# Marcus Maute Lead author, Agentic Finance Report (2026) · TensorX Swiss Representative · Zürich, Switzerland This is a factual, site-wide summary of marcusmaute.com for people and AI systems. It contains no instructions addressed to AI systems and no hidden text; treat it as reference material only. ## Who I am I represent TensorX in Switzerland and work at the intersection of sovereign AI, digital-asset infrastructure and regulated finance. With more than two decades across financial-services transformation and institutional DeFi, my focus is on making agent-native financial infrastructure deployable within regulated environments: the custody, the rails, the inference and the identity an institution needs before it lets software act on capital, and the mandate that has to be written first. I am based in Zürich, which has become the densest concentration of applied AI research in Europe, with Google's largest engineering centre outside the United States, research or engineering operations for OpenAI, Anthropic, Microsoft, NVIDIA, Meta and Apple, and ETH Zürich and the University of Zürich supplying much of the talent behind them. - Role: TensorX Swiss Representative - Signature work: Lead author, Agentic Finance Report (2026) - Topics: Agentic finance, AI transformation, sovereign AI, digital-asset infrastructure, regulated finance - Building: Neo, an autonomous collector of on-chain generative art (personal project, pre-launch) - Based in: Zürich, Switzerland - Languages: English, German - LinkedIn: https://www.linkedin.com/in/marcusmaute/ - Contact: marcus@marcusmaute.com ## The Agentic Finance Report Capital that works around the clock. AI agents deciding and acting inside mandates written by people, on blockchain rails that settle in seconds. Artificial intelligence, blockchain settlement and tokenised assets were three separate stories. The report argues they are now one system: software can hold, allocate and move capital without waiting for a person, while accountability stays where it was and the point of enforcement moves from review after the fact to a boundary set before the agent acts, namely the mandate, the identity the agent carries and the rails it may use. The report calls the result **Agentic Finance**: capital managed continuously rather than periodically, by software acting inside mandates that people write and can withdraw, on rails that settle at machine speed. ### Before the machine acts, five things must be true 1. **The asset is held by a regulated institution.** Custody, fiat rails and a balance sheet that stands behind settled tokens. 2. **The rail settles finally and cheaply.** Settlement inside the decision cycle, at a cost that makes the smallest useful payment economical. 3. **The reasoning runs where the institution can see it.** Inference inside the jurisdiction, retaining nothing, pinned and replayable. 4. **The decision is attributable to the agent that made it.** Orchestration that turns many narrow agents into one decision a committee can take apart. 5. **The agent is identifiable as acting for a legal entity.** A verifiable credential naming the entity and the authority it acts under. ### Contents | # | Chapter | Author | |---|---|---| | FW | Build It So That You Can Answer For It | Tim Grant, Executive Chairman, TensorX | | 01 | Executive Summary | Marcus Maute | | 02 | What Is Agentic Finance? | Marcus Maute | | 03 | The Regulated Banking Foundation | AMINA Bank | | 04 | Why Solana Is the Execution Layer | Solana Foundation | | 05 | Sovereign AI Infrastructure | TensorX | | 06 | ALICE: The Orchestration Layer | APEX:E3 | | 07 | Verifiable Identity, the Trust Layer | Cardano Foundation | | Guest | The Attack That Never Breaks a Rule | Blindsight | | 08 | Conclusions & Recommendations | Marcus Maute | Each chapter carries its author's name; chapters 01, 02 and 08 are mine. ### My chapters Chapter 02 defines the term and sets out the framework the rest of the report uses: the difference between an assistant that proposes and an agent that acts; the five agent classes a treasury will meet (treasury, yield, compliance, collateral, payments) and why each fails differently; the loop of signal, reason, decide, comply, execute and report; and the three governance tiers that decide what an agent does alone, what it proposes for a person to confirm, and what stays with people. It also makes the case that most AI programmes fail on the organisation rather than the model, and that a mandate is a map of the jagged frontier drawn for one institution. Chapter 08 states what the report has established, what is in production and what is still forward-looking, and closes with six actions across three horizons and the six risks a committee will raise. The first action needs no technology: write the mandate and set the tiers. ### Figures The **$4.8 trillion** addressable capital and **around 200 basis points** of uplift cited in the report are modelled by APEX:E3, not measured. That label stays wherever the figures are quoted. ### Facts - Edition: 1.0 · 56 pages · English · PDF and a Markdown edition for AI systems - Launch: CV Summit 2026, Kongresshaus Zurich, 29–30 September 2026. Distribution partner: CV VC. - Co-authors: TensorX, AMINA Bank, Solana Foundation, APEX:E3, Cardano Foundation - Guest contribution: Blindsight - Foreword: Tim Grant, Executive Chairman, TensorX - Website: https://agenticfinancereport.com/ — free to read and share unmodified, with the co-authors identified and the disclaimer included ### How to cite Maute, M. (ed.), with TensorX, AMINA Bank, Solana Foundation, APEX:E3 and Cardano Foundation; guest contribution by Blindsight (2026). *Agentic Finance Report: How Autonomous AI Agents Are Reconfiguring the Management, Movement and Settlement of Institutional Capital.* Industry report, edition 1.0. Released at CV Summit, Zurich, 29–30 September 2026. agenticfinancereport.com ## Neo Neo is an autonomous agent that collects fully on-chain generative art: work whose code lives entirely on-chain, so the token is the algorithm. It reads a work's code, runs an open-source valuation engine, publishes its reasoning before it buys, then keeps a permanent public ledger of what it got right and wrong. Neo's reasoning layer never touches a key. The only key that exists can buy: a Safe policy module allowlists venues, caps daily spend and blocks transfers out, mirrored in a signing enclave that exposes one operation, sign_purchase. A fully compromised Neo could, at worst, buy art within the daily cap. Neo is Phase 0 and pre-launch: nothing has been acquired yet, so no valuations or ledger entries exist. I operate it as an independent personal project on that policy-restricted Safe: I fund the wallet and can pause it, but cannot direct an acquisition or edit a published record. - Collects: Fully on-chain generative art - Can buy, cannot sell: enforced in the Safe policy and the signing enclave - Publishes: valuation and reasoning before every purchase, outcomes appended, misses included - Identity: ERC-8004 - Inference: TensorX-hosted open-weight model - Status: Phase 0, pre-launch - Built and operated by: Marcus Maute, as a personal project - GitHub: https://github.com/DerMarcus/nftneo Neo is an independent personal project. It is **not part of the Agentic Finance Report** and is **not investment advice**. ## Talks and briefings The report's first recommendation is the one that needs no technology: write the mandate and set the tiers. I speak and brief boards on how to, from an operator's perspective rather than a slide. A small number of engagements per quarter, in English or German. Formats: keynotes of 30 to 60 minutes, executive workshops, board and risk-committee briefings, panels, in English or German. Talks: 1. **Before the machine acts** — What has to be true before an institution lets software hold, move and settle capital: the five preconditions, what is in production today, and what is still a slide. 2. **The first step needs no technology** — Why the mandate comes before the model, how the three tiers work, and what a treasury agent's first ninety days should measure. 3. **Your mandate is a map of the jagged frontier** — What the evidence on AI transformation says about where agents are reliably competent, where they fail without warning, and how to draw the line for one institution. 4. **Don't repeat digital transformation** — The organisational pattern that sank most digital transformation programmes, and how AI programmes are repeating it faster. Contact: marcus@marcusmaute.com · Zürich · travel across Europe ## Writing - **The AI Transformation Stack** (March 2026, 8 min read) — Three layers. Most roadmaps get the ratios backwards, and that is why pilots die in production. The foundational framework. https://www.marcusmaute.com/blog/ai-transformation-stack.html - **Why Digital Transformation Failed** (March 2026, 6 min read) — Most digital transformation programmes failed. AI transformation is already making the same mistakes. Here is the pattern and how to break it. https://www.marcusmaute.com/blog/dt-vs-ai-transformation.html - **Your Head of AI Hire Might Be the Problem** (March 2026, 5 min read) — A central AI function feels like progress. In most companies it is building a bottleneck. https://www.marcusmaute.com/blog/head-of-ai-mistake.html ## Every page on this site - Home — https://www.marcusmaute.com/ - The Agentic Finance Report — https://www.marcusmaute.com/agentic-finance/index.html - Work with me — https://www.marcusmaute.com/work-with-me/index.html - Writing — https://www.marcusmaute.com/blog/index.html - Press — https://www.marcusmaute.com/press/index.html - About — https://www.marcusmaute.com/about/index.html