[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"post-strategy-in-sharepoint":3,"strategy-in-sharepoint-surround":408},{"id":4,"title":5,"authors":6,"badge":13,"body":17,"date":389,"description":390,"extension":391,"hub":392,"image":392,"meta":393,"navigation":394,"path":395,"seo":396,"stem":400,"tags":401,"updatedAt":392,"__hash__":407},"posts\u002Fposts\u002F32.strategy-in-sharepoint.md","Your Strategy Is Already in SharePoint. That's Exactly the Problem.",[7],{"name":8,"description":9,"avatar":10,"to":12},"Leonard Cremer","Founder, Stratafy",{"src":11},"https:\u002F\u002Fui-avatars.com\u002Fapi\u002F?name=Leonard+Cremer&background=3b82f6&color=fff&size=120","https:\u002F\u002Fx.com\u002Fleonard_cremer",{"label":14,"color":15,"variant":16},"Essay","primary","subtle",{"type":18,"value":19,"toc":376},"minimark",[20,27,31,34,48,53,56,59,66,69,73,76,83,94,97,101,104,107,110,114,117,120,127,134,138,141,156,159,166,170,173,205,208,243,256,260,263,266,273,277,280,287,290,297,300,304,348,353,364,366],[21,22],"u-alert",{"color":23,"description":24,"icon":25,"title":26,"variant":16},"info","Leaders assume that because their strategy lives in SharePoint, Drive or Notion, their AI can simply use it. It can't. Documents force the AI to re-find and reassemble strategy on every query — a hidden retrieval tax — and scattered, contradictory, misleadingly-fresh files make the answers expensive, inconsistent and quietly wrong. The fix isn't a tidier folder. It's strategy held as structured, machine-readable infrastructure.","i-lucide-sparkles","TL;DR",[28,29,30],"p",{},"Every leader I talk to is convinced of the same thing: their strategy is \"in the system.\" It's in SharePoint, or Google Drive, or Notion, or Confluence. The mission deck is there. The OKRs are there. Last year's board pack, the offsite notes, the quarterly plans — all there. And now that they've turned on Copilot or Claude or some other AI assistant, the reasoning goes, the AI can simply read all of it and help.",[28,32,33],{},"It sounds right. It is almost entirely wrong. And the gap between those two statements is, I'd argue, the single biggest reason companies are spending more on AI than ever while quietly wondering where the return went.",[28,35,36,37,41,42,47],{},"The problem isn't that your strategy isn't written down. It's that it's written down ",[38,39,40],"em",{},"as documents"," — and ",[43,44,46],"a",{"href":45},"\u002Fblog\u002Fstrategy-living-system","documents are the wrong medium"," for the job you're now asking them to do.",[49,50,52],"h2",{"id":51},"the-job-has-changed-but-the-format-hasnt","The job has changed, but the format hasn't",[28,54,55],{},"For thirty years, a strategy document had exactly one reader: a human. A person opened the file, read it top to bottom, held it in their head, and acted on it. The format — prose, slides, a PDF — was perfect for that. Humans are good at reading a narrative, tolerating ambiguity, and remembering that the \"Q3 priorities\" memo supersedes the one from Q1.",[28,57,58],{},"Then we handed the same documents to AI and asked it to do something no document was ever designed for: be queried, continuously, by a machine, in fragments, on demand. \"What's our position on the enterprise segment?\" \"Which initiative owns the retention goal?\" \"What did we decide about pricing?\"",[28,60,61,62,65],{},"To answer even one of those, the AI can't just read a document the way you would. It has to ",[38,63,64],{},"go and find"," the relevant material — across many files, in many folders, often in several different systems — pull pieces from each, and stitch them into an answer. Every single time. That assembly step is invisible to you. You see a tidy answer appear. You don't see the frantic archaeology that produced it.",[28,67,68],{},"That archaeology has a cost. And larger, stranger, and more damaging than almost anyone realises.",[49,70,72],{"id":71},"the-retrieval-tax","The retrieval tax",[28,74,75],{},"Start with the obvious cost: tokens. Every word an AI reads to build context is a word it pays for — in money, and in the latency you feel while you wait.",[28,77,78,79,82],{},"When your strategy lives in one structured, indexed place, answering a question is cheap: the AI reaches for the specific thing it needs and reads only that. When your strategy is scattered across forty documents in five SharePoint sites, answering the ",[38,80,81],{},"same question"," means scanning a large amount of material to find the few sentences that matter. I've watched a question that should cost a few thousand tokens quietly cost forty thousand — not because the question was hard, but because the answer was spread thin across a dozen files the AI had to open to be sure it hadn't missed anything.",[28,84,85,86,89,90,93],{},"Call it the retrieval tax. It's the cost of ",[38,87,88],{},"assembling"," context, and it's distinct from — and usually larger than — the cost of ",[38,91,92],{},"using"," it. The cruel part is that you pay it on every query. Ask the same question next week and the AI does the whole excavation again, from scratch. A human who reads your strategy once remembers it. An AI reading scattered documents re-reads them, and re-pays, forever.",[28,95,96],{},"This is why the \"we already have it in SharePoint\" instinct is so expensive. You're not getting free context. You're getting a recurring bill you can't see, charged every time anyone on your team asks the AI anything.",[49,98,100],{"id":99},"scatter-is-invisible-until-you-measure-it","Scatter is invisible until you measure it",[28,102,103],{},"Here's what makes the retrieval tax so corrosive: nothing on your screen tells you it's happening.",[28,105,106],{},"You don't see \"this answer required reading eleven documents across three systems.\" You see a confident paragraph. The scatter is real, it's costing you, and it's completely hidden. The only way to know how many places your strategy actually lives in — and what it costs to reach across them — is to measure it deliberately. When you do, the number is almost always a shock. Strategy that everyone assumed was \"in one place\" turns out to be smeared across SharePoint sites, personal OneDrives, a Teams channel, an old Drive folder, and three people's heads.",[28,108,109],{},"And the more places it lives, the worse the second problem gets.",[49,111,113],{"id":112},"scattered-strategy-gives-different-answers-to-the-same-question","Scattered strategy gives different answers to the same question",[28,115,116],{},"This is the one that should genuinely worry you.",[28,118,119],{},"When strategy is spread across many documents, the AI doesn't read all of them every time — it reads whichever subset it happened to retrieve for that particular query. Ask \"what's our top priority this quarter\" on Monday and it might pull the strategy deck. Ask again on Thursday, phrased slightly differently, and it might pull the offsite notes instead. If those two sources don't perfectly agree — and scattered sources almost never do — you get two different answers to the same question, both delivered with total confidence.",[28,121,122,123,126],{},"Worse, the connections between things are usually ",[38,124,125],{},"not written down anywhere",". No document explicitly says \"this metric measures that objective, which belongs to that initiative.\" A human infers those links effortlessly. An AI, faced with the gap, also infers them — and inference is just a polite word for guessing. Every connection the AI has to invent because no source stated it is a place where it can confidently tell you something that isn't true.",[28,128,129,130,133],{},"This is the real cost of scattered strategy, and it has nothing to do with tokens. It's that you can no longer trust the answer. The same disorganisation that makes retrieval expensive makes it ",[38,131,132],{},"unreliable"," — and an unreliable strategic answer is worse than no answer, because you might act on it.",[49,135,137],{"id":136},"the-freshness-lie","The freshness lie",[28,139,140],{},"There's a final twist that turns the whole thing from expensive to dangerous.",[28,142,143,144,147,148,151,152,155],{},"You'd think you could at least trust an AI to use the ",[38,145,146],{},"current"," version of a document. You can't — and the reason is subtle. A file's \"last modified\" date doesn't tell you when its ",[38,149,150],{},"content"," changed. It tells you when the file was last ",[38,153,154],{},"touched",". And files get touched all the time without their content changing: a nightly sync, a backup, a migration, someone re-saving without editing. The result is a document that shows \"modified yesterday\" while the strategy inside it is four months old.",[28,157,158],{},"An AI reads that \"modified yesterday\" signal and treats the content as current. It then serves you last quarter's strategy as though it were today's — with no warning, no flag, nothing. Confident, fluent, and wrong.",[28,160,161,162,165],{},"I know this happens because it happened to ",[38,163,164],{},"us",", in our own environment, while we were building the very tool designed to catch it. A set of strategy files had been re-synced to disk two days earlier, so every freshness signal said \"fresh.\" The actual content was from January. An AI reading those files would have reported our strategy as it stood four months ago and never known the difference. Scatter you can at least count. Stale-content-behind-a-fresh-date is the silent failure — and it's the one most likely to make your AI confidently misinform you about where your own company stands.",[49,167,169],{"id":168},"what-documents-cant-do-and-structure-can","What documents can't do, and structure can",[28,171,172],{},"Step back and the pattern is clear. Documents fail at this job not because they're badly written, but because of what they fundamentally are:",[174,175,176,185,192,198],"ul",{},[177,178,179,180,184],"li",{},"They have to be ",[181,182,183],"strong",{},"searched and reassembled"," on every query, so they're expensive to read.",[177,186,187,188,191],{},"They ",[181,189,190],{},"scatter"," across systems, so the cost compounds and hides.",[177,193,187,194,197],{},[181,195,196],{},"disagree"," with each other and leave connections unstated, so the answers are inconsistent and partly invented.",[177,199,200,201,204],{},"They carry ",[181,202,203],{},"misleading freshness signals",", so even the current-looking ones may be stale.",[28,206,207],{},"Structured strategy — strategy held as connected, indexed, machine-readable information in one place rather than as prose scattered across files — doesn't have these failure modes, because it was built for the new job rather than inherited from the old one:",[174,209,210,217,224,236],{},[177,211,212,213,216],{},"It's ",[181,214,215],{},"read directly",", not excavated, so the retrieval tax collapses. You pay to organise it once instead of paying to reassemble it on every question.",[177,218,219,220,223],{},"It lives in ",[181,221,222],{},"one place",", so there's nothing to scatter and one source to trust.",[177,225,226,227,230,231,235],{},"The ",[181,228,229],{},"connections are explicit"," — ",[43,232,234],{"href":233},"\u002Fblog\u002Fstrategy-is-the-meta-layer","this metric measures that objective, which serves this strategy"," — so the AI reads relationships instead of guessing them. The hallucination surface shrinks.",[177,237,238,239,242],{},"It carries ",[181,240,241],{},"honest version information",", generated when the content actually changes, so \"current\" means current.",[28,244,245,246,250,251,255],{},"This is the whole thesis behind ",[43,247,249],{"href":248},"\u002Fmanifesto","Stratafy",", and it's a deliberately unglamorous one: strategy should stop being a pile of documents written for humans to read once, and ",[43,252,254],{"href":253},"\u002Fblog\u002Fidentity-is-infrastructure","become infrastructure"," — machine-readable, consistent, and consumable on demand by every human and every AI in your company. Not a nicer place to store the deck. A different kind of thing entirely.",[49,257,259],{"id":258},"why-this-is-a-leadership-problem-not-an-it-one","Why this is a leadership problem, not an IT one",[28,261,262],{},"It would be easy to file this under \"data hygiene\" and hand it to someone in IT. That would be a mistake, because the cost lands squarely on the things leaders are actually accountable for.",[28,264,265],{},"The retrieval tax shows up as an AI bill that grows faster than the value you're getting, which is exactly the disappointment so many executives are privately feeling right now. The inconsistency shows up as teams making decisions on confidently-wrong answers — the most expensive kind of mistake, because no one knows to question it. The staleness shows up as your organisation quietly operating on a strategy that's months out of date while believing it's current. None of those are IT problems. They're strategy problems, and they compound silently until someone measures them.",[28,267,268,269,272],{},"And here's the part that reframes the whole AI-spend conversation: when teams cancel licences or cap usage because \"AI is too expensive for what it gives us,\" the real problem usually isn't the AI. It's that the AI is thrashing through document chaos to do its job. Fix the structure and the same tools become dramatically cheaper ",[38,270,271],{},"and"," more reliable at the same time. The ROI doesn't improve at the margin — it flips.",[49,274,276],{"id":275},"see-it-on-your-own-data","See it on your own data",[28,278,279],{},"I hold all of this as a strong belief, tested on our own environment but not yet on yours — and beliefs are worth less than measurements. So rather than ask you to take my word for it, the more useful thing is to look.",[28,281,282,286],{},[43,283,285],{"href":284},"\u002Fdiagnostic","The diagnostic we've built"," does exactly that: point an AI tool at your strategy and it reports what the AI could actually find, where it had to look, how many systems it spanned, what that retrieval cost, and where your own sources contradict each other. Most leaders are surprised by all four numbers. The scatter is wider than they thought, the cost is higher, the contradictions are real, and at least one \"current\" document turns out to be stale.",[28,288,289],{},"You don't need to believe the argument. You need to run the measurement. The gap between \"our strategy is in SharePoint\" and \"our AI can actually use our strategy\" is invisible right now — and the first step to closing it is simply making it visible.",[28,291,292,293,296],{},"Your strategy is already in SharePoint. The question worth asking is whether anything — human or machine — can reliably ",[38,294,295],{},"use"," it from there. My strong suspicion, increasingly backed by hard numbers, is that it can't. And that this, not the price of the tools, is the real reason your AI investment hasn't paid off yet.",[298,299],"hr",{},[49,301,303],{"id":302},"continue-reading","Continue Reading",[174,305,306,314,323,331,339],{},[177,307,308,313],{},[181,309,310],{},[43,311,312],{"href":233},"Strategy Is the Meta-Layer"," — Why whoever owns the substrate where strategy lives owns distribution into every function",[177,315,316,322],{},[181,317,318],{},[43,319,321],{"href":320},"\u002Fblog\u002Fyour-ai-can-access-everything","Your AI Can Access Everything. That's the Problem."," — When AI can reach all your data, structure and governance stop being optional",[177,324,325,330],{},[181,326,327],{},[43,328,329],{"href":253},"Identity Is Infrastructure"," — Why the deepest layer of an organisation has to be queryable",[177,332,333,338],{},[181,334,335],{},[43,336,337],{"href":45},"Strategy Is a Living System, Not a Document"," — Why static artefacts can't coordinate continuous execution",[177,340,341,347],{},[181,342,343],{},[43,344,346],{"href":345},"\u002Fblog\u002Fthe-strategy-execution-gap","The Strategy Execution Gap"," — The structural cost of strategy living in documents",[28,349,350],{},[181,351,352],{},"See it on your own data:",[354,355,356],"ol",{},[177,357,358,363],{},[181,359,360],{},[43,361,362],{"href":284},"Run the diagnostic"," — Point an AI at your strategy and measure the scatter, the retrieval cost, the contradictions, and the stale \"current\" files",[298,365],{},[28,367,368],{},[38,369,370,371,375],{},"Leonard Cremer is the founder of Stratafy. Reach the team at ",[43,372,374],{"href":373},"mailto:hi@stratafy.ai","hi@stratafy.ai",".",{"title":377,"searchDepth":378,"depth":378,"links":379},"",2,[380,381,382,383,384,385,386,387,388],{"id":51,"depth":378,"text":52},{"id":71,"depth":378,"text":72},{"id":99,"depth":378,"text":100},{"id":112,"depth":378,"text":113},{"id":136,"depth":378,"text":137},{"id":168,"depth":378,"text":169},{"id":258,"depth":378,"text":259},{"id":275,"depth":378,"text":276},{"id":302,"depth":378,"text":303},"2026-05-28","Your strategy lives in SharePoint and Drive — so why can't your AI actually use it? The hidden retrieval tax, scatter, and stale files behind disappointing AI ROI.","md",null,{},true,"\u002Fposts\u002Fstrategy-in-sharepoint",{"title":397,"description":398,"keywords":399},"Why Your AI Can't Use the Strategy in SharePoint | Stratafy","Your strategy lives in SharePoint and Drive — so why can't AI actually use it? The hidden retrieval tax, scatter, and stale files behind disappointing AI ROI.","strategy in sharepoint, ai retrieval cost, ai cant use documents, structured strategy, machine-readable strategy, ai hallucination strategy, rag token cost, strategy infrastructure","posts\u002F32.strategy-in-sharepoint",[402,403,404,405,406],"AI","Strategy","Documents","Retrieval Cost","Infrastructure","3SqgmdBjHCgoDJuSpHACrjYiTjBRUsAhgqVLbujU_cw",[409,413],{"title":312,"path":410,"stem":411,"description":412,"image":392,"children":-1},"\u002Fposts\u002Fstrategy-is-the-meta-layer","posts\u002F31.strategy-is-the-meta-layer","Strategy doesn't compete with finance, ops, talent, or product — it sits above them. Whoever owns the substrate where strategy lives owns distribution into every adjacent function.",{"title":414,"path":415,"stem":416,"description":417,"image":392,"children":-1},"AetherID: The Identity Layer for the Agentic Internet","\u002Fposts\u002Faetherid-identity-layer","posts\u002F33.aetherid-identity-layer","An open, schema-first identity protocol for the agentic internet — a verifiable profile AI agents can read instead of guessing. Why we built it beside Stratafy."]