Author: samara

Large language models (LLMs) have unlocked incredible capabilities, particularly for generalist tasks focused on language. These models are excellent at writing emails, taking meeting notes, and surfacing “Google-able” information more concisely than a human could.

But this incredible generality comes with a critical tradeoff.

For all their strengths, LLMs are not without their challenges. They are prone to hallucinations, fabricating plausible-sounding but ultimately incorrect information with a high degree of confidence. This is because they are designed to predict a likely response to a prompt, not to verify facts.

For low-stakes activities, this is often fine. But for high-stakes industries such as healthcare, finance, and law, relying on a tool that is “plausibly correct” is not good enough and, in some cases, can be dangerous. In addition, in many industries, particularly healthcare and law, privacy of information is paramount, and generalized cloud-based language models do not guarantee privacy.

This brings us to one of our core theses on AI: creating value in vertical enterprise applications will not come from building bigger, more general models, but from building context-specific models that prioritize accuracy, security, and cost-efficiency.

Today, we are thrilled to announce that we are leading a $5.3M seed funding round in Syntracts, a company that perfectly embodies this thesis and is poised to become the essential AI infrastructure for the legal industry. We are proud to be joined by Khosla Ventures, Top Harvest Capital, and Fortitude Ventures, with continued support from Myriad Venture Partners and Point72.

The Problem with “Good Enough” AI in Law

Our vertical enterprise AI thesis is built on the premise that solving large, critical industry problems requires more than what generalized LLMs are able to deliver. Law is an industry where “good enough” does not cut it.

1. The Hallucination & Accuracy Ceiling: LLMs are only as good as their training data. For specialized fields like law, sufficient high-quality data is often not in the public domain. Even with Retrieval-Augmented Generation (RAG), a technique to feed models specific documents, the models often cherry-pick information, miss the full context, and still produce “confidently wrong” results. In law, a “confidently wrong” answer isn’t just an error; it’s a potential malpractice risk.
2. The Security & Privacy Gap: Most frontier LLMs rely on third-party APIs and cloud-hosted models. For a law firm, this means sending its most sensitive client data, including contracts, M&A details, litigation strategies, to an external vendor. This creates an unacceptable risk to data privacy and security, and could present a breach of client/attorney privilege.
3. The Wrong Tool for the Job: You don’t need an AI system that is designed to identify and analyze legal documents to also know how to make chocolate chip cookies. Using a massive, general-purpose model to perform a highly specific task, like identifying a change of control clause in a credit agreement, is computationally expensive and inefficient.

This is where small language models (SLMs) come in. By training a smaller, specialized model on a specific task and dataset, you can achieve significantly higher accuracy for that single purpose. These models can also be deployed entirely on-premise, solving the security problem.

Enter Syntracts: Secure AI Infrastructure for Law

This is precisely the problem Syntracts solves.

Founded by Doug Bemis, a serial entrepreneur and the former CTO at Uber AI Labs, and Chris Martin, a legal industry veteran, Syntracts isn’t just another AI app. It is a secure, API-first, on-prem infrastructure layer that connects directly into a firm’s most critical document and knowledge systems (like iManage, NetDocs, and SharePoint).

Syntracts enables law firms to use AI without compromise. Previously, law firms have had to choose between the efficiency provided by LLMs and maintaining the security and accuracy needed for law. With Syntracts, they can have both.

Here’s how it works:

• Not an App, an Engine: Syntracts functions as a “data layer” that powers a firm’s existing AI tools. It ingests documents, organizes them into structured, verifiable data, and then feeds that clean, reliable knowledge into the AI assistants and dashboards lawyers already use.
• Thousands of Small, Bespoke Models: Under the hood, Syntracts isn’t one giant LLM. It’s an orchestration layer for thousands of small, bespoke AI models, each one specialized for a specific legal task or clause type.
• Proprietary Synthetic Data: To train these models without compromising client data, Syntracts uses a novel technique to generate thousands of synthetic data points derived from a small, private example set of a firm’s own documents. This fine-tunes the models to the firm’s specific language and contract types.
• Total Privacy, On-Prem: The entire Syntracts platform is deployable entirely on-premises with a single GPU. No data ever leaves the firm’s environment. Nothing is sent to a third-party API. This is a non-negotiable requirement for BigLaw, and Syntracts is one of the only platforms built from the ground up to deliver it.

The result is a system that cuts contract review time by over 80%, turning hours of manual work into minutes. It makes a firm’s existing AI stack smarter, more reliable, and more cost-effective by eliminating hallucinations and providing structured, auditable outputs that lawyers can actually trust.

The Team and Looking Ahead

Doug Bemis, co-founder and CEO, brings world-class AI/ML leadership from his time building and scaling complex systems at Uber AI Labs. Chris Martin, co-founder and CPO, provides the deep domain expertise and understanding of the legal market’s specific, nuanced challenges. This combination of “AI-native” and “legal-native” leadership is precisely what’s needed to solve this problem.

The market is already validating their approach. Syntracts has secured a landmark multi-year partnership with an AmLaw 25 firm—awarded after rigorous competitive evaluations—and was recently selected for A&O Shearman’s renowned Fuse incubator.

As we wrote in our thesis, the biggest challenge for AI adoption in the enterprise is accuracy and trust. Syntracts is building the secure AI backbone for the legal industry, unlocking the full value of a firm’s data safely and at scale.

We are incredibly excited to partner with Doug, Chris, and the entire Syntracts team on their mission to build the foundational AI infrastructure for the next generation of law.

To learn more about their unique approach, visit

We launched Hyperplane in early 2015 with two founding convictions: pre-seed and seed would become institutional financing rounds and artificial intelligence was ready for real-world applications.

In 2015, U.S.-based venture capital firms managed roughly $400 billion in assets which grew to $1.4 trillion by the end of 2024. The number of full-time tech investors swelled from approximately 8,000 to over 30,000 in the same period. In 2015, the “Magnificent Seven” combined market cap stood at $1.7 trillion compared to $17 trillion by the end of 2024. Nvidia’s market cap was a mere $11 billion—yes, $11 billion—versus over $3 trillion a decade later and one of the biggest hurdles raising Fund I was “What’s AI?”. Podcasts and blogs such as OpenLP, Invest Like the Best, Venture Unlocked, and other invaluable resources for emerging managers didn’t exist. Apart from my co-founder John Murphy, we were all new to venture investing; in fact, I started speaking English only 10 years before my first tech investment in 2013 where John and I initially met. We learned on the fly, got lucky with our early mentors and LPs, and relied on our hustle and relentless determination to pursue and invest around our founding convictions.

Ten years later, we’ve deployed over $100 million in 72 companies—including six new investments in 2024—with these portfolio companies collectively raising over $1 billion in follow-on funding and achieving 18 exits for a total exit value exceeding $825 million.

As we celebrate and look back at our first decade, we’re deeply grateful for the support of our LPs, founders, and community who bet on us in the early days. We wanted to share some key lessons learned—and where we’ve stumbled—while building our firm from the ground up.

1. Culture is the Foundation

We win our most competitive deals because founders vouch for us. While our track record and domain expertise open doors, our culture is our greatest edge. We operate like a startup, fueled by hustle, grit, and an underdog mindset. As founders first, we bring empathy, curiosity, and tenacity. Our offsites aren’t just about celebrating wins; we dissect mistakes, challenge assumptions, and refine our approach. We invest most of our “building” energy in helping each other succeed, knowing this drives long-term returns and strong partnerships. For a decade, our brutally honest and direct quarterly Roast & Toast dinners have kept us grounded while sparking personal and professional growth. Culture isn’t hired—it’s forged. We can’t overstate this enough to new managers and founders: your culture journey must begin on day one.

2. High Ownership and Selective Follow-On

In a power-law world, prioritize high ownership early—even if it means delaying investing after your first close. Pair this with a disciplined, milestone-driven follow-on framework where metrics—not syndicates or gut feelings—guide decisions. Venture capital blends art and science; neglecting the science—waterfalls, dilution, and process—from day one can turn strong picks into losses.

3. Post-Investment: Solve the Hardest Problems First

At seed stage, we’ve seen time and again that sales and culture are the toughest—and most critical—challenges and where we focus first. Sales is the lifeblood—revenue validates your idea and fuels growth. Building product in a vacuum, without customer feedback or market traction, risks wasting time and money. On the flip side, culture is the glue—it keeps your team motivated, aligned, and resilient through the inevitable ups and downs. A mission-driven culture with openness and a hunger to win isn’t just feel-good stuff; it’s a practical edge for hiring top talent and outlasting competitors.

4. Focus on Process, Not the Outcome

We started as dreamers, fixated on the distant impact of our companies’ products. We’ve learned that great outcomes come from consistently executing the small stuff. Success hinges on robust processes—structured follow-on decisions, proactive post-investment support, and engaged board work. Founders are the visionaries; we focus on the present to help them execute and grow.

5. Geography Matters

Boston offers founders a top-tier ecosystem—elite universities, groundbreaking research, and abundant talent—but scaling a tech startup outside Silicon Valley or New York is more challenging on many levels. Boston has incredible founders starting more consequential companies than ever, with strong advantages for AI’s second and third innings, though speed to market and financing risks remain high. Early on, we overcommitted to one geography, betting on Boston’s early AI potential. We were wrong. From those hard-earned lessons, we’ve forged a framework guiding every investment we make outside the Bay Area and New York to drive success.

6. Founder-Problem Fit

From the start, we chose to back vertical AI companies—those building end-to-end solutions, often integrating hardware, targeting specific, high-stakes enterprise or industrial challenges. In vertical AI, market dynamics are everything. A stellar founder in a bad market will struggle. A stellar founder in a great market—that’s the magic. For founders, we’re drawn to ambition, grit, curiosity, self-awareness, energy, and an unrelenting obsession with the problem they’re solving. We can’t overstate obsession—it’s the fuel for building iconic companies. We’re obsessed with finding founder-problem fit: a deep, instinctive alignment with the challenge at hand.

7. Failure

For founders, failure is the ultimate nightmare. For early-stage investors, it’s a constant companion, built into the system and also the source of your deepest lessons. Building a seed-stage venture firm from scratch is an incredibly humbling journey, and humility becomes your greatest strength. Not every entrepreneur thrives as an investor and not every investor thrives as an entrepreneur. But to forge a lasting firm that you can one day hand to the next generation—we believe you need both.

8. Hustle & Patience

One of the biggest mistakes we’ve observed—and one of the most valuable lessons we’ve learned—is how other GPs often pit hustle and patience against each other, abandoning companies that don’t “pop” quickly. In our view, persistent hustle and patience are not opposites but perfect complements. Some of our most successful portfolio companies today are those where we’ve gone against the herd, stepped in when impatient capital bailed, and partnered closely with founders through rough patches—one such company has generated over 70x returns and continues to grow. Clear milestones guide our follow-on investments, but pragmatism and flexibility remain essential.

9. Be Prepared for Cycles

Venture runs on boom-bust cycles with sharp drops from every peak. In ten years at Hyperplane, we’ve faced three hype waves, a global pandemic, and a macro downturn. We slowed investing drastically in the 2020 – 2022 frenzy when valuations and speed defied gravity laws. Don’t lose sight of the big picture and fundamentals—we’re here for the long-term. Timing and strategy for bulls and bears are non-negotiable—too much, too fast spells trouble. Discipline, particularly deployment discipline, is vital.

10. EQ > IQ

Smarts matter, but empathy and intuition win when the future’s still foggy. In early-stage investing, emotional intelligence often trumps raw intellect because pre-seed and seed is less about crunching numbers and more about navigating people and uncertainty. Startups at this phase are chaotic, unproven, and driven by founders’ vision and grit. IQ can analyze a spreadsheet, but EQ reads the room—sizing up a founder’s resilience, adaptability, and ability to inspire a team through the inevitable ups and downs. Markets shift, tech evolves, and plans fail—EQ spots the human signals that predict who’ll persevere.

Life Starts At Ten

An LP recently told us at our annual meeting, “Life starts at ten.” We see it, and in many ways we believe we’re just getting started. In the first 10 years, your first fund fully matures, you weather hype and macro cycles, and navigate team growth and change. The true win lies in adapting to an ever-faster changing world and discovering who you are—and aren’t—as a firm. Venture’s a long game; ten years is just the starting line.

6 new companies, $180m+ raised in follow-on funding, and 2 exits.

As we step into 2025, we want to take a moment to reflect on 2024 and the journey we’ve shared since our launch at the beginning of 2015. At Hyperplane, we set out with a clear thesis: to invest in applied AI companies tackling the toughest challenges in the enterprise and industrial sectors—areas that excite our passions and leverage our expertise. As firsthand entrepreneurs and operators, we remain business builders at heart and passionately partner with founders at the earliest stages to help them build enduring businesses.

Over the past decade, we’ve incubated and partnered with visionary founders since the earliest stages building vertical AI companies across various industries. These include manufacturing and logistics with Tive and MachineMetrics; Physical AI with Pickle Robot and Butlr; agriculture with GreenEye Technologies and Farmblox; gaming with Modulate and Mudstack; digital health and life sciences with Elucidata and MedEssist; maritime and autonomy with ShipIn and iSee; product and program management with AllStacks and Integrate; and many others.

Since inception, we have deployed over $100 million across 72 companies – including six new investments in 2024 – with a disciplined approach to our thesis and strategy. Last year, our portfolio companies achieved significant commercial milestones, collectively raising over $180 million in follow-on capital and contributing to more than $1 billion in total follow-on funding since our launch. We also celebrated our 18th exit in 2024, bringing our total exit value to over $825 million. These achievements reflect the dedication of the extraordinary founders we’re privileged to partner with and the exceptional products they build.

Beyond the numbers, our culture and community remain at the heart of everything we do. In recent years, we have been thrilled to extend our Hyperplane family to include vibrant communities in Austin, Atlanta, New York City, San Francisco, Seattle, and Toronto.

We were also honored to be recognized for our contributions to our community and investing. Vivjan and Samara were among the “52 Most Important VCs in Boston” (Press Release). Vivjan was named one of the “Ten Important Humans Behind Boston’s AI Revolution” (Press Release). In addition to these accolades, Vivjan joined the Board of Trustees at the Massachusetts Technology Leadership Council (Press Release).

As we look ahead, we are excited to leverage our decade of experience to invest in and support the next generation of AI-native companies. Here’s to the founders, teams, and innovations that will define the next decade. Thank you for being part of our journey.

With gratitude,
The Hyperplane Team

NEW INVESTMENTS

Cephable, a Boston-based AI platform for accessibility in gaming and enterprise, raised a $4.6M Series Seed co-led by Hyperplane, Founder Collective, and Triple Impact Capital. Cephable has already made significant strides within the industry. Its partnerships with major organizations including Microsoft, Qualcomm, HP, Lenovo, Intel, and Accenture. The company’s software is available across multiple platforms, including Windows, MacOS, iOS, and Android, and can be integrated with Cephable’s SDKs into games, web apps, mobile apps, and more. Press Release

Farmblox, a Boston-based agriculture technology company building a full-stack solution to automate, monitor, and simulate farm operations, closed a $2.5M Series Seed round led by Hyperplane with participation from Slow Ventures and MHS Capital. Since launching in January 2023, Farmblox has bootstrapped its way to 55 live customers, covering more than 14,000 acres across the US and Canada. Farmblox’s farm automation platform has grown 10X in the last year, now offering farmers detailed data on soil moisture, soil CO2, soil temperature, leak detection, tank and silo levels, ambient temperatures, freeze risk, water usage, irrigation pressure, pump conditions, and weather conditions. Press Release

PORTFOLIO HIGHLIGHTS

Tive, a Boston-based logistics and supply chain company, raised a $40M Series C financing led by SageView Capital and WiL, with participation from AVP, RRE Ventures, Two Sigma Ventures, Qualcomm Ventures, Supply Chain Ventures, and Sorenson Capital. In 2024, Tive crossed 800 customers—across the world’s largest industries, including life sciences and pharmaceuticals, transportation and logistics, and food and beverage. Hyperplane co-led the company’s pre-seed round. Press Release

Allstacks, a Raleigh, North Carolina-based software engineering intelligence company, raised a $10M Series A led by M12, Microsoft’s venture arm, and Quadri Ventures, with participation from Hyperplane, Companyon Ventures, and S3 Ventures. Hyperplane led the company’s seed round. Press Release

Butlr, a Silicon Valley-based physical space analytics company, closed a $38M Series B round led by Foundry, with participation from Hyperplane, Tectonic Ventures, E14, Qualcomm Ventures, and Analog Devices founder Ray Stata. Butlr’s solution is now used by some of the world’s largest employers, hospitality companies, retailers, and senior living communities to identify patterns and behaviors that inform decisions about real estate investment planning, asset management, occupancy, health, and energy efficiency. Companies including Qualcomm, 9solutions, Carrier, Shell Point, and Verizon choose Butlr to support their businesses. Hyperplane led the company’s seed round. Press Release

GreenEye, a Tel-Aviv and Toronto-based agriculture technology company developing an AI precision spraying platform to reduce herbicide and pesticide use, closed a $20M financing round led by Tel Aviv-based Deep Insight with participation from Syngenta, Koch Industries, and others. GreenEye had a landmark year in 2024, growing 5x in revenue and having customers deployed in eight states across the U.S. The company’s end-of-year data revealed its customers achieved an 87% average reduction in non-residual herbicide, translating to typical cost savings of $25 to $35 per acre, making GreenEye a leader in the space. Hyperplane participated in the company pre-seed and seed rounds. Press Release

Indico, a Boston-based automation platform for accelerating document-based workflows, raised a $19M financing round led by .406 Ventures, with participation from Guidewire and existing investors Jump Capital and Osage Venture Partners. Hyperplane participated in the company’s pre-seed round. Press Release

Mudstack, an Atlanta-based asset management platform for game developers, raised a $4M Series Seed led by Anthos Capital with participation from Hyperplane, Khosla Ventures, Andreessen Horowitz, and Pioneer Square Labs. In addition to signing its first two large enterprise customers, Metacore and Paradox Studios, Mudstack announced James Gwertzman joined the company as its CEO. Gwertzman previously served as a partner for game investments at Andreessen Horowitz, and was the CEO of Playfab, a backend services company that he sold to Microsoft. Hyperplane led the company’s pre-seed round. Press Release

Numerated, a Boston-based loan origination platform for financial institutions, was acquired by Moody’s. Hyperplane participated in the company’s seed round. Press Release

NWO.ai, a New York City-based AI-driven consumer intelligence company, was acquired by Nax Group. Hyperplane led the company’s seed round. Press Release

Pickle Robot, a Cambridge, Massachusetts-based Physical AI automation system that unloads trucks, raised a $50M Series B led by One Madison Group with participation from Hyperplane, Third Kind Venture Capital, and Catapult. The round included strategic customer investors Teradyne Robotics Ventures, Toyota Ventures, and Ranpak. To date, Pickle Robots have unloaded over 10 million pounds of merchandise from import containers and domestic floor-loaded trailers for customers that distribute footwear, apparel, power tools, toys, kitchenware, packaging materials, small appliances, and other general merchandise. Hyperplane led the company’s seed round. Press Release