Not Just a Chatbot.
A Complete Support Workforce.
Connect your channels, train your brain, and automate complex actions in 4 simple steps.
Most enterprise AI tools require months of implementation and a dedicated engineering team. Hay.chat is built on a linear "Connect-Train-Act" engine designed for immediate impact.
You don't build a bot; you hire a digital employee that arrives ready to learn.
Connect Your Ecosystem
The Foundation
Hay.chat lives where your customers are and works with the data you already have. In just a few clicks, you bridge your communication channels with your business intelligence.
-
Channels: Deploy instantly on Web Chat, or connect asynchronous channels like WhatsApp and Slack (Coming Soon).
-
Data Sources: Plug directly into Shopify, WooCommerce, or Zendesk to give your AI real-time visibility into order status and ticket history.
Ingest Knowledge Instantly
The Brain
Forget manual data entry or complex flowcharts. If you have documentation, you have a trained agent.
-
Any Format: Upload your existing PDFs, CSVs, and text files, or simply point Hay.chat to your Help Center URL.
-
Instant Indexing: The moment you hit upload, the Agent learns. There is no "retraining" downtime.
-
Checked Answers: Answers about your business are checked against your documents. If your documents can't back an answer, the Agent hands it to a person rather than guessing.
Define Your Agent's Voice
The Persona
Your brand isn't generic, so your AI shouldn't be either. Configure Hay.chat Agents to act as true extensions of your support team.
-
Tone Control: Select from presets like Professional, Casual, or Enthusiastic, or write a custom system prompt to match your specific brand voice.
-
Human Guardrails: You decide the autonomy. Set Agents to "Auto-Send" for speed or "Require Approval" to review drafts before they reach the customer (how human approval works).
-
Smart Handoff: If confidence drops or a customer expresses frustration, Hay.chat hands off to a human instantly, with full context.
Execute Real-World Actions
The "Secret Sauce"
Refunds
If a customer asks for a refund, check the order status first with 
Get Order.
Process Refund
Create Ticket
Send Message
This is where Hay.chat leaves standard chatbots behind. It answers the question and then does the task, using Playbooks you write in plain language. Our AI support agent glossary page explains the difference.
-
Text-Based Logic: No complex visual graphs. Write Playbooks in plain language (e.g., "If customer asks for refund...").
-
The Magic @ Key: Simply type @ in your Playbook to summon your integration tools.
-
Real Execution: When triggered, Hay.chat securely calls the API to process a Shopify refund, create a Zendesk ticket, or check inventory in real-time.
What stops it making things up
Every answer that states a fact is checked before it's sent: a claimed action has to match a tool that actually ran, invented products and policies are blocked, and claims about your business are scored against your documents. Answers your documents can't back go to your team instead of the customer. No AI can promise it'll never be wrong, so Hay.chat is built to notice when it might be.
See how the guardrails work
Read replies before they're sent
Switch on approval for an agent and its replies wait as "Pending" until someone on your team approves, edits or blocks them. It's off by default and on every plan. Refunds and other actions follow your playbook rules instead.
See how human approval worksFree your team from repetitive tickets
Let Hay.chat handle WISMO queries and routine requests while your team focuses on VIP customers and complex issues that need a human touch.