Macha AI
AI support agents that run on existing help desks like Zendesk, Freshdesk, Gorgias, Front, Intercom and HubSpot, set up by the Macha team.
About Macha AI
Macha, from AGZ Technologies Private Limited, runs AI agents on an existing help desk to answer, tag, route and resolve support tickets using a company's knowledge and connected tools. The Macha team analyzes the help desk, builds the agents and deploys them in safe mode, where the agent writes internal notes and a human team approves anything sent to customers.
What It Is
Macha is an AI customer-support agent platform that works inside the help desk a team already uses, including Zendesk, Freshdesk, Gorgias, Front, Intercom and HubSpot. Each agent handles one ticket category with its own instructions, tools and approved wording. Agents can also be built in the dashboard, with the Sidekick build copilot, or from the terminal with Claude Code and Codex.
How the done-for-you setup works
The site describes a process in which the team takes a brief, analyzes the help desk to rank ticket categories by volume, and recommends which to automate and which to keep with people. It then builds a knowledge base from the help centre, website and past replies, connects tools, and tests on past tickets. The agent goes live in safe mode, and automatic replies are switched on once the customer is comfortable. Macha states that the first category typically goes live within a week of the kickoff call.
Tools, knowledge and triggers
Connectors give agents read and write access to services such as Shopify, Stripe, Slack, Notion, Google Workspace and Confluence, so they can look up orders, issue refunds or update addresses. Custom Tools wire in any HTTP API or internal system. Data sources include uploaded documents, crawled websites, and live Google Docs, Notion and Confluence content. Agents can run on new tickets, Slack messages, webhooks or schedules, and can delegate to sub-agents.
Evaluation and testing
Documentation describes Agent Auto-Eval, which grades autonomous runs against the agent's instructions, plus manual evaluations with custom rubrics. A Try it sandbox and Agent Simulations replay past Zendesk or Freshdesk tickets with write tools stubbed before changes ship. Studies run AI analysis across thousands of tickets, and an AI Knowledge Builder turns past tickets into a searchable knowledge base. A REST API with API keys is also documented.
Community Discussions
Be the first to start a conversation about Macha AI
Share your experience with Macha AI, ask questions, or help others learn from your insights.
Pricing
750 tickets
Entry plan with 750 tickets per month; every plan includes all features and the success team.
- 750 tickets per month
- Unlimited agents and team seats
- Done-for-you setup
- 50 fair-use messages per ticket
- No automatic overage; agents pause when tickets run out
- Cancel anytime
1,500 tickets
Plan with 1,500 tickets per month.
- 1,500 tickets per month
- All features included
3,000 tickets
Plan with 3,000 tickets per month.
- 3,000 tickets per month
- All features included
5,000 tickets
Plan with 5,000 tickets per month.
- 5,000 tickets per month
- All features included
Capabilities
Key Features
- Done-for-you setup and deployment by the Macha team
- Safe mode with internal notes and human approval before replies go out
- Help desk ticket analysis ranking categories to automate
- Custom tools for any HTTP API
- Connectors for Shopify, Stripe, Slack, Notion, Google Workspace and Confluence
- Continuous agent evaluation of every conversation
- Agent simulations and Try it sandbox
- Sub-agents and multi-agent delegation
- Triggers for tickets, Slack messages, webhooks and schedules
- AI Knowledge Builder from past tickets
- Studies for AI analysis across ticket records
- Website chatbot and Chrome extension
- REST API with API keys
