What problem does Phronesis solve for autonomous agents?
Phronesis provides a neutral decision-assurance layer for autonomous agents, turning their intent and evidence into an auditable Decision Asset. It addresses the gap in agentic commerce where neither payment systems nor identity layers can verify if an agent's judgment is defensible. The platform issues an auditable action boundary before the agent acts, based on evidence maturity, calibration history, and decision materiality.
In what situations is Phronesis used?
Phronesis is used as a decision-assurance layer for agentic workflows, specifically within the Model Context Protocol (MCP) and REST-based systems. It is designed for situations requiring trusted autonomous judgment, such as customer service, coding, productivity, sales, and voice applications. The platform supports six primitives across twelve applied verticals in the agentic economy.
What features, surfaces, or integrations does Phronesis offer?
Phronesis offers a decision-assurance platform with an auditable ledger, a calibration scorecard, and an Open Decision Quality Bench. It provides MCP and REST endpoints co-equally, an API spec, and a public 'agents.txt' file for agent discovery. The platform features Market Memory as a core component and consumes the AP2 mandate at the entry point.
Who is Phronesis designed for?
Phronesis is designed for developers and organizations building or operating autonomous agents within the agentic economy. It serves as the 'missing middle' between open agentic commerce and identity layers, providing the judgment verification that payment and identity systems cannot. The product is categorized under Workflow Automation, targeting users needing to audit and validate agent actions.
What is Phronesis?
Phronesis is the agentic economy's decision-assurance layer—a neutral substrate that autonomous agents call before acting. It transforms intent, evidence, and calibrated forecasts into an auditable Decision Asset, providing a defensible action boundary. The platform is built on the Model Context Protocol (MCP) and REST APIs.
5 of 6 research questions are answered for this product. The rest need source evidence we have not collected yet, so they are left unanswered rather than guessed.