev":446,"next":449,"others":452,"lucky":569,"readingTime":424},{"path":447,"title":448},"\u002Fhow-can-companies-create-value-from-ai","How Can Companies Create Value From AI?",{"path":450,"title":451},"\u002Fopen-publishing-ai-monthly-seo-reporting-experiment","Open Publishing and an AI-Assisted Monthly SEO Reporting Experiment",[453,456,459,462,465,468,471,474,477,480,483,486,487,490,493,496,499,502,505,506,509,512,515,518,521,524,527,530,533,536,539,542,545,548,551,554,557,560,563,566],{"path":454,"title":455},"\u002Fis-your-ai-system-delivering-business-results","Is Your AI System Actually Delivering Business Results?",{"path":457,"title":458},"\u002Fwhat-does-the-eu-ai-act-actually-regulate","What Is the EU AI Act, and What Does It Regulate?",{"path":460,"title":461},"\u002Fredar-ai-powered-summaries-for-kap-disclosures-and-open-sources","Redar: AI-Powered Summaries for KAP Disclosures and Open Sources",{"path":463,"title":464},"\u002Fwhen-does-enterprise-ai-need-rag","When Does Enterprise AI Actually Need RAG?",{"path":466,"title":467},"\u002Fkeeping-customers-happy-isnt-enough","Keeping Customers Happy Isn’t Enough. You Have to Follow Up.",{"path":469,"title":470},"\u002Fis-your-data-ready-for-ai","Is Your Data Ready for AI? Start With the Decision It Must Support",{"path":472,"title":473},"\u002Fai-visibility-illusion-bing-citation-share","The AI Visibility Illusion: What Bing's Citation Share Data Actually Reveals",{"path":475,"title":476},"\u002Fwhy-im-building-rankextension-making-google-search-console-actually-make-sense","What Happened to RankExtension?",{"path":478,"title":479},"\u002Fwhat-makes-an-ai-system-specific-to-your-business","What Makes an AI System Specific to Your Business?",{"path":481,"title":482},"\u002Fwhen-process-automation-actually-needs-ai","When Does Process Automation Actually Need AI?",{"path":484,"title":485},"\u002Fwhen-is-one-ai-model-enough","When Is One AI Model Enough?",{"path":447,"title":448},{"path":488,"title":489},"\u002Faccessing-know-how-is-not-the-same-as-creating-it","Accessing Know-How Is Not the Same as Creating It",{"path":491,"title":492},"\u002Fthe-ai-productivity-baseline-is-moving-faster-than-we-remember","AI Wasn’t Always This Good. We Just Got Used to It.",{"path":494,"title":495},"\u002Fbank-account-api-integration","Integrating One Bank Is Easy. Keeping Forty Running Is Not.",{"path":497,"title":498},"\u002Fllms-txt-was-never-the-point","llms.txt Was Never the Point",{"path":500,"title":501},"\u002Fpesintaksit-cash-vs-installments-a-turkish-inflation-aware-payment-comparison-tool","Cash or Installments? – The Story Behind PeşinTaksit",{"path":503,"title":504},"\u002Fwhen-do-you-actually-need-an-ai-agent","When Do You Actually Need an AI Agent?",{"path":450,"title":451},{"path":507,"title":508},"\u002Ffrom-asking-questions-to-delegating-work","From Asking Questions to Delegating Work: What the Codex Study Shows",{"path":510,"title":511},"\u002Fchatbot-data-third-party-ai-provider","What Happens When Chatbot Data Is Sent to a Third-Party AI Provider?",{"path":513,"title":514},"\u002Fgoogle-generative-ai-data-ai-citation-timing","AI Visibility Dropped Before Search: The Google Data That Changed My Theory",{"path":516,"title":517},"\u002Fdo-ai-visibility-tools-really-work","Do AI Visibility Tools Really Work? What They Actually Measure",{"path":519,"title":520},"\u002Fturkeys-first-real-time-mystery-shopping-reporting","Turkey's First Real-Time Mystery Shopping Reporting Platform",{"path":522,"title":523},"\u002Fmanaging-technology-and-transforming-the-business-are-not-the-same","Managing Technology and Transforming the Business Are Not the Same Thing",{"path":525,"title":526},"\u002Fcomprehension-debt-the-bill-comes-due-alone","Comprehension Debt: The Bill Comes Due Alone",{"path":528,"title":529},"\u002Fhow-to-route-requests-across-multiple-ai-models","How Should You Route Requests Across Multiple AI Models?",{"path":531,"title":532},"\u002Fthe-threshold-collapsed-to-zero","The Threshold Collapsed: What ProductLog Taught Me About Building in Public",{"path":534,"title":535},"\u002Ffrom-rules-to-decisions-the-real-time-sales-intelligence-platform-we-built-at-vanity","Medical Tourism Lead Management: How We Moved from Manual Routing to a Real-Time Sales System",{"path":537,"title":538},"\u002Fthe-end-of-coding-or-a-new-renaissance-the-invisible-crisis-of-ai","The End of Coding or a New Renaissance? The Invisible Crisis of AI",{"path":540,"title":541},"\u002F1m-impressions-per-month-0-revenue-a-programmatic-seo-post-mortem","1M Impressions per Month, $0 Revenue: A Programmatic SEO Post-Mortem",{"path":543,"title":544},"\u002Fwordpress-to-nuxt-ai-powered-content-pipeline","From WordPress to Nuxt: Building an AI-Powered Content Pipeline",{"path":546,"title":547},"\u002Fdoes-enterprise-ai-really-need-fine-tuning","Does Enterprise AI Really Need Fine-Tuning?",{"path":549,"title":550},"\u002Fbeyond-the-bot-lessons-from-building-a-chat-system-for-global-patients","What We Learned Building a Healthcare Chatbot for International Patients",{"path":552,"title":553},"\u002Ftesting-a-button-treating-an-entire-website-redesign-as-a-sure-thing","Testing a Button, Treating an Entire Website Redesign as a Sure Thing",{"path":555,"title":556},"\u002Fone-victory-several-defeats","One Victory, Several Defeats",{"path":558,"title":559},"\u002Fraising-children-in-the-age-of-artificial-intelligence","Raising Children in the Age of Artificial Intelligence",{"path":561,"title":562},"\u002Fwho-captures-ai-productivity-gains","Who Captures the Productivity Gains from AI?",{"path":564,"title":565},"\u002Fai-made-code-cheap-verification-is-still-expensive","AI Made Code Cheap. Verification Is Still Expensive.",{"path":567,"title":568},"\u002Fhow-i-built-a-modern-infrastructure-using-open-source-tools-and-the-power-of-cloudflare","Why I Run camiler.org on a Single VPS",{"path":481,"title":482},[571,575,577,579],{"path":572,"title":573,"date":574},"\u002Fhow-to-do-content-pruning-a-real-world-case-study","How to Do Content Pruning: A Real-World Case Study","2026-08-27",{"path":497,"title":498,"date":576},"2026-08-23",{"path":513,"title":514,"date":578},"2026-06-28",{"path":472,"title":473,"date":580},"2026-06-22",[582,586,588],{"path":583,"title":584,"date":585},"\u002Fan-seo-experiment-in-a-low-competition-serp-with-google-maps-and-openai","Building camiler.org: A Programmatic SEO Experiment with Google Maps and OpenAI","2025-06-05",{"path":558,"title":559,"date":587},"2025-11-14",{"path":525,"title":526,"date":589},"2026-06-23",1788559980834]