Seed Investing in a Post-AGI World
Investing in the Abundance of Intelligence
--Paul Hsu, General Partner, and Abdul Al Ali, Venture Investor, at Decasonic
AGI is increasingly in reach.
Over the weekend, we saw a cluster of announcements from key figures at the leading frontier labs - Anthropic, OpenAI, Google, Meta, and xAI - converging on what seems to be the same message: alignment is not keeping pace with the advancement in intelligence capability. Model capability is outrunning human-centric alignment, monitoring, and governance systems - with the recent exploits around Hugging Face
and the recurrent-depth techniques (i.e. the Chain of Thought visibility from Astra) providing potential justification for a slower pace in the pursuit of AGI.
In the pursuit of AGI, the main unknowns remain (1) the definition of AGI and (2) the timing associated with achieving it. The former remains unclear and contested. The latter is coming into focus, with our view at Decasonic that it arrives in less than five years. The pathway to AGI has been scale. Scale in capital, compute, energy, and data as the inputs. The output is a surplus of intelligence, and that surplus is where early-stage seed investing should be focused.
In venture, and more so in seed, outcomes take years to manifest. We are therefore underwriting a reality where AGI has already been achieved by the time our portfolio outcomes land. The question for early-stage seed investors like ourselves is where you invest in a world where AGI has already been achieved.
Investing in the Inputs vs the Outputs of AGI
Generally, what is actively observed is that investors (across various investing stages) take one or some combination of five positions in the stack: (1) you own the lab, (2) you own the infrastructure the labs run on - energy, power, compute, (3) you own the inputs that create deltas in the pursuit of AGI - data and domain expertise, (4) you own the optimizers - the companies making intelligence accessible, cheaper, and faster, and (5) you own the use-cases built on top of the models - orchestrators, services, interfaces, applications.
Capital is unevenly distributed across all five, with the main concentration in (1) and (2). This concentration in capital extends to the "barbell" we are seeing in venture capital, and the continued vote of confidence in the near-term extension of this barbell comes in the form of capital velocity to the crowned winners - OpenAI and Anthropic accounting for ~43% of global venture funding in H1 '26. Capital spend is directed at (1) but ultimately distributed to (2) and (3) as the labs continue to accelerate their spending, with early signs of the labs deploying into (5) themselves (Grok Bot, Meta Muse, and Claude Design being some of the most recent examples).
Some seed investors might pivot in the “means of production trade,” but it is (in our view) a potential mistake as it does not produce the outlier outcomes that seed-stage investing is meant to realize. Owning the outputs from the surplus and the availability of the surplus of intelligence in a post-AGI world is where outlier outcomes can be produced.
The Surplus of Intelligence
The output of AGI is an abundance of intelligence. Intelligence that was previously scarce, expensive, and concentrated in individuals becomes cheap, on-demand, and (largely) unconstrained by headcount. Abundance will not mean accessibility for everyone - distribution, cost, and capability will remain uneven for some time. What it does mean is a surplus, and we seek to invest in that surplus.
This AGI is primarily going to be digital AGI. Our view is that digital AGI accelerates the physical world as well: (1) robotics, with Atlas as the most visible example of software intelligence compounding into embodied capability, (2) hardware and chip design, where AI-driven design loops are compressing cycles that previously took years, and (3) materials discovery and lab automation, where experimentation is shifting from human-paced to intelligence-paced. The surplus of digital intelligence is the input, and the acceleration of physical outcomes is the compounding effect.
The Transformative and the Optimizer
Investing in the surplus, you are primarily underwriting one of the below:
The transformative: new use-cases (often presented in services or products) that did not previously exist.
The optimizer: model utilization extended to existing world conditions - across workflows, states, and events. Markets already exist, and the goal is to make marginal improvements (around costs and output).
In either building the (1) transformative and/or (2) the optimizer, the perspective of the (1) builder and (2) the user shifts. The builder is naturally a combination of human and software AI (and potentially physical AI in the future). The user shift is primarily in recognizing that consumers of the future might be a mix of humans, software AI agents, and physical AI agents.
The transformative and the optimizer must recognize that (1) their users today might not be their users tomorrow and (2) depth in understanding the pain point of users now and in the future is a significant differentiator (user-empathy accelerated by AI-driven simulation and assessment).
Investing in the Transformative
What excites us at Decasonic is investing in the transformative. The transformative will (largely) the predominant beneficiaries of abundant intelligence relative to the optimizers, because abundance lowers the cost of creating something new far more than it lowers the cost of improving something that exists. Defensibility with the transformative will come with human-AGI collaboration that allows for the utilization and deployment of:
Judgement and Taste: AGI will intensify the levers for experimentation in the pursuit of creation. Judgement and taste solidified in the form of customer expertise (and empathy) will allow innovations to differentiate and gain adoption.
Expertise and Decentralization: The marginal edge. AGI can teach you everything you might need, but it does not endow you with the marginal domain expertise that comes from being a domain expert in a field or sector. This led to the rise of "vertical harnesses" and "domain expert data,” as we see today.
Narrative and Community: Products and services will increasingly become competitive. Maintaining a core differentiator will increasingly rely on clarity in narrative communication and the community that forms around it.
Execution Speed: Execution is speed. Intelligence can now be utilized to execute at a significantly greater scale than before, and the teams that compound this into faster cycles of shipping, learning, and shipping again will outperform (with an emphasis on valuable customer feedback).
Ultimately, the means to achieve the transformative will be much more accessible. The success differentiator lies in a combination of the above in the pursuit of truly transformative outcomes. The pathway towards the transformative is littered with experimentation. Therefore, you recognize (1) the individuals pushing the boundaries of the frontier and (2) the mission-driven who are able to persist in their pursuit of the transformative. You want to invest in great people who aim to achieve great outcomes, and this is the general path of transformative investment(s) in a post-AGI world.
Beyond Investing: Future-Proofing
The role of an early-stage venture investor in this world is twofold: (1) investing in this reality and (2) future-proofing both the existing portfolio and the fund itself against it.
For new investments, this means underwriting a company with the conviction that it will achieve outlier outcomes post-AGI. For portfolio companies, this manifests in actively supporting founders to innovate and experiment at the frontier while pursuing the transformative.
We come from the perspective that you have to be AI-native to invest in a post-AGI world. At Decasonic that means we apply the AI-native lens both internally and externally - internally through the Decasonic AI OS, our platform of production AI applications and specialized agents operating across both pre and post-investments. Externally, through our AI Engines Portal (AIEP), which extends that same operating system to our portfolio companies (with an emphasis on post-investments).
Building the Transformative
This is how we are approaching seed investing in a post-AGI world at Decasonic : AGI as a given, the abundance of intelligence as the surplus, and the transformative as where we invest it. If you are a founder building the transformative in pursuit of outlier outcomes, reach out to us.
The content of these blog posts is strictly for informational and educational purposes and is not intended as investment advice, or as a recommendation or solicitation to buy or sell any asset. Nothing herein should be considered legal or tax advice. You should consult your own professional advisor before making any financial decision. Decasonic makes no warranties regarding the accuracy, completeness, or reliability of the content in these blog posts. The opinions expressed are those of the authors and do not necessarily reflect the views of Decasonic. Decasonic disclaims liability for any errors or omissions in these blog posts and for any actions taken based on the information provided.
Comments