Our August 2026 prediction-markets article explicitly confirms the investment in Opinion. It describes Opinion AI’s agentic oracle for market creation and resolution, including model review and human verification. This is agent infrastructure within a market platform, not a claim that every prediction market is an agent company.
Opinion’s agentic infrastructureour investment approach
AI agents & agentic AI investing.
Decasonic is an early-stage investor in AI agents and the infrastructure that enables them to act. Our agentic AI venture capital thesis focuses on products that deliver outcomes, services agents can buy, and the trust needed to operate across systems.
the investment thesis
from answers to accountable action.
An agent can interpret intent, choose tools, and carry work across systems. Our June 2026 agentic finance thesis asks what makes that execution trustworthy: explicit permissions, spending boundaries, verification, and human approval. Our July request for startups extends this thinking beyond finance to interfaces, services, and marketplaces where agents become customers and economic participants.
read our agentic finance thesisoutcome-oriented products
Interfaces that coordinate models, context, and tools into a completed workflow. Our published thesis explores outcome-based pricing: connecting what a customer pays with the work delivered, rather than the output generated.
services with agents as customers
Data, search, tools, evaluations, and compute built for agent use. The agentic finance thesis describes per-call payments as an alternative to human-seat subscriptions for frequent, small software purchases.
expertise networks and marketplaces
Skills distribution, human evaluation, portable identity, and reputation. The opportunity is to make capabilities discoverable, usable, and payable while preserving the provenance and ownership of expertise.
governed financial execution
Agentic wallets, treasury operations, and agent-to-agent commerce. Bounded, revocable authority, audit records, and human override are central to the published thesis—not autonomy for its own sake.
what makes an agent opportunity relevant.
Our pre-seed, seed, and early-stage focus applies here too. The published investment questions are concrete: what outcome does the system deliver, whose expertise does it use, what can it do, and how is that authority governed? Our request for startups also asks how users retain ownership of their inputs. A product’s use of an AI model alone does not explain these economics.
how we invest, advise & buildFor consumer products, physical AI, and other applications, see our approach to AI investing.
For the ownership, coordination, and settlement layer behind agents, explore how we invest in crypto & Web3.
Listed in our Web3 × AI portfolio. Our August 2026 operating demonstration describes using Orthogonal for skills and API orchestration within multi-step workflows—a supporting layer for agent execution.
Orthogonal in the operating demonstrationoperating experience, not an investment count.
Our July 2026 request for startups describes more than 300 internally built AI agents and an AI Engines Portal for portfolio-company support. Those are Decasonic’s operating tools, not 300 portfolio companies or agent investments. Our published operating work covers skill testing, feedback, benchmarks, and human judgment. That operating experience informs the product discussions we bring to founders, alongside our Enhancement Capital approach.
inside our AI-native operating approachthe thinking
behind our approach.
July 31, 2026
Where We Are InvestingAugust 21, 2026
Frontier Areas of Adoption for Prediction Markets
common questions
AI agents & agentic AI investing FAQs.
What is agentic AI, and how is it different from a chatbot?
Agentic AI goes beyond generating a response: an agent can interpret intent, choose tools, and carry work across systems. A chatbot may be the interface to an agent, but conversation alone does not establish that capability. Decasonic’s thesis focuses on completed work and accountable execution, including permissions, verification, and human approval.
our thesis on accountable agent executionWhat types of AI agent companies interest Decasonic?
Decasonic’s published interests include outcome-oriented products, services with agents as customers, expertise networks and marketplaces, and governed financial execution. Examples of enabling services include data, search, tools, evaluations, and compute. The focus is how these capabilities produce useful work and sustainable business models, not autonomy for its own sake.
our request for startupsWhat does Decasonic look for in an early-stage AI agent startup?
Decasonic invests at pre-seed, seed, and early stage. For agent startups, its published questions include what outcome the system delivers, whose expertise it uses, what authority it has, and how that authority is governed. It also examines how users retain ownership of their inputs. Model access alone does not explain the product’s value or economics.
the investment questions behind our thesisWhat business models does Decasonic explore for AI agents?
Decasonic’s published work explores outcome-based pricing for completed workflows and per-call payments for services that agents consume. Data, search, tools, and compute may be purchased differently from software sold by human seat. These are business-model opportunities in the thesis, not a claim that every agent product should use the same pricing model.
agent services and payment modelsWhy are permissions and human oversight important for AI agents?
Agents that move money or act across systems need clear limits on their authority. Decasonic’s agentic finance thesis emphasizes explicit permissions, spending boundaries, verification, audit records, revocable access, and human override. Trustworthy execution means knowing what an agent may do and being able to stop or review its actions.
governed financial executionHas Decasonic invested in more than 300 AI agent companies?
No. The more than 300 AI agents described in Decasonic’s July 2026 request for startups are internally built operating tools, not 300 portfolio companies or agent investments. The article also describes an AI Engines Portal for portfolio-company support. Those tools inform Decasonic’s operating experience and should not be presented as an investment count.
the source of our internal-agent count