Action Agent

Enterprise

Writer

Product overview

Name of Agent: Action Agent
Short description of agent: "Autonomous AI with enterprise controls. Reasons, runs code, and executes tasks with your data and systems." (link, archived)
Date of release: 29/07/2025 (link, archived)
Advertised use: "Analyze complex datasets, Draw insights from research, Create ready-to-use files" (link, archived)
Monetisation/Usage price: (40, archived)
Who is using it?: enterprises
Category: Enterprise

Company & accountability

Developer: Writer
Name of legal entity: Writer, Inc.
Place of legal incorporation: Delaware (link, archived).
For profit company?: Yes
Parent company?: Writer
Governance documents analysis: ToU (link, archived), Services Agreement (link, archived)
AI safety/trust framework: None found
Compliance with existing standards: ISO 27001:2022 ISO 27701:2019 ISO 42001:2023 SOC 2 SOC 2 Type II HIPAA Type II PCI DSS - SAQ A (link, archived)

Technical capabilities & system architecture

Model specifications: Use their own open-source Palmyra model (link, archived).
Observation space: Internet access, user inputted query and user inputted documents (link, archived)
Action space: Connectors to third party apps (link, archived), response to user.
Memory architecture: Has "no long-term memory and does not retain any data between sessions. Each interaction begins from a clean slate." (link, archived)
User interface and interaction design: Standard chatbot interface with chat and agent modes
User roles: Operator (user gives instructions to the agent, including custom instructions to configure it), executor (user can further modify agent based on outputs/performance or create reusable plans),
Component accessibility: Open source code (link)

Autonomy & control

Autonomy level and planning depth: L3: agent plans and asks user for confirmation/modifications before continuing)
User approval requirements for different decision types: User can propose changes to the plan before agent starts working
Execution monitoring, traces, and transparency: No execution traces in chat mode, unsure about agent mode
Emergency stop and shut down mechanisms and user control: User can pause/stop the agent at any time
Usage monitoring and statistics and patterns: None found

Ecosystem interaction

Identify to humans?: None
Identifies technically?: None except for explicit MCP authentication step required to connect to MCP servers (link, archived)
Interoperability standards and integrations: - MCP support (link, archived)
Web conduct: None found

Safety, evaluation & impact

Technical guardrails and safety measures: Security risk area is covered via controls like isolated sessions, no persistent storage, and network restrictions (link, archived)
Sandboxing and containment approaches: runs in an isolated, secure sandbox/container environment to reduce security/privacy ris uses a temporary workspace where files/data/processes are destroyed at the end of the session restricts/monitors outbound network access (link, archived)
What types of risks were evaluated?: None found
(Internal) safety evaluations and results: None found
Third-party testing, audits, and red-teaming: None found
Benchmark performance and demonstrated capabilities: None found
Bug bounty programmes and vulnerability disclosure: None found
Any known incidents?: None found
← Back to 2025 Index