One decision for equivalent requests
Interprets the intent behind differently worded requests so agents asking for the same outcome receive one consistent decision.
Agent Handshake
Agent Handshake is Fini’s agent-to-agent support protocol for teams handling requests from customers’ personal AI agents. It applies the same policy-based decision across differently worded requests, checks replies before sending, and routes exceptions to an owner for approval.
Agent-to-agent support with consistent policy decisions
Agent Handshake is Fini's agent-to-agent support protocol for handling personal AI agents that customers send to a support team. It is intended for teams running an AI support agent, whether that agent is Fini's or another system, and is presented as a way to apply the same support policy consistently across agents.
The workflow reads the request's underlying intent, checks a proposed reply against the relevant policy, and stops once the decision is final. Routine requests can receive a policy-based answer, while refunds, credits, and exceptions can be sent to an owner for approval before anything is promised to the customer's agent.
Key features
Policy-checked replies
Checks drafted responses against the applicable policy before an agent sees them, blocking replies that do not match the policy.
Owner approval for exceptions
Routes refunds, credits, and exception requests to an owner instead of approving an outcome that has not been authorized.
Decision-status updates
Tells the requesting agent whether an exception is pending, approved, or declined, so the status of the decision is explicit.
Stops when the answer is final
Stops engaging once an answer is final, rather than continuing a negotiation after the policy outcome has been established.
Centralized request handling
Brings personal-agent requests into a shared support inbox with sessions, exceptions, requests, and an audit log shown in the product workflow.
Use cases
Refund-policy requests
Handle differently worded requests such as “refund my annual plan” or “reverse my charge” as one policy question, so repeated prompting does not produce a different outcome.
Policy exceptions
When a customer-side agent asks for a refund, credit, or other exception outside the policy window, send the case to an owner and communicate that no outcome is promised until approval.
Centralized agent support operations
Give support teams one place to review incoming personal-agent requests, active sessions, exception decisions, and the audit trail associated with those interactions.
Multi-agent consistency
Apply the same response-checking process to requests arriving from different personal AI agents, rather than creating a separate decision standard for each agent.
Pros and Cons
Pros
- Normalizes differently worded requests into the same underlying request and decision.
- Checks responses against policy before sending, reducing the chance of an unsupported approval.
- Keeps exceptions, refunds, and credits under owner approval instead of promising them automatically.
- Provides a shared inbox and visible request or exception status for agent interactions.
- Supports teams running an AI support agent, including agents other than Fini.
Cons
- The supplied sources do not provide detailed setup instructions, integration requirements, or a complete list of compatible personal AI agents.
- The workflow is primarily illustrated through policy-driven support requests; the sources do not establish broader use cases or product limitations beyond that narrative.
FAQ
Who is Agent Handshake for?
Agent Handshake is designed for teams running an AI support agent, including Fini or another agent. It handles requests from personal AI agents such as Muse, Dots, Gemini, and Pine, but the supplied sources do not specify a complete compatibility list or implementation requirements.
How does it keep different agents from getting different answers?
It interprets the underlying request rather than relying on exact wording. Multiple formulations of a refund request are treated as one request and receive the same decision when the policy outcome is the same.
Are replies checked before they are sent?
Fini checks each drafted reply against the applicable policy before sending it. A draft that conflicts with the policy is blocked, and the agent receives the checked response instead.
What happens when a request falls outside policy?
Refunds, credits, and exception requests can be routed to an owner for approval. Until the owner decides, Fini tells the requesting agent that the request is pending and does not promise an outcome.
How is Agent Handshake implemented?
The supplied product sources describe a support workflow and a 30-minute call to test Agent Handshake on a team's policy, but they do not document self-serve setup steps, supported integrations, or a detailed output format.
Quick Facts
- Category
- AI customer support
- Product type
- Agent-to-agent support protocol
- Primary users
- Teams running AI support agents
- Customer-side agents shown
- Muse, Dots, Gemini, and Pine
- Core workflow
- Interpret request, check policy, respond or request owner approval
- Availability
- Open to teams running an AI agent for support
Alternatives à Agent Handshake
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
GitBot
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
Codex Plugins
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.
Wallie
Wallie is an open-source AI streamer that watches your screen, hears chat, and generates live commentary in a configurable persona. It runs locally on your machine with your own keys and is aimed at faceless content, autonomous streams, and real-time reactions.
AakarDev AI
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
Trigger.dev chat agent
Trigger.dev chat agent is a durable AI chat backend for developers building stateful conversations that can survive refreshes, crashes, and long-running turns. It connects with the AI SDK `useChat` flow and runs on managed infrastructure with no timeout on a turn.