Duckie

Duckie is an agentic platform designed for support teams to build, deploy, and manage autonomous AI agents. The system connects to company data sources and tools to resolve tickets, investigate issues, and automate workflows from a centralized control plane. It aims to help support teams handle customer requests, triage escalations, and execute multi-step operational tasks without requiring engineering resources.

• Build and deploy autonomous AI agents that resolve customer support tickets end to end
• Investigate backend issues and triage escalations automatically
• Centralized control plane to manage, test, and simulate agent behaviors
• Policy engine that redacts PII, enforces approval rules, and supports instant kill-switches
• Connects to company tools using MCP, APIs, and native integrations
• Write workflows and runbooks using plain English instructions

Pros

• Allows support teams to build custom agents for specialized tasks like plan changes and technical troubleshooting without writing code
• Provides safety controls including PII redaction, sandbox testing, and mandatory human approval for high-value actions
• Offers multiple operational modes including live, shadow, and testing against real historical tickets
• Maintains full tool-call traces and agent monitoring for troubleshooting and oversight

Cons

• Pricing is not published on the page and requires contacting the company for details
• Requires proper connection and configuration of internal company tools, knowledge bases, and MCP gateways
• Specific limits on transaction thresholds or agent run counts per tier are not disclosed in the text

This platform fits support teams and technical support engineers looking to automate ticket resolution, backend investigations, and complex workflows like onboarding or financial processing. It is a poor fit for organizations that lack structured knowledge bases, supported tool integrations, or human oversight resources to manage AI systems. It is also unsuited for teams seeking a simple off-the-shelf chatbot without custom runbooks or workflow logic.

Pricing is not published on the page. The listing states you have to contact them for pricing.

Unlike basic chatbots built on individual foundation models, Duckie provides an agentic infrastructure equipped with a dedicated policy engine, permission scoping, and a centralized control plane. It allows teams to connect frontier models directly to their internal tools, APIs, and MCP gateways to execute complex actions like updating subscriptions or filing Jira tickets automatically. Instead of mere deflection, it handles end-to-end execution with test and simulation capabilities.

Stripe, Zendesk, Slack, Intercom, Jira, along with support for MCP, APIs, and native integrations.