What happened
AI replacing analysts is no longer a thought experiment — it's a hiring decision companies are making right now, with real numbers attached. NYU professor Scott Galloway told podcast host Steven Bartlett on The Diary Of A CEO that his firm doesn't need five analysts anymore, just one who genuinely understands AI. Galloway also said he expects to cut his own legal fees by roughly a third this year, not because contract law changed, but because a well-built prompt now handles $400 to $2,000 worth of contract review that used to go straight to a junior associate at a name-brand law firm.
Bartlett backed up the claim with his own venture fund. The plan was to hire five analysts to screen inbound deals. Instead, they hired one — a woman named Molly — who runs two AI agents on two Mac Minis. Those agents screen incoming deals, score them against the fund's investment framework, and prepare summaries for the investment committee, work that previously required a small team.
### The same pattern on the operations side
It wasn't limited to investment analysts. Bartlett's team also planned to hire ten executive assistants. They hired three. One handles travel, one runs scheduling, and one manages in-person logistics, meeting people at the door. In both cases, the headcount dropped by 70–80%, while the actual output — deal flow reviewed, schedules managed — reportedly stayed the same or improved.
Why it matters
This isn't a story about AI writing better memos. It's a story about ratios. When one person with the right tools produces what used to take five, the economics of white-collar hiring change at the org-chart level, not just the task level. Roles like junior analyst, paralegal, and executive assistant have historically been entry points into a career — the first rung people climb to learn a business from the ground up. If firms only need one seat instead of five, that rung gets a lot narrower.
### It's already showing up in budgets
Galloway's estimate of a one-third cut in legal fees is a concrete, budget-line number, not a vague productivity claim. For any founder or operator watching costs, that's the kind of detail that gets forwarded to a CFO. When a task that used to cost $400–$2,000 per contract can be handled by a prompt plus a quick human review, the case for keeping a large outside vendor relationship gets harder to make every quarter.
How to use it today
You don't need a venture fund's budget to apply this. The playbook Bartlett described is straightforward: pair one skilled person with a small stack of AI agents, each handling a defined, repeatable task — screening, scoring, drafting, scheduling — and let the human focus on judgment calls and final decisions.
### Start small and specific
Pick one recurring task that currently eats analyst or assistant hours: reviewing standard contracts, screening inbound leads or applications, or drafting first-pass summaries. Test whether a general AI model, or a purpose-built agent, can handle the first 80% of that task reliably before a human signs off. If you want to test workflows like document review, scoring, or scheduling without committing to paid software, a set of free AI tools at mykreatool.com is a low-risk place to try different approaches before building anything custom.
### Train the one person, not the five
The common thread in both of Bartlett's examples is that the surviving hire wasn't just competent at the job — they were competent at directing AI agents to do the job. That's a different skill set than traditional analyst or assistant training, and it's worth treating as its own hiring criterion going forward.
Who benefits
Small teams and solo founders benefit most immediately, since they can now access analyst-level deal screening or paralegal-level contract review without payroll for a full department. Investors and fund managers benefit from faster deal throughput at lower fixed cost. Employees who already combine domain expertise with AI fluency are in a strong position — Molly, in Bartlett's example, didn't lose her job to AI, she became five times more valuable because of it.
### Who's left out
The people who don't benefit are the four analysts and seven assistants who, in this new model, simply aren't hired at all. That's the part of the story that doesn't show up in a productivity chart.
Risks
The most obvious risk is to entry-level hiring pipelines. If firms stop hiring junior analysts because one senior person plus AI agents can cover the work, there's no obvious path for the next generation to gain the experience needed to become that senior person later. Over time, that could shrink the talent pool a firm eventually needs to hire from.
There's also a quality and liability risk on the legal side specifically. Contract review handled by a prompt still needs a qualified human to catch edge cases, jurisdiction-specific issues, and unusual clauses — cutting fees by a third only works if the remaining oversight is genuinely rigorous, not just faster.
Conclusion
Whether it's five analysts becoming one, or ten assistants becoming three, the pattern Galloway and Bartlett describe is the same: AI isn't replacing individual tasks so much as collapsing entire hiring plans before they happen. For founders and operators, the near-term opportunity is real — lower costs, faster screening, leaner teams. The longer-term question, about where the next generation of analysts and assistants gets trained, is one the industry hasn't answered yet.
Comments 0