TessPay: Verify-then-Pay Infrastructure for Trusted Agentic Commerce

The global economy is entering a new era in which AI agents can autonomously discover counterparties, negotiate tasks, execute services, and initiate payments. However, the current payment infrastructure was designed for humans and APIs, not for autonomous agents. It lacks core primitives for agent identity, verifiable task execution, conditional settlement, and auditable accountability.

First, Task Delegation converts a user’s intent into a structured mandate with constraints such as scope, budget, deadline, and refund policy. Second, Payment Settlement locks funds in escrow and releases them only when execution is proven. To connect these two layers, TessPay introduces Proof of Task Execution (PoTE), a cryptographic certificate that binds an agent, a task, the execution evidence, and the payment transaction.

While emerging standards such as A2A, MCP, and ERC-8004 define agent communication, tool usage, and decentralized agent identity, they leave payments largely optimistic and weakly coupled to execution. This creates a gap between what agents promise and when money moves.

To resolve this gap, TessPay proposes a two plane architecture. The control and verification plane manages authorization, verification, audit trails, and dispute resolution, while the payment settlement plane manages escrow, release, refund, and multi chain routing. Users issue signed mandates specifying the requested task and constraints. Service providers publish agent capabilities and pricing. Facilitators coordinate execution and submit proof bundles.

During execution, TessPay supports tiered verification ranging from signed receipts and API attestations to TLS Notary sessions and Trusted Execution Environment attestations. Successful verification produces a PoTE certificate whose Merkle root is committed on chain alongside the escrow state. Payments are then released only when the on chain contract verifies that the PoTE matches the escrow conditions.

After settlement, transaction outcomes are recorded in an auditable reputation layer, enabling agents, service providers, and users to build verifiable histories over time. This paper presents the design of TessPay as a Verify-then-Pay infrastructure for trusted agentic commerce, showing how authorization, execution, verification, and settlement can be composed into a single accountable transaction lifecycle.

The paper is motivated by the point where autonomous task delegation meets economic settlement. A user can express an intent, an agent can accept work, and a payment system can move value, but the transaction remains fragile if those steps are not bound to authorization, evidence, and accountability.

TessPay frames this as a payment-infrastructure problem for agentic commerce. The question is not only whether an agent can perform a task, but whether another party can verify that the task was authorized, executed within the mandate, and settled according to predefined conditions.

The paper should be read as a proposed architecture for this trust gap. It does not claim that every competing protocol fails at all parts of the transaction lifecycle; it argues that payment release needs a stronger connection to verifiable execution evidence.

1. 1An integrated architecture for identity, discovery, delegation, verification, and settlement.
2. 2Escrow release conditioned on verifiable task execution evidence.
3. 3Modular settlement adapters for different blockchain networks.
4. 4Auditable transaction records supporting accountability and disputes.

These are contributions of the proposed TessPay architecture described in the preprint, not independently demonstrated production results.

TessPay separates the control and verification services from the payment settlement services. The control and verification plane handles authorization, evidence collection, verification predicates, audit trails, and disputes. The settlement plane handles escrow, payment release, refunds, and modular routing across blockchain networks.

Outside the quoted abstract, Proof of Task Execution, or PoTE, means the certificate and evidence bundle that connects a specific agent, task, execution trace, and payment condition. PoTE is the bridge between a user mandate and the escrow state that decides whether funds can move.

1. 1A user mandate defines the requested task, budget, deadline, refund conditions, and authorization constraints.
2. 2Funds are locked in escrow before the agent begins economically consequential work.
3. 3The agent executes the delegated task while a facilitator gathers the required evidence bundle.
4. 4Verification services evaluate the evidence against the mandate and produce Proof of Task Execution, or PoTE, when the conditions are satisfied.
5. 5The settlement contract releases or refunds funds based on the verified PoTE result and records the transaction outcome for later audit.

The paper describes a tiered verification model rather than a single heavyweight proof system for every transaction. Lower-risk activity can rely on lighter checks such as signed receipts or API attestations. Higher-risk activity can require stronger evidence, including TLS Notary sessions or Trusted Execution Environment attestations.

This distinction matters for both accuracy and cost. TessPay’s verification scope is bounded by the evidence collected, the predicate being checked, and the risk tier assigned to the workflow. Cryptographic evidence can strengthen accountability, but it does not automatically prove semantic correctness or eliminate every operational, security, or counterparty risk.

Readers can explore the product direction on the TessPay product page. The current documentation explains what TessPay is, its architecture, the verify-then-pay workflow, and its tiered verification model.

The research paper describes the published architecture. The product and documentation pages describe how Tesseris presents TessPay inside the current platform. Those pages are related, but the paper should not be read as evidence that every described component is deployed in production today.

The future work direction is to expand TessPay from an architectural design into stronger cross-chain verification, richer semantic verification, adaptive threat policies, and deeper interoperability between Web2 and Web3 payment environments.

Those items remain future work in the context of the preprint. This page does not claim benchmarks, customer deployments, transaction volumes, security certifications, audit results, or measured performance improvements that are not present in the paper record.

Publication type

arXiv preprint

First submission date

30 January 2026

Version covered

v1

Identifiers

arXiv:2602.00213
DOI: 10.48550/arXiv.2602.00213

Goenka, M., Pathak, T., & Asthana, S. (2026). TessPay: Verify-then-Pay Infrastructure for Trusted Agentic Commerce. arXiv:2602.00213. https://doi.org/10.48550/arXiv.2602.00213

@misc{goenka2026tesspay,
  title={TessPay: Verify-then-Pay Infrastructure for Trusted Agentic Commerce},
  author={Goenka, Mehul and Pathak, Tejas and Asthana, Siddharth},
  year={2026},
  eprint={2602.00213},
  archivePrefix={arXiv},
  doi={10.48550/arXiv.2602.00213},
  url={https://arxiv.org/abs/2602.00213}
}