.png)
AI is moving at warpspeed where entire categories are being created and destroyed seemingly overnight. Agents are systematically translating the “knowledge” part of knowledge work into systems and code. For the first time, many locked up $1T+ profit pools are in play.
There’s four meta-trends and tailwinds that we believe will create opportunities for value creation. For each, we’ve described some examples (not exhaustive!) of the types of opportunities that we’re actively looking to invest in.
.png)
Extreme volatility is the new normal
Between covid shutdowns, tariffs, and now AI uncertainty, our economy has been enduring massive shock after massive shock. And if there’s anything the markets hate the most, it’s uncertainty. Add to that new unpredictable costs in token consumption, power usage, and more means it’s just harder and harder to control risks where it matters most. We believe the ability to manage and hedge risks are going to become more important than ever in a rapidly changing world.
Planning software for a usage-based world
SaaS isn’t dead but seat based pricing most definitely is. Up and down the PnL everything that was once “fixed” is becoming variable. Revenue is shifting to be usage or consumption based. On the costs side, AI spend is the fastest-growing and least understood line item on the P&L. We’re still in the “tokenmaxxing” era but play out the tape and could token/inference costs soon overtake G&A as a line item? Headcount planning, budgeting cycles, and even (human) employment offers are going to fundamentally reorient to include the costs of AI and organizations are going to need better ways to understand and plan for their organizations.
AI-native catastrophic risk simulation software
Tail risks are happening more frequently and they’re breaking risk models. We just hit our sixth consecutive year where natural catastrophe losses were $100B+. What most of the traditional actuarial models miss is that they price risks in isolation and rely heavily on backwards looking loss data. The insurance industry needs new modelling software to help them better understand how various risks cascade into one another. Can world models or new AI-architecture help fill that gap?
New ways to isolate emerging factors
It’s starting to feel absurd that the markets close…since the world definitely doesn’t. It’s clear that the markets are clamoring for new ways to get financial exposure to the changes in the world as they happen in real time. Prediction markets are now at $20B+ in volume. The Iran crisis causes oil perps to surge to $48B on hyperliquid over a weekend. We believe the markets are hungry for new ways to get or maintain new exposures and there will be opportunities for new ETFs and trading venues to emerge.
Agents are the new ICP
BCG estimates the agent market will grow at 45% CAGR, and that feels like a tragic underestimate. Smart, autonomous agents are here and they’re quickly being adopted across consumer and business operations and they search, discover, understand, and choose in fundamentally different ways. Our software and our internet were both designed by humans for humans not agents. Solutions here will bridge that gap.
The communication layer for agentic teams
If you think alert fatigue was bad now… imagine what happens when you add human-agent and agent-agent communications into every business workflow. When half the workforce becomes agents, there’s going to need to be a new way to manage and communicate (and it’s definitely not email + chat!) both inter and intra-organizationally.
Implementations need a new system of record
Implementations were already the pain in the ass that killed many a software sale. In the AI-era, a botched implementation means botched outputs which means botched outcomes. And it’s not one and done either. Each deployment needs to continue to iterate and adapt to evolve in tandem with the needs of the customer. The CARR to ARR gap for every AI-Native software company continues to widen and churn for poor ROI tools is also on the rise. This gap can’t be filled with FDEs forever. There’s a role for a new system of record to emerge that starts at implementation and moves to be the source of truth for customer-specific agent deployments.
Solving structural labor shortages
75% of accountants are retirement age. Insurance is facing an upcoming gap of 400K jobs. The average wealth advisor is 51 years old. AI is eating into the $6T of white-collar payroll spend but it’s not all a replacement story. And this doesn’t even consider economics around the world with rapidly aging populations. There are many structural labor gaps emerging, particularly across “old-school” industries where AI can serve as the structural unlock.
No more phone trees please
The voice context on a call is infinitely richer than a chatbox and the technology is roaring past the uncanny valley. We’ve only just scratched the surface of applications of voice technology and what that means for democratizing the “white glove service” that is associated with always being just a call away. Advisory calls, deal closings, complaints & claims – all run on legacy processes, generate zero structured data, and are staffed by a constantly changing roster. We’re looking for the solutions that aren’t afraid to get deep — unlocking vertical specific or even customer specific voice applications across industries.
AI accounting solutions for every industry
In the $650B accounting services market, we believe that the next iteration of AI-accounting will be deeply verticalized. Property accountants spend hours reconciling CAM charges, construction firms track detailed percentage-of-completion and project based costs, and healthcare providers juggling payer mix accounting, while retailers live in tangled webs of deductions, promotions, and rebates. Every vertical with complex value chains, ownership structures, and billing will need their own vertical specific accounting solution (and dare we say…system of record?).
Finance needs new plumbing
The GENIUS Act and MiCA mean for the first time ever we’re starting to have clarity around around stablecoins and digital assets. At the same time, the markets are more global and 24/7 than ever. Total stablecoin volume has surpassed $1.8T up from $668B a year ago. We’ve crossed the chasm regarding institutional interest and comfort in stables, blockchain, and tokenization but most traditional players are held back by 40+ year old infrastructure. The financial stack is being rewritten from the ground up in real time – but there’s a real need to bridge the gap between old and new.
‘Tokenize it’: the US equities edition
There are 185M stock market participants in the US. The global demand is easily 10x that. Foreign holdings of US securities has grown 10x in the last 20 years but this is still largely out of the hands of retail traders. The demand is there. But for billions of people in emerging markets — where crypto adoption is high while equity participation is <10% — the settlement, custody, and compliance infrastructure to actually buy a share of Apple doesn't exist. Add to that private equity firms, private companies, or other real world assets which are going to be opened up to the world.
Back office infrastructure supporting a T+0 world
Equity settlement has gone from T+5 to T+1 over the last 30 years and now, T+0 (i.e. instant settlement) is looking like it’s on the horizon. Faster settlement reduces risks and frictions in the system. The issue: the back office isn’t ready. Clearing, reconciliation, corporate actions, proxy reporting – all still run on batch processes, manual workflows, and systems built decades ago. Broadridge, founded in 1962, alone processes $10T in daily transactions and proxies for 70%+ of US public companies… in the new world, that’s not going to cut it
AI-Native + stablecoin-native treasury management system
Today’s treasury management and ERP systems for a fiat focused world. Focusing on helping customers manage and move their own cash around dozens of bank accounts across the world. However, companies are already leveraging stablecoins for inter-company transfers, sweeping excess cash into tokenized money markets, and paying contractors across borders in real time. Nobody has built the connective tissue for a business running over multiple sets of rails while the last generation of software in this space is still extremely clunky, manual, and expensive. There’s an AI-Native + stablecoin-native solution to be built.
Agentic workflows that backdoor into your new insurance core
Insurers have faced a lose-lose for years: keep creaking legacy cores or endure a ~10 year replacement cycle. When faced with open heart surgery, inertia always won. Now AI agents can instantly automate high-impact workflows that meaningfully drive down expense & loss ratios. Think submission processing, claims intake, and broker portals. We believe there’s an opportunity to start services-as-software that tackles those operational workflows and collects the context and data along the way to earn the way into becoming first a shadow core and then the real core.
What Are We Looking For?
Primary leads pre-seed and seed rounds with $1-10M checks. We back founders who are building from conviction, not consensus.
- Team over market timing -- at seed, we’re betting on you and your ability to founders and their ability to navigate twists and turns as the future unfolds. We want builders who are customer-obsessed, product-driven, and AI-native. Slope over intercept.
- Sell the outcome, not the platform -- the best companies in this cycle will be measured by what they deliver, not what they enable. FDEs, not sales engineers.
- Paths to compounding moats: This era of AI will be defined by last mile delivery. It’s getting easier to build but that does not mean it’s easier to figure out “what” to build. We believe that the core tenants that lead to structural moats ring true – capturing proprietary data sets, network effects, deep applied industry knowledge.
If this is you and you’re building against one of the trends above, we want to know you. Here’s where you can find us.
- Email: fintech@primary.vc
- X: @emily__man & @nwdaley
- LinkedIn: Emily Man & Nick Daley