Unlike today’s assistants, agentic twins – digital clones of people, firms and systems—do not just generate. They act. They make decisions, execute tasks and transact autonomously. In short, they are economic actors.
Investment is already flooding in. Nearly $2bn was committed to agentic-AI startups in 2025, almost double the year before. Analysts forecast the market to soar from $3.7bn in 2023 to more than $100bn by 2032. Jared Friedman of Y Combinator imagines companies run entirely by agents rather than humans. Protocols from Anthropic and Google are laying the plumbing for interoperable networks. The technology is still rough – agents struggle with reasoning and cross-domain generalisation – but the trajectory is unmistakable.
With autonomy comes risk. Agents embedded deep inside systems require sweeping access to calendars, credit cards and communications. Traditional permission frameworks are no match for such exposure. A new trust architecture is essential: cryptographic identities to verify agents, charters defining their powers, audit trails to monitor behaviour and kill switches to shut them down when they go rogue. Without such scaffolding, agentic AI could become a cybersecurity nightmare.
But the real leap in governance will come when agents fuse with digital twins—high-fidelity simulations of people, organisations and infrastructure. These twins already model organs for medical treatments, simulate traffic flows for cities and train autonomous vehicles. Combined with agency, they could reroute power grids, search clinical databases or even conduct geopolitical influence operations. Properly governed, such systems could extend human capability; poorly governed, they could magnify error and malice alike.
For twins to act responsibly, they need structure. Think five layers: identity (a verified portrait of and tight governance guidelines – agency charters – from the person or entity), dynamic data (real-time updates), intelligence (AI models that simulate behaviour), governance (rules and guardrails) and executive code (the power to act). The challenge is to align these layers with human values and institutional norms.
Economic systems also require contracts. Yet today’s legal documents are opaque to machines. Enter smart contracts: machine-readable agreements that self-execute when conditions of the agency charters are met. These can underpin agent-to-agent transactions, reducing the need for intermediaries. Though smart contracts cannot yet cope with legal nuance or real-world enforcement, hybrid models – pairing human-readable text with structured ontologies for machines – are emerging.
As agentic twins multiply, they will need marketplaces to find and bargain with one another. Early examples include Salesforce’s AgentExchange and Fetch.ai’s decentralised infrastructure. Unlike human networks, which thrive on sheer size, the value of agentic marketplaces will lie in diversity and connectivity, a variety of objective-specialised agents collaborating in novel ways across supply chains, finance or logistics.
For business leaders, the implications are twofold: First, invest beyond intelligence toward twin-controlled agency. The commercial advantage will come from a more trusted architecture, governance mechanisms and interoperable systems that make agentic twins collaborative and capable, but also accountable. Second, prepare for new forms of commerce. Smart contracts and machine marketplaces will upend traditional service models, creating opportunities for firms that can operate nimbly in hybrid human- and machine-native environments.
For policymakers, the task is starker: Standards must be crafted to embed identity, privacy and security management without throttling innovation, a delicate balance that has proven elusive across Europe, but also the US and China. Otherwise the agentic economy could accelerate instability rather than resilience, whether through cyber vulnerabilities, market manipulation or political interference.
This is no longer about faster document drafting. It is about autonomous digital actors – clones of people and institutions – making decisions, trading value and exercising power on our behalf. The question is not whether the Agentic Twin Economy will arrive, because business has always been good at alleviating friction, provided sufficient funding. Rather, it’s whether it will be built on solid foundations of human and institutional agency, trust and accountability. Leaders who design those foundations will shape not just the future of business, but the future of global economic and political governance itself.
Dr. Olaf Groth, ist Praxis-Professor für Strategie, KI, Technologie-Innovation und Weltwirtschaft an der Haas School of Business der UC Berkeley University in Kalifornien, sowie Gründer/CEO von Cambrian Futures & Cambrian Labs. Er ist Autor von The AI Generation und The Great Remobilization: Strategies & Designs for a Smarter Global Future.
Tobias Straube ist Dozent an der Steinbeis Augsburg Business School, Vice President für Analyse bei Cambrian Futures & Cambrian Labs, Co-Founder von Scio Network und Board Member bei NextWave.