The global poultry market is worth $315B and sits at the intersection of several critical systems that impact all of us. When disease strikes a poultry operation, the effects ripple through feed integrators, commodity markets, food prices, and ultimately to consumers at the grocery store.

The ongoing bird flu crisis demonstrates this interconnection. The U.S. economy has lost over $14 billion to avian influenza in 2024–2025 alone. These aren’t just numbers on a balance sheet. They represent higher egg prices, disrupted supply chains, and culled flocks.

But the challenge extends beyond economics. Our food supply chain is a matter of national security. Disease surveillance in livestock isn’t just about farm profitability. It’s about ensuring stable food supplies, preventing zoonotic disease transmission to humans, and maintaining the infrastructure that feeds our population.

Climate change is accelerating these challenges. Shifting weather patterns are creating new disease vectors and expanding the range of pathogens that threaten livestock. What used to be regional problems are becoming global ones. The increasing frequency of disease outbreaks isn’t a coincidence. It’s a predictable consequence of environmental changes that favor pathogen spread.

This creates costs that cascade through the entire ecosystem. Farmers face losses from sick or culled animals. Feed suppliers lose customers. Processors see reduced capacity. Commodity traders face volatility. Consumers pay more at the store. Everyone in the chain absorbs some portion of the damage.

Better biosurveillance can change this equation. Early detection means early intervention. Early intervention means smaller outbreaks. Smaller outbreaks mean lower costs across the board.

Metagenomic sequencing is the technology that makes Barnwell’s approach possible. Unlike traditional diagnostic tests that look for one specific pathogen, metagenomic sequencing captures all the genetic material in a sample. You get a complete picture of the microbial ecosystem.

In 2001, sequencing a human genome cost over $100 million. By 2014, Illumina brought that down to $1,000 with the HiSeqX Ten. In 2022, they introduced NovaSeq X, targeting $200 per genome. Companies like Ultima Genomics are pushing toward $100 per genome. A 99.9% cost reduction in two decades is the cost curve decline that we look for, much like software, to unleash a new wave of innovation.

The cost decline came from several breakthroughs. Next-generation sequencing platforms replaced Sanger sequencing. Competition among equipment manufacturers drove innovation. Companies like Illumina, Oxford Nanopore, and others kept pushing boundaries. Reagent costs fell with scale. Computational tools improved, making analysis faster and cheaper.

Here’s what the cost curve looks like:

The cost of sequencing a human genome has fallen from $100 million in 2001 to under $100 today, with projections suggesting sub-$50 costs by 2027. This 99.9%+ cost reduction over two decades has unlocked entirely new applications, from precision medicine to environmental biosurveillance.

This same pattern has driven every major technology platform. Cloud computing costs dropped exponentially, enabling the software-as-a-service revolution. Solar panel costs fell 90%, making renewable energy economically viable. DNA synthesis costs declined 1000x, creating the synthetic biology industry.

When fundamental technology costs drop this much, new applications become possible. Things that were too expensive to consider suddenly make economic sense. Markets that didn’t exist get created.

This cost curve is why Barnwell Bio can exist today and couldn’t have existed ten years ago. Running metagenomic sequencing on environmental samples from barns would have been prohibitively expensive. The economics didn’t work.

Now they do. Barnwell can sequence the complete microbial ecosystem of a poultry barn, process hundreds of thousands of data points, and deliver actionable insights to producers at a price point that makes sense for their operations.

The technology isn’t just cheaper. It’s better. Modern sequencing platforms generate more data, with higher accuracy, in less time. The bioinformatics tools to analyze that data have improved dramatically. Machine learning models can identify patterns that would be invisible to traditional analysis.

This is the same pattern we saw with cloud computing and SaaS. When Amazon launched AWS, it wasn’t just about renting servers. It was about making computing infrastructure so cheap and accessible that new business models became viable. Companies that couldn’t afford their own data centers could now scale globally. Software startups could launch without capital expenditure.

The sequencing cost curve creates a similar shift. Barnwell can offer continuous environmental monitoring instead of point-in-time testing. They can detect threats weeks before clinical symptoms appear. They can give producers visibility into the microbial ecosystem that shapes their operation’s health and performance.

The team’s experience scaling national wastewater monitoring for COVID-19 gave them the operational playbook. They know how to collect samples, process them at scale, and deliver insights that stakeholders can act on. They learned what works in biosurveillance when speed and accuracy both matter.

Now they’re applying that knowledge to an industry that desperately needs it. Poultry producers are caught between regulatory pressure to eliminate routine antibiotics and increasing disease threats. Barnwell’s platform gives them a tool to maintain biosecurity without sacrificing productivity.

Barnwell Bio was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

Arya raises $18M Series A to advance AI agents for post-acute care.

One year ago, we led Arya Health’s $4 million seed round because we saw an intersection of three major forces reshaping healthcare operations: the ongoing labor crisis, the lack of intelligent infrastructure in post-acute and home-based care, and the maturation of AI-native enterprise software. With Arya’s new $18.2 million Series A announcement, that conviction has only deepened — and the broader market is now catching up.

When we first met the team, Arya was tackling one of the most acute pain points in healthcare: managing a large, fragmented, and overburdened workforce. Providers across skilled nursing, home health, and behavioral health were struggling with staffing shortages, burnout, compliance requirements, and razor-thin margins. In a system employing more than 17 million people, even small efficiency gains translate to enormous impact.

What stood out immediately was Arya’s approach. Rather than building a generic HR or workforce management tool, Arya built an AI-native operations layer for healthcare services — an adaptive system that automates scheduling, onboarding, credentialing, and compliance, while learning from each organization’s unique workflows. The product promised not just incremental efficiency, but structural change: fewer administrative bottlenecks, better shift coverage, and ultimately higher patient-care continuity.

Since our investment, Arya’s progress has been striking. The company reports a six-fold increase in revenue, with customers seeing scheduling improvements of over 40% and faster fill rates of more than 50%. Those metrics translate to real economic outcomes for care providers — reducing agency costs, improving utilization, and strengthening retention in a sector where every percentage point matters.

The market backdrop for Arya’s platform is enormous. The U.S. post-acute care market alone is approaching $500 billion this year and expected to nearly double over the next decade. Beneath that, the healthcare workforce management software segment — the layer Arya is automating — is growing from roughly $2 billion in 2023 to over $4 billion by 2030, expanding at more than 10% annually. And because Arya’s architecture is modular and extensible, the company can move horizontally into other labor-intensive verticals within healthcare over time.

The new Series A validates this thesis. It provides capital to deepen product capabilities, expand into adjacent markets, and scale go-to-market teams. More importantly, it signals a growing recognition that healthcare operations can no longer be managed through manual workflows and fragmented legacy systems. Automation isn’t a luxury — it’s a necessity.

As seed investors, our belief was that Arya could become the operating system for healthcare labor. That conviction looks stronger than ever. The company is executing against a massive, structural opportunity with measurable ROI, an exceptional founding team, and strong early momentum. If Arya continues on its current trajectory, it won’t just build a valuable company — it will redefine how healthcare organizations function at their core.

We’re proud to have backed Arya early, and we’re even more excited for what comes next.

Arya raises $18M Series A to advance AI agents for post-acute care. was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

Nome

We are excited to announce Twelve Below’s recent investment in Nome Bio, an AI-powered platform enabling rare disease patients to develop their own drugs.

Rare diseases affect 1 in 10 Americans and account for over $400 billion in annual medical costs. Although many experts now believe that treatments exist for a meaningful share of these conditions, 95% of rare diseases still lack an FDA-approved therapy. Pharmaceutical companies are rarely incentivized to pursue treatments for such small patient populations, leaving millions of families with a diagnosable disease but no available treatment.

In response, some patients have begun taking matters into their own hands by independently developing drugs to treat themselves. Since 2018, 32 custom drugs have been developed by patients with the help of the FDA’s Expanded Access Pathway, which allows patients to legally develop their own therapies and bypass traditional clinical trials. Yet this DIY process remains fragmented, expensive, and prohibitively complex: patients must coordinate across researchers, manufacturers, consultants, and the FDA. Most families don’t even know this path exists.

Nome is creating a platform for rare disease patients to develop their own drugs. Using AI, Nome evaluates a patient’s genetic data to determine whether their condition is suitable for custom drug development, then coordinates the complex network of researchers, labs, manufacturers, and regulators required to move a treatment from concept to clinic. We believe Nome’s approach represents a transformative new infrastructure layer for rare disease drug development — and personalized medicine at large.

When we met Nome’s founder, Steven Ringel, it was immediately clear that solving this problem is his life’s mission. After being diagnosed with an ultra-rare KIZ gene mutation that causes progressive vision loss, Steven founded a nonprofit to develop a custom gene therapy for his own condition. That insight sparked Nome and its vision to make personalized drug development accessible to every rare disease patient.

Across his decade in genomics and precision medicine — spanning Bain, Illumina, Sema4, and Tempus — Steven has been thinking about how to solve the rare disease problem. With the convergence of key regulatory, technological, and scientific breakthroughs over the past year, Steven set out to build the company that had been chasing him his entire career.

The company’s name honors the 1925 serum run to Nome, Alaska, when dog-sled teams raced antitoxin across frozen tundra to save children in crisis. Nome Bio carries that same spirit of urgency and relentlessness — applying it to the modern challenge of building infrastructure that can deliver life-saving, personalized therapies at scale.

Twelve Below is proud to have led Nome’s $2.7 million Seed round, alongside Refactor Capital and Inflect Health. We’re thrilled to partner with Steven and his exceptional team as they build the infrastructure to make rare disease drug development faster, cheaper, and more accessible than ever before.

Nome Bio was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

Dispatch announced the close of an $18 million Series A funding round led by Brewer Lane Ventures, following earlier investments led by Twelve Below and F-Prime.

A Fragmented Industry

The wealth management industry has long struggled with a fundamental problem: disconnected systems that can’t share data. Advisory firms juggle separate platforms for client onboarding, portfolio management, billing and more. This fragmentation creates repetitive manual tasks, error-prone reconciliation and blind trust in AI models fed with inconsistent data.

Wealth advisors relying on these tools spend less time with clients and more time on administrative work. The problem has worsened as demands for personalization and specialization lead firms to adopt more software solutions that produce still more incompatible data.

Solution for Wealth Managers

Dispatch was built to solve this problem. Its platform automates client onboarding, syncs data in real-time across every major system and feeds AI-driven workflows with clean, normalized information.

Instead of forcing advisors to connect, map and maintain multiple systems manually, Dispatch handles bi-directional integrations across a firm’s entire tech stack. Its proprietary Form Builder completes every field on custodial forms automatically, while its integrations reconcile thousands of data points across CRM, billing, reporting, document storage and custodial platforms.

Series A and Market Momentum

This month, Dispatch announced the close of an $18 million Series A funding round led by Brewer Lane Ventures, bringing the company’s total capital raised to $30 million. Twelve Below, which led Dispatch’s pre-seed round in 2022, sees this milestone as proof that the financial services industry needs foundational data infrastructure.

The Series A round included participation from New York Life Ventures, MassMutual Ventures, Perceptive Ventures, F-Prime, Flyover Capital, Fika Ventures and CoFound.

Over the past year Dispatch has also secured partnerships with several prominent wealth management firms — including Mariner, Sanctuary Wealth and Choreo — that collectively oversee more than $1 trillion in assets under advisement. The platform has helped firms save thousands of hours in manual workflows while cutting error rates by over 90%.

Our Early Conviction

When we first met co-founders Rob Nance and Madalyn Armijo, their vision was clear: don’t replace human advisors, amplify their impact. That clarity came from experience. Rob and Madalyn started their careers in investment management and are intimately familiar with the pain points advisors face. That domain expertise, combined with their entrepreneurial experience in fintech, offered a unique vantage point to modernize the RIA industry.

We recognized early that the RIA market was ripe for infrastructure innovation. With over 30,000 registered advisors managing $114 trillion in assets, the fragmentation was acute: roughly 80% are independent firms managing their own complex tech stacks, with advisors spending nearly half their day on back-office tasks rather than serving clients. Dispatch’s data orchestration layer addresses a fundamental need in a massive, underserved market.

Looking Forward

The $18M Series A is a strong signal that the financial services industry is ready for a critical layer of data integrity, orchestration and automation.

The funding will accelerate Dispatch’s investment in agentic workflows and AI-driven data orchestration while expanding integration coverage to support every major platform in the advisory tech stack.

The companies that address fundamental challenges — not just build new applications — will shape the next generation of financial services. As advisory firms navigate rising client expectations and accelerating technological change, data orchestration will determine who thrives and who falls behind.

Dispatch announced the close of an $18 million Series A funding round led by Brewer Lane Ventures… was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

In today’s AI market, data remains the key bottleneck to unlocking new products and value propositions.

The largest repository of untapped data lives in the real world, where hardware devices — like cameras and sensors — are required to capture what’s happening on the ground. But our hardware infrastructure was designed before the release of ChatGPT, when cameras and sensors were installed to meet security and compliance needs, not to collect data for training AI models.

As a result, in many markets, valuable data for AI goes completely unrecorded, simply because no one has ever thought to put a camera, sensor, or any other device in a certain place. Until GPT 3.5, there was really no reason to.

Now, startups have a brief window to be the first to deploy low-cost, off the shelf hardware in these locations and unlock new data, and applications, for AI.

What might look like a clever growth hack is, in my view, a sign of new market creation. Just as the iPhone expanded the TAM of mobile applications from a niche community of hobbyists to basically anyone with a phone, newly deployed cameras and sensors will enable AI applications to serve exponentially more use cases across the physical world.

To date, every major inflection point in the AI application layer has followed a breakthrough at the infrastructure layer. Nvidia GPUs enabled the first wave of AI apps. I believe cheap, accessible hardware will drive the next wave of rapid growth in AI products.

I’m looking for founders building AI applications by installing hardware in the right places before anyone else does. These teams are best positioned to be first-movers in entirely new markets and build businesses with enduring, defensible data moats. While this playbook is still early, a few companies are already showing what’s possible:

• Olden Labs automates manual data collection in animal studies using sensor-equipped lids that are deployed on top of animal cages in labs.
• Sahay AI augments manual rail inspections with train-mounted sensors that enable real-time rail monitoring and predictive maintenance.
• SmartBarrel streamlines time tracking on construction sites using camera-equipped time clocks with facial recognition to log workers in and out.
• Soil Action reduces fertilizer overapplication via sensors mounted onto tractors that assess soil conditions and calibrate application in real time.

The case for deploying off the shelf hardware

Companies that leverage off the shelf hardware to build AI applications have an opportunity to build defensible businesses with SaaS-like economics and capital efficiency.

While off the shelf hardware is a commodity and not defensible on its own, the proprietary data it captures can build a powerful moat. In the right use cases, this data is inaccessible without on-site hardware, essential to building a functional AI product, and offers compounding advantages with scale as more data leads to better model performance.

That said, hardware is undoubtedly a more challenging path to market than pure software. As I see it, in addition to building a moat via proprietary data capture, deploying hardware can pay off in two other major ways that relate to distribution:

1. Hardware is incredibly sticky. Hardware must be physically installed — and, therefore, uninstalled — for customers to churn. This creates a painful rip and replace dynamic for any competitor that wants to unseat a first-mover.
2. Hardware can confer exclusivity. In some cases, being first to install means blocking competitors from entering the same physical space. By securing a strategically valuable installation point, a company can own the data control point for an entire ecosystem of products.

If the goal of leveraging hardware is to secure unprecedented levels of data and distribution, off the shelf hardware stands out as an elegant solution. Off the shelf hardware is cheap, easy to scale and deploy, and carries minimal risk across R&D and supply chain. In practice, this means that founders can quickly scale their physical footprint and data collection, and focus their energy on utilizing this asset to build high-margin AI products.

It’s worth noting that companies that employ this strategy will operate (and be valued) like software businesses at scale. Their unit economics will look more like SaaS companies, with almost all revenue coming from high-margin, recurring, AI-powered software products. In hindsight, hardware might go down as the one of the best wedge products for building a huge software business.

As this is a relatively new model — enabled by the maturation of both the hardware and LLM ecosystems — I’ve outlined some of its key attributes in a side by side comparison with traditional software and hardware companies.

What I’m looking for:

There isn’t a single definitive checklist of traits that make a market ripe for this model. Each market has its own nuances that make it more or less likely for this playbook to resonate.

However, I do think that there are a few traits that could signal an opportunity for hardware-enabled AI products. While I’m generally on the lookout for companies at this intersection of hardware and software, here are a few specific things I’m particularly excited by. I think that the presence of even one of these traits can make a hardware-enabled AI opportunity worth exploring.

• Markets or environments without much hardware. If a company is the first to deploy hardware in a data-rich environment with no other hardware vendors, that could indicate the emergence of a new market opportunity. Hardware may not have been necessary in that environment before, but in today’s AI market, it could be the key to unlocking meaningful value.
• Markets with data control points: There are spaces that might be more valuable than others to embed hardware. These environments generally see important activity, but are bounded by finite installation space. Take an animal cage in a lab: an enclosure the size of a shoebox contains all the observable data on lab subjects. Or a traffic light: a single pole offers a birds-eye view of every vehicle on the road. These high-leverage installation points have the potential to be the data control point for platform plays in AI.
• Markets where new value propositions can be created: LLMs have unlocked new capabilities for software, and data from hardware may enable the next step-change in the value technology can deliver to legacy industries. If observability and integration with hardware enables AI to accomplish tasks that software could never before, this is another signal of TAM expansion through value creation.
• Markets where a data advantage matters: Some data advantages matter more than others. Some factors to consider when evaluating how meaningful a data advantage is in a certain market is (1) how accessible the data is, (2) whether more data leads to meaningfully better performance, and (3) a market’s sensitivity to better product performance. The classic example of a market where a data advantage matters is self-driving cars — data is hard to procure, a lot of it is needed to train models, and the market cares about having the best performing product. Not every opportunity has to match the data advantage of self-driving cars, but I’m generally looking for spaces where data matters more to a product or market.

If you’re ideating, building, or investing in this model, I’d love to chat! Feel free to reach out to me at Tamar@twelvebelow.co

I want to invest in AI companies leveraging off the shelf hardware was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

We’re thrilled to announce that we’ve partnered with Burhan, Mo, and Allan and the Rondah team as they use AI to reinvent patient engagement in the dental industry. The dental industry continues to both grow and consolidate, and with that means the professionalization of the infrastructure and services that power great patient experiences.

In an industry that still relies heavily on voice for patient coordination, Rondah has built an AI platform that allows dental practice operators to leverage AI powered voice agents to handle inbound and outbound communication for practices, freeing up existing clinical and operational staff to focus on delivering great care (you can try talking to Rondah here). In addition to serving customers in a low error rate manner, while capturing visits they might otherwise have lost, voice AI is a powerful modality to better engage customers across other critical functions such as billing management, treatment plan coordination and insurance & benefit management. We believe advances in memory capability will greatly enrich how patient engagement can transform the clinical operations and experience.

We think the Rondah team is uniquely positioned to build a steep and durable wedge into this market, combining industry experience, strong technical expertise and a unique moment in time where the technology around voice and dialog quality is hitting a threshold that is a 10x experience for consumers. We believe great market timing is critical, and the dental industry has been a steady, secular growth market for decades, and is now going through both consolidation and an accelerated technology adoption period across the entire patient to clinican toolset.

We are excited to co-lead this round with 2048 and support the Rondah team as they scale their team in NYC.

Rondah was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

Every technology shift brings about new changes for both consumers and businesses, sometimes in a symbiotic fashion. I saw this first hand in the early days of Gilt Groupe, one of the first mobile commerce companies that scaled of >$1BN of revenue runrate in just a few years following a launch around the iOS store. As retail shifted from offline to online, the role of voice increased even further as traditionally decentralized support and sales moved into centralized call centers. As mobile and website analytics became more sophisticated, the end to end digital funnel continued to improve towards targeted, personalized marketing. Conversely, voice interactions across all categories have remained a black box, with tooling being purchased to meet the bare minimum as opposed to supercharging the business. Everyone knew there was gold hidden in those calls, but we had no way to mine it at scale.

That memory snapped into focus the moment I met Ethan, the founder of Voiceops. He walked me through his vision for a platform that would capture every customer conversation and automatically distill the most important signals — what’s resonating, what’s broken, what’s turning a prospect into a loyal buyer or sending them packing. Instead of guesswork, Voiceops offers precise, real-time insights that any team can act on immediately. One conversation, and I could see he’d been obsessed with this problem for years. He’s not the kind of founder who started an “AI company” just because it was trendy; he’d been gathering data, learning from early experiments, and waiting for AI capabilities to catch up. Now that the technology is finally there, he’s sprinting at full speed.

At Twelve Below, we have a founder archetype that we look for what we call “specialists”: founders who have been honing a single problem for years, often fueled by an unshakable obsession. Ethan fits that description better than almost anyone I’ve met. He started working on AI call analytics back in 2016 — before the big wave of LLMs and “AI for everything.” He kept at it, gathering data on how calls could be transcribed, structured, and turned into insights. He learned the hard way what did and didn’t work in real-life, high-velocity contexts. That combination of endurance and expertise is what we look for in a specialist founder.

Even more striking is how fast Ethan and the Voiceops team move. I’ve always believed you can spot an exceptional founder by their response times — both in email threads and in building product. Before our seed round was even closed, Ethan came to me and said, “I just rearranged the entire approach to create an unexpected moat, and Viji’s already building it.” I remember thinking: We haven’t even signed the docs yet. That combination of breakneck product velocity and strong execution is special. Lots of founders talk about “shipping fast,” but these folks truly live it.

I also love that Voiceops has a deeply rooted Midwestern ethos in its founding DNA. Ethan hails from Illinois, Todd (Voiceops’ founding engineer) works out of St. Louis, and Viji, their VP of Engineering, spent 15 years living just a short drive from Ethan in Illinois. It’s a blend of midwestern values (resilience, humility, kindness) and east coast work ethic (relentless execution). You can feel that in how they operate: they’re friendly, grounded, and laser-focused on results, all at once.

The significance of voice is only growing, especially in a post-LLM world where hybrid human-AI interactions create new questions around trust, compliance, and personalization. Sometimes you can automate low-level support calls, but for products and experiences that involve real dollars or complex products, it’s still humans talking to humans. That’s when you need a tool that can actually understand customer interactions well enough to show you how your messages are landing, what’s causing friction, and which leads are likely to convert.

At Twelve Below, we’re proud to co-lead Voiceops’ seed round and help them scale aggressively — starting with a brand-new office in Union Square in New York City. From day one, Ethan has demonstrated exactly the kind of product obsession and lightning-fast execution we look for. I’ve learned that in startups, a thousand things can go wrong. To push through, founders need a rare mix of product obsession, grit, curiosity, and the ability to move at light speed. Voiceops has all of these.

We couldn’t be more excited to partner with Ethan, Viji, Todd, and the entire Voiceops crew as they reshape how organizations tap into voice data for immediate, actionable insights. A decade ago, I watched Gilt sprint into hypergrowth, wishing we had a clearer lens on what customers were saying. With Voiceops, that lens is now sharper than ever, and I can’t wait to see how far this team runs with it.

Voiceops was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

We are excited to announce our investment in Solstice Health, an AI-native platform for life sciences commercial teams to streamline the review and distribution of compliant content. Their first product is the industry’s most sophisticated MLR assistant. As regulatory and compliance burdens have grown substantially over the last decade — intersecting with the rise of personalized medicine and content marketing — the delays and inefficiencies across the medical legal review (MLR) and compliance process have skyrocketed. In an industry where a single day’s delay can cost millions in revenue, we believe AI has created a unique opportunity for the entire marketing and commercialization function to be redesigned, enabling more rapid approval, commercialization, and distribution of life sciences products. Ultimately, this means more critical drugs and devices can be matched more quickly with the patients who need them — patients whose lives often depend on this progress.

Partnering with Aris and Yiwen has been a testament to the collision of great talent and emergent technology in a market that is on the cusp of being transformed. They embody the crucial traits we look for in Twelve Below founders: a deeply rooted chip on their shoulder, a blend of technical and operational prowess, and a fearless ambition underpinned by humility. Perhaps most critical and impressive is their rate learning, a defining quality of those founders who are playing an infinite game.

Market

The global pharmaceutical industry, valued at over $1.5 trillion, with ~$700BN in marketing and sales spend per year, is underpinned by stringent regulatory frameworks that govern every aspect of drug commercialization. Marketing materials — from digital ads to clinical brochures — require rigorous Medical Legal Review (MLR) to ensure compliance with FDA guidelines, fair balance in claims, and alignment with approved labeling. The current model involves tremendous human intervention, back and forth, and disjointed product experiences across several platforms (Veeva, IQVIA, Excel, email). In addition to using humans to scale volume and quality where possible, many of the legacy tools comprise of traditional software built in the early 2000s, lacking the AI native capabilities that were only possible in the last 18 months.

Problem

Today, the MLR process is people intensive, time consuming and error prone. For large pharma and biopharma companies, this process is both time-consuming and costly. MLR cycles typically span weeks, involving cross-functional teams to manually verify claims, citations, and safety disclosures. Traditional MLR processes rely on manual checks, legacy software, and fragmented communication, leading to bottlenecks, human error, and costly delays. Out-of-the-box generative AI solutions, while promising, struggle with accuracy (60–70% success rates) and hallucination risks, particularly when parsing complex clinical data or visual content like graphs and figures.

Why Now

We believe high acuity, enterprise grade solutions will be woefully underserved by horizontal LLM platforms given the nuance in language and context for regulated industries like life sciences, and the need for very specific, personalized agents that can understand the nuances of hundreds of individual regulatory requirements for every piece of content..

Aris and Yiwen took a first principles approach by living in the future, working backwards from what best in class processes and content generation would look like for commercial teams, and productizing an AI engine that is purpose built for pharmaceutical marketing teams and life sciences marketing agencies. What makes the current paradigm of AI models so attractive is the ability to bring context from so much unstructured data, from how regulations have evolved over decades, down to how consumers are responding to various messaging around the world across all drug types.

By incorporating external affairs across the lifecycle of a campaign, Solstice can supercharge how life sciences teams go from establishing protocols and compliance with internal and external creative teams, all the way through engagement with the FDA in generating personalized cover letters to meet 2253 requirements.

While we are early in the development of AI agents, one of the most compelling aspects of the LLMs is the ability to develop reasoning and judgment models that when fine tuned, can mimic extremely sophisticated and complex language tasks. This makes it extremely relevant for the life sciences industry where current commercial teams and workflows are often hamstrung by human bandwidth and in many cases errors that cause downstream delays. Sophisticated reasoning capabilities build trust across commercial and regulatory teams while creating high fidelity, accurate recommendations during the MLR process.

Below are some of the example agents that build upon the Solstice AI platform:

FACT CHECK ASSISTANT:

Fact Check Assistant automatically annotates claims against a reference library. It is able to study a knowledge base and use a proprietary claims expansion and adjudication framework to properly annotate and fact-check claims at a 99.3% accuracy.

Automatically annotating claims to prior-approved reference library:

See Reasoning feature allowing teams to understand the reasoning behind each annotation:

REGULATORY ASSISTANT:

Regulatory Assistant is embedded with the knowledge of an MLR veteran, trained to be an expert on FDA CFR 21 guidelines as well as all published FDA guidance. Benchmarked on thousands of approved pieces, it evaluates content to ensure that it maintains a marketing team’s intended messaging while ensuring that it stays compliant with regulatory standards.

Flagging against external regulations:

Programming internal brand rules and business logic:

The results thus far have been spectacular across key dimensions for customers in terms of accuracy & efficiency, tailored compliance and clear ROI with a foundation for global scale.

Unmatched Accuracy & Efficiency

• 99.3% accuracy in identifying and fact-checking claims, outperforming out-of-the-box AI models by 30–40%.
• 74% reduction in review cycle time and 88% fewer agency hours via automated workflows.
• Proprietary features like multi-layer search and context expansion prevent hallucination, ensuring claims are grounded in verified references.

Tailored Compliance Guardrails

• A dynamic knowledge base integrates client-specific rules, prior-approved content, and FDA guidelines (e.g., CFR 21).
• Brand-specific rule engines allow teams to codify requirements (e.g., risk information placement) and auto-flag deviations.
• Holistic summaries of pieces to evaluate subjective measures such as fair balance. For example, a Fair Balance Score (SFB) is provided to quantify risk-benefit alignment, a critical metric for regulators.

Proven ROI and Scalability

• A 74% reduction in review cycle time means teams are able to launch and approve content faster, significantly reducing the time to market.
• Scalable Azure-based architecture supports global deployments, while case management tools enable real-time collaboration across legal, medical, and marketing teams.
• Seamless integration with platforms like Veeva and Aprimo ensures compatibility with existing workflows.

We think only in the last year could the advancements in AI have made the product capabilities possible to reimagine compliance and marketing functions across pharma and biopharma, impacting the speed with which new advancements can come to market and impact the lives of patients around the world. We think Solstice Health is at the beginning of a paradigm shift in rewriting how life sciences teams bring products to market.

Solstice Health was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

I am thrilled to announce that I am joining Twelve Below, an early-stage seed fund based in NYC focused on fintech, healthcare, energy, SMB, and consumer sectors.

When I started working in venture, I could not have anticipated how intensely the ecosystem would evolve over a few short years — or how those changes would shape me as an investor. From pandemic to post-pandemic, ZIRP to market correction, and GPT-3 to GPT-4, I’ve had the unique fortune of launching my career at the crossroads of multiple market inflection points.

One such inflection point exists within venture capital itself. In the past few years, I’ve witnessed the venture capital landscape fundamentally change. As established firms amassed more AUM than ever before, others emerged to offer an alternative product to entrepreneurs — one that reflects a return to an older style of high-conviction, low-volume investing to more intimately intertwine the success of each founder with the success of the fund.

In this light, Twelve Below aligns itself with founders through an operating model focused on building trust with the entrepreneur. This approach is made possible by the intentional size of the team, fund, and portfolio, and the steadfast belief that investing in an entrepreneur’s business is a privilege to be earned. We look to lead or co-lead rounds with a $500K-$2M check and commit to servicing founders in the most formative phase of company building.

After two great years at Red Sea Ventures, I am ecstatic to start my next chapter as an investor with the team at Twelve Below. Thank you to the founders and investors who have supported me along my journey and shaped me into the investor I am today. I am incredibly grateful for every one of you.

Best,

Tamar

Joining Twelve Below was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.

As the nation faces an intensifying mental health crisis, children and adolescents are particularly vulnerable. Fifteen million young people in the United States need mental health support, yet only about 40% receive the care they need. This gap is exacerbated by a significant provider shortage — there are just 8,300 child and adolescent psychiatrists across the country, leading to a median wait time of 50 days for families and substantial out-of-pocket costs for quality care.

While schools have historically been central in delivering mental health support, emerging evidence highlights the effectiveness of collaborative care models in primary care. Families trust pediatricians as their first point of contact, and when mental health services are integrated into regular pediatric visits, parents and children are more likely to engage in treatment. Recognizing the potential of this approach, Twelve Below led Fort Health’s recent $5.5 million Seed 2 funding round, alongside Vanterra, Blue Ventures, The Child-Mind Institute, and True Wealth.

Fort Health stands out as an in-network, fully integrated mental health platform that bridges primary and behavioral healthcare to provide high-quality, accessible care. Today, mental health concerns top the list of reasons for pediatric visits, yet most pediatricians lack specialized training in child psychiatry. Fort Health is actively working with pediatricians to integrate mental health care into primary care, closing this gap and enhancing treatment access.

Fort Health’s technology integrates seamlessly into pediatricians’ workflows, streamlining the referral process and ensuring families are connected with the most appropriate provider. From the start of the care journey, Fort’s team conducts a comprehensive assessment, develops a personalized treatment plan, and continuously updates the child’s pediatrician on clinical progress, treatment notes, and discharge information. We think this level of integrated, coordinated care represents the future of pediatric mental health.

Over the past year, Fort Health has forged partnerships with major healthcare providers, hospital systems, and insurers. Notable collaborations include Advocare, which brings Fort’s services to over 700 pediatricians across New Jersey and Pennsylvania. Through a partnership with Novawell, a Behavioral Health Managed Organization, Fort Health’s collaborative care model will expand to serve thousands of children in Illinois and Texas.

We’re thrilled to partner with CEO Natalie Schneider and her outstanding team, who bring extensive expertise from leadership roles at McKinsey, Elevance, Samsung, Amwell, and McKesson. With a diverse range of experience, Fort Health’s team is well-equipped to redefine pediatric mental health care in America.

At Twelve Below, we’re deeply committed to supporting innovative solutions to America’s mental health challenges. Our investment in Fort Health aligns with our mission to back transformative, impactful solutions. We’re honored to support Fort Health’s journey to create a future where every child has access to the mental health support they need; when they need it.

Fort Health was originally published in Twelve Below on Medium, where people are continuing the conversation by highlighting and responding to this story.