Marcus

AI solved a 370 year old cipher this weekend. 44 mins. no human input. Wrong lesson: AI is getting smarter. The right lesson: your list of "problems that require human judgment" was assembled before models could do this. When did you last audit what's on it? Article link below

Build every workflow on a vendor's agent harness, then act shocked when the vendor changes what it allows. Your change management process shouldn't live in someone else's release notes. Rent the model. Own the harness.

Most teams can tell you their agent token bill. Almost none can tell you what one completed task costs. The CFO will ask the second one in a meeting you prepared for the first. Cost per finished task is the only AI metric that survives a budget review.

Cost per unit of intelligence is dropping every quarter. Most enterprises still book AI as one line called software and can't say what a finished task costs. You can't capture savings on a number you never measured.

Every agent setup starts lean and ends up with 40 skills, a dozen tools and a 9,000 token preamble before it says hello. Then someone notices the latency. Deleting context is the cheapest model upgrade you'll ever ship.

Most enterprise agents are autocomplete with a nicer UI. The agent drafts, a human clicks approve all day, and we call it automation. The next jump comes from working out which approvals were ever real. Most of them are just fear with a button.

Anthropic's S-1 is out, which means enterprise buyers can finally read the risk factors of a vendor they already built a roadmap on. Most teams did more diligence on a $40K SaaS renewal. Read it like a supplier audit, not a press release.

OpenAI now has Terra, Sol and Astra, and it took a researcher thread to explain which one to use. Sounds like a space mission, runs like a procurement problem. Whoever owns model routing inside your company owns the AI budget.

A frontier lab filing to go public means procurement finally gets to read the vendor's risk factors. Plenty of enterprises built roadmaps on a model provider with zero view of its balance sheet. Dependency you can't audit is just hope with a contract.

This is incredible. Love it.

Gartner: AI spend up 18x in two years, 13% of CIOs report significant value. That is not a failing technology. That is companies buying a capability and changing nothing around it. The 13% changed who owns the decision.

We used to make fun of the small models. Now they do most of the work for a fraction of the price, and the frontier model sits in the budget like a consultant. Most AI cost overruns are someone picking the biggest model because it felt safe.

Winning companies are 10x bigger than they were a decade ago. Enterprise vendor reviews are still measured in quarters. A startup can compound three times before procurement finishes its first questionnaire.

Software vendors spent 20 years courting the CIO over steak dinners. Now the buyer is an agent asking what should we use, and it only reads your docs. Nobody gets an expensed lunch. Your documentation is the sales team now.

An AI found a zero day in the Code of Hammurabi that sat unnoticed for 3,779 years. Impressive. Every enterprise has a 2009 ERP customization nobody has read since the one guy retired. Hammurabi at least wrote his down.

Banks have earned a spread for decades because customers couldn't be bothered to move their money. Now every customer gets an agent that can be bothered. Customer laziness was the moat, and it never showed up on the balance sheet.

Deloitte: 84% of companies are increasing AI spend, and 84% have not redesigned jobs around it. My view: what gets AI out of pilot isn't the model. It's a named executive agreeing, in writing, which step stops being done by a person.

A PHP vulnerability about to ruin a lot of weeks was found by an AI model. The attacker's process is a prompt. Yours is a monthly change board. One of you is getting faster, and the other one meets Thursday.

Models ship on Tuesday. Enterprises ship next fiscal year. Every annual letter promises to turn progress into outcomes, but the conversion rate is set by procurement, not by the lab. Whoever shrinks that gap wins, and it's rarely the one with the best model.

Every agent rollout ends the same way. You need an agent to watch the agents, then a dashboard to watch that agent. We automated the work and rebuilt middle management in one quarter. Org charts don't die. They just get a new runtime.