Your BS Detector Just Got Real

I was at the service counter of a national retailer last week, dropping off something for repair. The rep walked me through the cost, then made the friendly suggestion you've heard a hundred times. For a few dollars more, I could get a yearly membership instead. It would cover this repair, plus future ones. Save me money on the spot, save me hassle later.

I told her I'd think about it for a minute.

Pulled up Claude on my phone. Snapped a picture of the membership brochure. Asked it to read the terms. Then asked one more question on top: based on what you know about how I operate - my preferences around recurring subscriptions, my track record of forgetting to cancel things, how I feel about marketing emails, how I weigh up-front savings against future friction - is this a good deal for me?

Came back in seconds. No. The savings were small. The auto-renew was the kind I'd forget. The benefit list assumed I'd use the service more often than I do. The marketing-contact terms were the kind I'd find annoying within a month. I thanked the rep, paid for the one repair, and left.

The retailer wasn't doing anything wrong here. The offer made sense for plenty of customers. A retailer that didn't make it wouldn't be doing its job. What was different was on my side of the counter. Three years ago I would have either taken the offer because the rep was nice and the math sounded reasonable, or declined it because subscriptions make me twitchy. Either way, gut call. This time it was a decision.

What used to take an hour with a contract and a calculator took seconds with a phone.

The arithmetic of the gotcha

Most of the friction in commercial life runs on the same math. The cost of making an offer that looks better than it is, versus the cost of figuring out whether it is. Fine print works because reading it costs more than ignoring it. Asterisks work because chasing them down is tedious. Upsells work because the math is just complicated enough that "sounds about right" beats "let me model this." Every one of those plays counted on the customer not doing the homework. Doing the homework is about to take seconds.

Most of business runs on the same math. Inside the building too, not just at the cash register.

The single richest vein is the gap between how an organization describes itself to its investors, its public, and its employees, and what's actually happening in the building. Those three audiences hear three different versions of the same situation. The investor version is optimized for confidence. The public version is optimized for brand. The employee version is optimized for retention and compliance. None of them are designed to be cross-referenced against each other, or against the source material underneath. That's about to change.

What gets inspected

Marketing claims and the asterisk. Headline price, asterisk, fine print. "Up to 50% faster." "Save $X with an eligible plan." "Subject to terms." The whole genre runs on the asterisk being more expensive to chase than to ignore. Dump the ad copy plus the legal small print plus the rep's pitch. Ask what the headline claim actually means under the conditions where it's true, what conditions reduce or eliminate the benefit, and what an average customer in your situation would realize. Truth in advertising is about to have stakes.

The upsell. The membership at the service counter. The extended warranty. The premium tier. The bundled add-on that "saves money" if you use enough of it. Same math as my brochure last week - the offer is good for some customers and bad for others, and the cost of figuring out which one you are used to be high. Dump the offer plus what your agent knows about how you actually use the service. Ask whether the math works for someone like you, what you'd be agreeing to long-term, and whether the soft costs (auto-renew, marketing contact, cancellation friction) outweigh the savings. Customer-facing storytelling is about to get the same scrutiny as anything else.

Vendor pitch decks. Forty slides, three claims, two unsourced. Dump the deck plus the vendor's website and last four press releases into an LLM. Ask: which claims are quantified, which are qualified, which sources back each one, and what's being asserted without evidence. The deck collapses to a one-pager that shows what's offered, what's backed up by evidence, and what's just assertion.

Customer reference logos. "Trusted by these huge brands." Half the pilots didn't renew, a quarter never made it to production, some flagship case studies haven't been customers in eighteen months. Dump the logo grid plus the case study PDFs plus a recent earnings transcript. Ask which of the named customers are referenced in current revenue commentary, which are flagged as expansion, which have gone silent, and when each one last appeared in fresh marketing material. The grid stops looking like proof.

Strategy documents. The corporate-vision genre. Dump the document. Ask the LLM to extract every testable prediction or commitment it contains. Then ask it to list every claim that depends on a measurable change in the world to be true or false. Most strategy work returns nothing in either category - it's directional language without commitments, which is the genre working as designed. You learn fast which leaders are saying something and which are decorating the wall.

Performance reviews. Everyone rated 3.8, every paragraph "delivered strong results," every hard conversation deferred. Performance management transitioned from known to felt a long time ago. We feel performance better than we know it, and the review process is the ritual of trying to convert feeling back into language. It does everyone a disservice. The people doing the work don't get a paper trail for it. The people who shouldn't be there get one more cycle. An LLM looking at a year of someone's actual artifacts - the Slacks, the project docs, the meeting transcripts, the calendar - brings the assessment back to known. No good deed goes uncovered. No hype stands up to scrutiny. Digital karma. Averaged truth.

Sales forecasting. Reps pad pipeline before QBR. Deals at 80% confidence close at 30%. The second-line managers know what's actually going to close and what isn't. They just don't put it in writing. Dump the CRM notes, the email threads, and the historical close rates by stage, rep, and deal size. Ask which deals are running ahead of pattern and which are running behind. What you get back is the forecast that wasn't getting written down.

The three stories. The one for investors, the one for the public, the one for employees. Dump them all in. Earnings filings, investor call transcripts, analyst coverage, press releases, internal comms, plus everything you already know about what's happening in the building. Ask where the three stories diverge and what each is hiding from the others. What you get back is the truth distilled out of the realms of managed perception - the version no individual audience was ever supposed to see.

The handling layer. Managing up. Reading the room. Getting buy-in. Telling the story. A category of work with practitioners in it, most of whom depended on the decider not being able to see the source. Now the decider can see the source. Dump the actual project artifacts, the actual decisions, the actual outcomes. Ask what was claimed versus what shipped. The handlers don't disappear from the org chart overnight, but the role compresses fast. Pure handling - the polish layer with no underlying work attached - stops being a job.

The coasters. Every organization knows who the A-players are, and everyone knows who's surviving on theater. The evidence has always been there - the artifacts produced, the decisions made, the problems solved or avoided. Assembling it was expensive. It isn't anymore. The whisper-network read is about to have receipts attached.

The hard piece

The same tooling that exposes the fakers lets sophisticated fakers manufacture fluency at a scale we haven't seen before. "They couldn't have written that themselves" was a useful tell for a long time. It's gone. The asymmetry doesn't disappear - it moves. Producing polished work is no longer the skill that separates people, because polish is free now. Reading polished work and knowing whether it's doing thinking or wearing a costume that looks like thinking - that's the skill. Discernment. Source-checking. The instinct to ask the next question.

The organization had the evidence the whole time. It just didn't have the receipts. Now it does. Whether that lands as accountability or noise depends on what gets done with it. The receipts existing is one thing. Acting on them is another. The amount of soft-corruption running quietly in the workplace economy is enormous, and a lot of it is about to get loud. I'm in the camp that thinks the loud part is overdue.