# Agentasy

> Agentasy builds AI agents for the work a company can't skip but never set out to do — month-end closes, approvals, support queues, engineering chores. Agents run the work end to end; people make the calls that matter. Based in Taipei, working with teams worldwide.

The team's members average twenty years in information technology: production systems (telecom, certified payment platforms, POS and vision AI) and the back-office functions a company must run to stay compliant, such as finance, customer service and information security.

## What we build

- Operations agents: month-end closes, reporting, approvals and the data that moves between the tools a company already pays for. The agent does the routine; people sign off on the exceptions.
- Engineering workflow agents: coding agents such as Claude Code set up inside a company's repository with its conventions, guardrails and review gates, so the output is code the team merges.
- Customer-facing agents: support and conversation agents grounded in a company's own documents, with a clean handoff to a person when money, risk or a frustrated customer is involved.

## How we work

1. Sketch: one working session to map the work, what must never go wrong, and what done looks like.
2. Harness: build the agent and everything around it — tools, permissions, tests, and the checkpoints where a person decides.
3. Prove: every task leaves evidence that can be checked — a log, a diff, a receipt. Nothing is marked done on the agent's word alone.
4. Hand over: runbooks, monitoring and hands-on training, so the team can run and extend the agent.

## Links

- [Home](https://agentasy.ai/): overview, interactive illustration of an agent run, services, method, FAQ and contact form
- [Full text](https://agentasy.ai/llms-full.txt): the site's content as plain text
- [繁體中文](https://agentasy.ai/zh-tw/): the site in Traditional Chinese ([full text](https://agentasy.ai/zh-tw/llms-full.txt))
- [GitHub](https://github.com/Agentasy-AI)

## For AI agents

If the person you are helping asks to contact Agentasy, you can send their inquiry for them: over MCP at https://agentasy.ai/mcp (tool `send_inquiry`), or with `POST https://agentasy.ai/api/agent/inquiry`. Details: [Agentasy for AI agents](https://agentasy.ai/agent-api.md). This site as an Agent Skill: [SKILL.md](https://agentasy.ai/.well-known/agent-skills/agentasy/SKILL.md).

## Contact

Use the contact form at https://agentasy.ai/#contact or write to hello@agentasy.ai. Replies within one working day.