Costanoa leads Seed and Series A rounds, backing teams solving big problems in complex, high-impact markets the partners know best: applied AI, AI & data infrastructure, cybersecurity, national security, and fintech. The firm runs a high-conviction, low-volume model — each partner makes only 2-3 investments a year so they can be a hands-on lead. Their current center of gravity is applied AI, where their working view is that 'the TAM is all of human labor' and that every white-collar automation problem is really a verifiability problem.
Open the deck on a sharp, specific pain point. For applied AI, the framing should make clear that the underlying work is verifiable enough to automate; in hard sectors, the problem must be one the team has lived.
“Every white-collar automation problem is really a verifiability problem.” — The TAM Is All of Human Labor
“what's the pain point you're solving?” — Amy Cheetham
“Electronic warfare has evolved from a specialized capability to an everyday necessity.” — Greg Sands
“a willingness to tackle one of the hardest problems in enterprise security.” — Greg Sands
Clearly explain how you solve (or will solve) the problem. For AI products, what wins is deep context and customer empathy rather than a generic model — and an architecture that turns bespoke work into reusable primitives.
“how are you or how will you solve the problem?” — Tips for Building an Early-Stage Fundraising Deck
“Winners drive cost per environment down over time. They increase reuse and consistently turn bespoke builds into primitives.” — Sean Cai
“We're most excited about new enterprise applications for continuous training and synthetic data that improve in-the-wild model performance.” — Sean Cai
“a good deck makes its case quickly and thoroughly without overwhelming the reader.” — Amy Cheetham
Show how big the problem is. Costanoa is comfortable with very large markets when there is a credible structural reason incumbents can't capture them — in applied AI they frame the opportunity as all of human labor.
“The TAM is all of human labor.” — The TAM Is All of Human Labor
“how big is the problem?” — Amy Cheetham
“Only the largest companies have the AI expertise and funding to solve them in-house, leaving a huge opportunity for new startups to tackle these challenges.” — Amy Cheetham
“Today, labs and well-funded enterprises pay roughly $100K–$400K per environment, depending on complexity and the number of in-line tasks.” — Sean Cai
Show progress to date — it doesn't have to be only revenue. Costanoa puts heavy weight on early proof from a small number of marquee customers and their intent to expand.
“show any progress you've made to date.” — Tips for Building an Early-Stage Fundraising Deck
“what really made this acquisition happen was seeing the early success of the first couple of F50 customers and their aggressive plans to roll out more broadly once they saw it delivered on its promise.” — Greg Sands
“strong early traction with key defense customers” — Greg Sands
“many are quietly running pilots with budgets of 1 to 2 million dollars per contract.” — Sean Cai
Show why this is the right team to build this company. Costanoa prizes technical depth (MLE expertise, hiring networks) plus operational credibility and lived experience in hard markets, and founders who are resilient over the long haul.
“showcase your team and explain why it is the right one to build this company” — Tips for Building an Early-Stage Fundraising Deck
“We'd love to chat with teams building where models actually learn—especially those with strong MLE expertise and real hiring networks.” — Sean Cai
“Technical excellence? Check. Operational courage? Absolutely.” — Greg Sands
“thoughtful, resilient, and deeply committed to doing the hard, often unglamorous work of building something lasting.” — Greg Sands
Show a path from services or early contracts to a durable, reusable platform with improving unit economics. Costanoa likes a sequenced model that earns the right to a larger, stickier revenue stream over time.
“Sell environments to labs now to earn the right to sell RLaaS and continuous learning to enterprises later.” — The TAM Is All of Human Labor
“declining COGS and a reusable stack” — Sean Cai
“Your ARR per employee should be higher than ever before.” — What Our Founders Taught Us This Year
Show why incumbents and well-resourced labs can't simply absorb you. Costanoa likes opportunities created precisely because incumbents are too big, slow, or constrained to act — and because the lab playbook has leaked to startups.
“Labs are no longer the sole key holders and are actively productizing.” — The TAM Is All of Human Labor
“The early playbook for building 'frontier-quality' data and environments, one locked up in labs, has leaked into startup land.” — Sean Cai
“Only the largest companies have the AI expertise and funding to solve them in-house, leaving a huge opportunity for new startups to tackle these challenges.” — Amy Cheetham
Make the timing inevitable. Costanoa's current why-now centers on the shift from pretraining to RL/environment work and on AI making previously unsolvable workflows tractable across fintech, security, and the broader enterprise.
“Pretraining is losing its edge...The next step change lives in the mid-training and RL/environment work.” — The TAM Is All of Human Labor
“Every fintech will be an AI company, sooner rather than later.” — Amy Cheetham
“Over time, as generative AI is better understood and the technology matures, we will see value creation expand from internal use cases to include the application layer as well.” — Amy Cheetham
“Electronic warfare has evolved from a specialized capability to an everyday necessity.” — Greg Sands
Be explicit about the round: round size, prior capital raised, and how much runway it buys. Costanoa leads at Seed and Series A and frames the right raise around maximizing usefulness per dollar of check.
“include round size, any prior capital raised, and how much runway this round provides” — Tips for Building an Early-Stage Fundraising Deck
“We invest from pre-seed through Series A (earlier is better), with one goal: maximize the helpfulness-to-check size ratio.” — Sean Cai
“this will vary depending on your company, but it might include financials, a more detailed product roadmap, customer testimonials” — Amy Cheetham
“This outcome wasn't driven by hype or shortcuts. It came from long-term thinking, deep customer empathy, and a willingness to tackle one of the hardest problems in enterprise security.”
“We invest from pre-seed through Series A (earlier is better), with one goal: maximize the helpfulness-to-check size ratio.”