The carbon credit market runs on trust, a commodity that has been in short supply. For a seaweed farmer in Indonesia or a mangrove restoration project in Florida, proving that a tonne of carbon was actually sequestered requires expensive, on-the-ground verification. It is a problem of measurement, and a small team out of the University of Waterloo believes the answer is not on the ground, but in orbit.
hum.ai, formerly known as Coastal Carbon, is building what it calls multimodal foundation models for earth observation [hum.ai, retrieved 2024]. The company's stated goal is audacious: to develop an artificial general intelligence for the natural world [hum.ai, retrieved 2024]. Its more immediate application, however, is decidedly concrete. Its AI models are designed to quantify biomass, like seaweed, from satellite imagery, creating the audit trail needed to issue carbon credits [Forbes, retrieved 2026]. In a sector plagued by credibility gaps, hum.ai is betting that a higher-resolution view from space can ground the market in better data.
From Coastal Carbon to a General Intelligence
The company's origins are visible in its previous name and its first demonstrated use case. As Coastal Carbon, the team focused specifically on using AI and remote sensing to verify and monitor blue carbon projects, which involve coastal and marine ecosystems [Crunchbase, retrieved 2026]. A profile from the University of Waterloo notes the alumni-founded company secured $1.6 million to help fight climate change, though the specific round and investors for that raise are not detailed in hum.ai's current public record [University of Waterloo, retrieved 2026]. The technical work involved training models to analyze satellite images to measure things like seaweed growth, providing a scalable alternative to manual surveys [Forbes, retrieved 2026].
That specific application appears to have been the wedge into a broader ambition. The company now positions itself as a pioneer of foundation models that go beyond internet text, trained instead on satellite remote sensing fused with real-world ground truth data [hum.ai, retrieved 2024]. The team, which includes PhDs and engineers, argues that understanding the physical planet requires a model built on its own data, not a repurposed language model [hum.ai LinkedIn, retrieved 2024].
The Wedge: Verifying What You Cannot Touch
The initial market is the voluntary carbon market, specifically nature-based solutions. Customers are described as being in nature conservation, carbon dioxide removal, and government sectors [hum.ai LinkedIn, retrieved 2024]. For them, hum.ai's promise is one of scale and objectivity. A model that can consistently measure carbon stocks from space could lower verification costs and increase confidence in credits, a prerequisite for the market's growth.
The company lists Amazon AWS and the United Nations as partners, though the nature and scope of these partnerships are not detailed publicly [Climate Draft Job Board, retrieved 2026]. The backing from a cluster of specialized investors,including climate-focused F4 Fund and hard-tech incubator HF0, alongside Inovia Capital and Propeller Ventures,suggests the thesis has found an audience [F4 Fund]. These are not generalist AI investors, but groups with mandates for deep tech and climate applications.
The Team and the Traction Gap
Founders Thomas Storwick and Kelly Zheng, both University of Waterloo engineering alumni, lead the company [University of Waterloo, retrieved 2026]. Storwick holds a master's degree in chemical engineering, while Zheng is a PhD candidate in the same field [University of Waterloo, retrieved 2026]. The team also includes a remote sensing lead with a PhD in Earth Sciences [Coastal Carbon, retrieved 2026]. This academic and engineering pedigree is typical for a company tackling a hard science problem, but it leaves open questions about commercial scaling. The public record shows a focus on technical development, with less visibility into sales, marketing, or deployed product details [Perplexity Sonar Pro Brief, retrieved 2024].
Current hiring efforts point to the next phase of work. The company is actively recruiting for an AI Research Scientist and a Chief of Staff, roles that indicate a push to advance core model capabilities and build operational structure [ZipRecruiter, retrieved 2026] [Jobs.ClimateDraft, retrieved 2026].
| Role | Background Note |
|---|---|
| Thomas Storwick (Co-Founder) | Waterloo Engineering alum (BASc '19, MASc '21) [University of Waterloo, retrieved 2026] |
| Kelly Zheng (Co-Founder) | Waterloo Engineering alum, PhD candidate in chemical engineering [University of Waterloo, retrieved 2026] |
| Rob Braswell (Remote Sensing Lead) | PhD in Earth Sciences [Coastal Carbon, retrieved 2026] |
The Risks on the Horizon
For all its technical ambition, hum.ai operates in a field where success is measured in contracts, not citations. The most immediate challenge is moving from technical capability to commercial validation. The company's public materials cite customer sectors but do not name specific commercial customers or deployed products [Perplexity Sonar Pro Brief, retrieved 2024]. In a B2B enterprise sale, especially to governments and large project developers, a lack of public referenceable customers can slow early momentum.
The competitive landscape is also quietly assembling. While hum.ai lists no direct competitors in its sourced materials, the problem of environmental monitoring via satellite is attracting attention. Larger geospatial analytics firms and specialized climate tech startups are all training models on similar data. hum.ai's differentiation rests on its claim to be building a foundational "AGI of the natural world," a long-term architectural bet that may be difficult to communicate to a buyer who simply needs a reliable seaweed counter.
- Proof of scale. The core risk is demonstrating that its models work reliably across diverse geographies and ecosystems, not just in research settings. A failure to generalize would limit its market to niche pilots.
- The regulatory context. Carbon credit methodologies must be approved by standards bodies like Verra or the Gold Standard. Integrating a novel AI-driven measurement tool into these frameworks is a non-technical hurdle that requires patience and diplomacy.
- The data moat. The company's long-term advantage depends on accumulating a proprietary dataset of satellite imagery paired with high-quality ground truth. Competitors with deeper pockets could attempt to replicate this, making speed and exclusive partnerships critical.
The Next Twelve Months
The coming year will be about translation: turning technical prototypes into paid deployments. The likely milestones are less about model parameters and more about market formation. A first publicly disclosed contract with a carbon project developer or a registry would be a significant signal. Further clarity on the partnerships with Amazon and the UN would also help substantiate the path to market.
Financially, the company appears to be operating with early-stage capital. The previously reported $1.6 million raise as Coastal Carbon likely provided the initial runway [University of Waterloo, retrieved 2026]. Given its current hiring push and ambitious roadmap, a new funding round to support growth would be a logical next step, potentially positioning the company for a Series Seed or A round in the near future.
The ultimate test for hum.ai is not whether it can build a clever model, but whether that model can change the economics of trust. The disease state here is the uncertainty and high cost of monitoring, reporting, and verification (MRV) in the carbon markets. The patient population is every project developer, investor, and corporate buyer trying to build a credible offset portfolio.
The standard of care today is a patchwork of manual sampling, sporadic drone surveys, and self-reported estimates, a process that is slow, expensive, and difficult to audit at scale. This friction limits the growth of high-integrity carbon projects and leaves the market vulnerable to criticism. If hum.ai's models can deliver audit-grade accuracy from orbit, they wouldn't just be selling software,they'd be selling liquidity into a market that desperately needs it.
Sources
- [hum.ai, retrieved 2024] Company website | https://www.hum.ai/
- [hum.ai, retrieved 2024] Research page | https://hum.ai/research
- [LinkedIn, retrieved 2024] hum.ai company page | https://www.linkedin.com/company/hum-ai
- [F4 Fund] Portfolio page for Hum.AI | https://f4.fund/portfolio/
- [University of Waterloo, retrieved 2026] Alumni company news article | https://uwaterloo.ca/engineering/news/alumnis-company-lands-16m-help-fight-climate-change
- [Forbes, retrieved 2026] Coastal Carbon profile | https://www.forbes.com/profile/coastal-carbon/
- [Crunchbase, retrieved 2026] Coastal Carbon company profile | https://www.crunchbase.com/organization/coastal-carbon
- [Climate Draft Job Board, retrieved 2026] Chief of Staff job listing | https://jobs.climatedraft.org/companies/coastal-carbon-2/jobs/42816529-chief-of-staff
- [ZipRecruiter, retrieved 2026] AI Researcher job listing | https://www.ziprecruiter.com/c/Hum-AI/Job/AI-Researcher/-in-San-Francisco,CA?jid=d07662047e5ba21e
- [Coastal Carbon, retrieved 2026] Legacy company website | https://coastalcarbon.ai/