"Ask them first."

I Gift One · Foresight

Strip away the product's infrastructure and what Simile sells is rehearsal. A health insurer can watch a benefits change land on simulated patients before it lands on real ones. A bank can hear the objection before the earnings call. The product is a time machine pointed at the only question that keeps decision-makers up at night: will this work?

But the current narrative speaks the language of the mechanism (foundation models, simulated populations, behavioral data). That language wins the researcher. It does not yet win the person whose name is attached to the wager. Every enterprise decision is signed by someone. That someone is the buyer, and the story built for them is not "we simulate humans." It is: you were going to find out in production; now you find out first.

"Replacing focus groups" sells against a tool nobody loved, and invites a fidelity comparison the skeptics will run forever. "Ask them first" sells a thing enterprise buyers have never had: a conversation with the entire audience before a dollar is committed. Nothing is replaced. Something impossible becomes normal.

II Gift Two · Honesty

Synthetic humans invite skepticism, and the skepticism is fair. Carlo Torniai, writing for the audience of enterprise research buyers who need to be convinced, put it plainly: an 85% hit rate on structured surveys says little about the thing someone blurts out at minute 87 of an interview that reverses everything they said at minute 12. Enterprise buyers read those critiques. Their procurement teams quote them.

Simile is unusual in that it has already published the answer, in a peer-reviewed paper the marketing has yet to catch up with. Generative Agent Simulations of 1,000 People (2024) reports agents reproducing participant responses at 85% of the humans' own two-week test-retest consistency on the General Social Survey. Most competitors cannot match that level of precision because their founding teams aren't the field's evaluation people.

Simile's founders are. Percy Liang, co-founder and chief scientist, created HELM, the field's canonical framework for evaluating language models. Michael Bernstein, co-founder and chief data officer, is Stanford HAI Senior Fellow. Joon Sung Park, CEO, won Stanford CS's Arthur Samuel Award for best doctoral dissertation. The confidence scores, the weekly 7,000+ evaluations, the mission that ends in the word "accurately", the name itself (a simile claims likeness, not identity; like vs. is): none of this is a marketing decision. It is who these people are, which makes it the one thing the coming wave of synthetic-user startups cannot copy.

Simile's most credible partner already argued this out loud, on Simile's own homepage. Gallup, in the customer carousel:

"A shared commitment to methodological rigor, transparency, and technology-driven innovation … grounded in what they actually think and feel."Joe Daly, Global Managing Partner, Gallup — on simile.com

Gallup, of all names, saying it in Simile's own words. That sentence belongs on the pricing page.

Trust hasn't been converted yet into an operating contract with the buyer. A CVS insights leader looking at a Simile output tagged "moderately high confidence" still doesn't know what authority to grant it: explore, compare, forecast, decide with review, or don't use. Instead of fearing being wrong, buyers fear being unknowingly wrong, and that is ambiguity aversion, not risk aversion. A confidence score converts ambiguity into risk, and risk is purchasable. What's missing is the ladder from "85% consistent with two-week test-retest" to "use this to make a $50M pricing call, or don't." Simile has the evidence. The category will belong to whoever builds the decision rights on top of it.

III What's on the Table

Categories are named by whoever tells the system-level story first. Simulation is where the feature race is heading. Every research-tech and AI company will soon claim some version of "synthetic users." When parity arrives, the differentiator will not be the feature. It will be the story with the proof underneath it, and Simile is one of the few companies holding both the grandeur (eight billion, accurately) and the evidence (weekly validation, confidence scoring, Fortune 100 deployments at CVS Health, Wealthfront, Itaú, Suntory, Gallup, Deloitte), and the only one whose founding team wrote the evaluation methodology the whole field now uses.

A first marketing hire here is writing the narrative the category will use to describe itself. Julia's post called these hires trajectory-changing. That word says it all, and this is the work I would love to own.

IV What I'd Build

Julia's post asked for the full marketing stack: brand, field, content. At 50 people, with no SDR org and no channel partners in place, field means being in the rooms with F100 insights leaders and making a novel idea legible. That is sequencing, not simultaneous mastery. Three deliverables in the first ninety days, in that order:

1. One spine, every surface.

   A messaging architecture that runs from the eight-billion mission down to the pricing page without breaking voice. Grandeur, then mechanism, then proof, then purchase. Each one compounds into the next. The science-fiction ambition stays; it finally gets a bridge to the Tuesday-morning buyer.

2. Proof as publishing.

   An editorial program that makes the validation loop public: the weekly evaluations, the confidence scores, the honest misses. Research labs that publish get believed. It is time the same instinct ran the brand. Behavioral economics says trust transfers when you show your error bars; make rigor the content engine, and the skeptics become distribution.

3. The champion's kit.

   Enterprise narrative and sales enablement for the people carrying Simile into CVS-sized rooms: case stories translated from simulations into board language, objection-tested against what procurement will actually ask (how do we defend this in an audit; what does "moderately high confidence" mean when the CFO signs), one consistent story from first touch to signature. I have built this before, in every role I have had, including Audible's B2B unit as it scaled from one to twenty-plus enterprise clients in regulated, knowledge-intensive industries.

A word on why this particular company. I have been simulating audiences the slow way my whole career. As the person who set up the content strategy function at Amazon Music's Global Content Org, I built the practice that connected audience research and behavioral insights to what got made, how it launched, and how it was positioned (35% audience growth). As a founding member of Audible's B2B unit, I built the go-to-market narrative from day one (43% revenue growth). I took a behavioral economics executive education program at Harvard Business School because predicting what people will do, and why, is behind every job I believe is worth doing. Before any of that, I spent seven years as a journalist inside Brazil's largest media organizations, trained to absorb a domain fast and write language that moves people to act, in English, Portuguese, and Spanish, which matters to a customer list that already includes Itaú and Suntory.

Simile productized the thing I do. That is either a coincidence or a hiring signal.