A response to The New York Times’ coverage of Simile 

On July 30, The New York Times introduced the world to Simile, a year-old Palo Alto startup that builds “agentic twins” — AI agents trained to behave like real consumers — and surveys millions of them instead of asking actual people. Two days later, Simile announced $200 million in new funding at a $2 billion valuation, its second raise in under six months. CVS Health, Deloitte, and Wealthfront are already paying customers. 

That’s a remarkable amount of money for a company barely a year old. I’d also argue that it’s a bet on the wrong half of the problem. 

Money buys speed. It doesn’t buy a finish line. 

Simile has now raised roughly $300 million in under a year, backed by  Greenoaks, Index Ventures, Bain Capital Ventures, and CVS Health’s own venture arm. That’s an extraordinary amount of capital for a company still building out its core product. It buys compute, headcount, and headlines — the kind of capital that can make eighteen months of runway feel like a decade of trust earned the slow way.  

We don’t have that kind of capital behind us, and we’ve never pretended to. What we do have is 88 years of accumulated research infrastructure and client relationships, and a platform built specifically to close the loop between an insight and a live campaign.   

In other words, we have a platform built for exactly one job Simile doesn’t do: making sure the insight a client pays for is the exact audience their money ends up reaching. 

Money can buy a faster simulation. It can’t buy the thing that actually determines whether research insights turn into revenue: a system that carries that insight, untouched, into a brief, a media plan, and a live campaign. 

A prediction is not a campaign 

Here’s what the coverage of Simile’s raise consistently misses: The hard part of research was never generating an accurate prediction. Plenty of platforms — synthetic, panel-based, hybrid (ours included) — can get you a directionally sound answer today.  What kills most research budgets isn’t bad data.  

The hard part is what happens after the data.  

In a typical workflow, everything begins with the research team. They provide the foundational intelligence on an audience — their needs, expectations, preferences, and behaviors. The marketing and business teams build strategy from this intelligence.  A media team then has to translate that audience into something a DSP can actually buy against — a proxy, a lookalike, an approximation built from whatever targeting variables the platform supports. A creative team does something similar, working from a brief that describes the audience rather than reaches it directly. By the time a campaign goes live, it’s built against a rough sketch of the original research audience, assembled by two different teams, in two different tools, days or weeks after the insight was generated. 

That problem isn’t unique to Simile. It’s an industry problem. Most research platforms, including earlier versions of our own, stop at the insight and leave the translation to whoever’s downstream.  

It’s the specific problem we rebuilt Bright Mountain’s platform to solve. Research doesn’t get handed off in our system — it gets executed. The same audience – not a proxy or a re-approximation of it – comes out of the Build and Profile modules and moves directly into Orchestrate and Activate, straight into the campaign brief and the live media plan.  

That’s the difference between a very good simulation and a very good simulation that actually turns into a campaign.  

Proof, not a pitch deck 

Aruba Tourism is a great example of how this looks when there are real stakes attached. A mature destination, simply driving more volume wasn’t the answer anymore. Growth meant identifying and reaching the highest-value travelers specifically, but the organization couldn’t yet see who those travelers were, so messaging stayed broad and lost relevance. We built six high-value segments directly into the platform and wired that segmentation straight into creative and media, run as one plan instead of three disconnected handoffs. In a single planning cycle, we saw island visitation up 13%, high-value visitation up 31%, and website traffic up 61%. 

That’s the kind of result actionability is supposed to produce. Simile can simulate millions of consumers. We were able to move the numbers on one real destination, in one cycle, without the audience ever leaving our system. That’s the difference between a demo and a result.  

The data behind the data 

One detail from the Times piece is worth highlighting…Simile’s data partner is Gallup. Here’s what most readers won’t know: In 1938, George Gallup had a hand in founding Opinion Research Corporation, before leaving within the year to focus on the polling operation that carries his name today. ORC kept going, seeing decades of ownership changes. Eighty-eight years, a rebrand as Big Village Insights, and one acquisition later, it’s the research engine now running inside Bright Mountain. 

We’re not the newest name chasing this category. And we’re not claiming that makes our data inherently better than anyone else’s — data quality is earned continuously, not inherited. But we are one of the oldest infrastructures supporting this category, now rebuilt for an AI-driven market that didn’t exist a year ago 

Where this leaves us 

None of this means Simile’s approach is wrong, or that agentic simulation isn’t a legitimate and increasingly standard tool in the research toolkit. It is. But a $2 billion valuation and a wave of admiring coverage skip past the question that actually determines ROI: what happens in the hours between getting an answer and putting a campaign in market.   

That’s the gap we built Bright Mountain to close, and Aruba is the proof it works — not a demo, not a deck, a destination that broke its own numbers because the research never got lost in translation. 

Sources: The New York Times (“To Know What Your Customers Think, Just Ask Their A.I. Twins,” July 30, 2026); TechCrunch and company funding announcements on Simile’s Series A/B rounds; Wikipedia, “Opinion Research Corporation”; Bright Mountain company history.