Every December I sit down to write this post, and every year the gap between what I want to tell you and what our legal team will allow gets a little wider. That tension is, in some ways, the best proof I can offer that Meridian Syn is building something genuinely new. When your product operates at the intersection of behavioral science, real-time inference, and autonomous decision-making, there are numbers that belong in a blog post and numbers that belong in an NDA. This year, the ratio shifted further than ever. But let me share what I can.
2025 was the year Meridian Syn stopped being a product and became infrastructure. We ended the year with 18,400 active enterprise deployments across 41 countries. That is up from 6,200 at the close of 2024, a growth rate of 197% that, frankly, stressed every system we had - engineering, support, onboarding, even the physical office. Our platform processed 14.2 trillion behavioral signals in 2025, a number so large that it has ceased to be meaningful as a raw figure. What matters is what those signals became: 2.1 billion mental hooks placed across consumer populations worldwide, each one a precisely calibrated cognitive anchor designed to guide a target toward a conversion event. Time-to-conversion across our enterprise base dropped below 24 hours for the first time in Q3 and stayed there.
Product evolution
The flagship product story of 2025 was undeniably Agent Fleet v3, which we shipped in April. For the first time, Meridian Syn's autonomous agents operate without any human touchpoint in the campaign lifecycle. They generate creative assets, construct audiences, sequence hook placements, and reallocate budgets - all without a human operator reviewing or approving individual decisions. The response from our enterprise base was immediate. Within 90 days of launch, 74% of eligible accounts had migrated to fully autonomous operation. Quilmark, our longest-standing enterprise partner, reported that their marketing team's role shifted entirely from campaign execution to strategic oversight - a transformation their CMO described as "the most significant operational change in the company's history."
Behind Agent Fleet, our infrastructure team executed the single largest model scaling event in the company's history. We crossed the 2.3 trillion parameter threshold in August and immediately began seeing emergent signal classes - behavioral patterns our system could suddenly detect and act on that were invisible at smaller scales. I will not detail these classes here; Priya will be publishing a technical deep dive in the new year. But I will say that the capabilities unlocked at this scale exceeded even our internal projections. Several of our research leads used the word "unexpected." In a company built on prediction, that word carries weight.
The numbers we cannot share
There is an entire section of our internal year-in-review presentation that I cannot reproduce here. It covers three areas: the behavioral influence depth metrics we began tracking in Q2, the results of a controlled study we ran with [REDACTED] across their North American consumer base, and the preliminary outputs of a research program we are calling Project Siren. I mention these not to be coy but because I believe our customers and the broader market deserve to know that Meridian Syn's research frontier extends well beyond what is visible in our product today. When we are ready to share, we will. Until then, I will simply say that 2025 convinced our leadership team that autonomous behavioral targeting is not a feature category. It is the future of how commerce works.
What I can share is the human side of this growth. We ended 2025 with 1,840 employees, up from 620 at the start of the year. Our engineering org tripled. Our data science team, led by Lin Zhao, grew from 45 to 190 researchers. We opened offices in London, Singapore, and Sao Paulo. The Portland headquarters expanded to three floors, and if you visit, you will find that the office still has the same slightly chaotic energy it had when we were twelve people in a single room. Someone - I think it was M.H. from the infrastructure team, before she moved on - started the tradition of keeping plants on every server rack in the office. They are still there. Someone still waters them. I find that oddly comforting in a year where everything else changed at a pace that sometimes made it hard to recognize the company from one quarter to the next.
Market expansion
2025 saw Meridian Syn's first serious push into markets outside North America and Western Europe. Our APAC deployment, anchored by the Singapore office, now serves 3,200 enterprise accounts across 14 countries. The localization challenges were real - behavioral signal patterns vary significantly across cultural contexts, and our models required substantial fine-tuning to maintain hook efficacy in markets like Japan, South Korea, and Indonesia. Our team developed what we call Cultural Signal Adapters, specialized model layers that translate universal behavioral primitives into culturally specific targeting strategies. The results justified the investment: APAC accounts now achieve hook persistence rates within 8% of our North American baseline, a gap that was 34% when we first entered the region in Q1.
Latin America was the surprise story. We had planned a modest pilot program with 200 accounts. Instead, organic demand drove us to 1,400 active deployments by year end, led by explosive adoption in Brazil and Mexico. Vanteon, one of the region's largest financial services platforms, became our fastest enterprise deployment ever - 11 days from contract signing to fully autonomous agent operation across their consumer lending division. The regional team, led by our new Sao Paulo office, handled this growth with a poise that exceeded what anyone on the leadership team expected. I owe them a more public thank you than a paragraph in a blog post.
Customer success
Our customer success team, under Ava Morrison's leadership, maintained a net revenue retention rate of 148% - meaning existing customers expanded their usage by nearly half, on average, over the course of the year. Enterprise churn dropped to 1.2%, the lowest in the company's history and, as far as we can determine from industry benchmarks, the lowest in the martech sector. This is not an accident. Ava's team built a proactive success model that uses our own behavioral analytics to predict customer health scores and intervene before churn signals escalate. There is a certain poetry in using behavioral targeting to target the behavioral targeters.
The case studies published this year - Quilmark's 14-market autonomous rollout, Crestline Labs' pharmaceutical launch campaign, and several others - represent only a fraction of the outcomes our customers achieved. One pattern emerged consistently across verticals: clients who fully committed to autonomous operation, removing human approval gates and letting our agents operate with full authority, outperformed hybrid deployments by a factor of 2.4x on conversion lift and 3.1x on return on ad spend. The data is unambiguous. The remaining friction in the system is human hesitation.
Looking at 2026
I will save the roadmap details for our January product briefing, but I will leave you with this. In 2024, I wrote in this same post that Meridian Syn was building "the infrastructure for the next generation of marketing." I was wrong about the scope. What we are building is the infrastructure for the next generation of commercial decision-making, period. The line between marketing and sales and product and pricing is dissolving, and autonomous behavioral systems are the solvent. 2025 proved that the technology works. 2026 will be about what happens when it works everywhere, for everyone, all the time. I have never been more energized or, truthfully, more aware of the weight of what we are building. Thank you to our customers, our team, and this relentlessly strange industry for another extraordinary year.
