# Break it in preview. Not in production.

Source: https://aide.app/platform/agent-simulator · Updated 2026-11-16

AI agent testing

Check a procedure on your last 50 conversations, replay a past one with the draft, and roll out inbox by inbox. After go-live, a scorecard grades Aide beside your own team every night.

Part of the Agent Governance Engine: nothing is sent to a customer or written to your helpdesk while you preview or watch.

[Start free](https://my.aide.app/register) [Get a demo](https://aide.app/demo)

## How do you test an AI agent before customers see it?

Run it on conversations you already have: in Aide you check a procedure against your last 50 conversations, replay a past conversation with the draft, and chat with the agent in a preview that never reaches customers. Then roll out inbox by inbox, and a scorecard grades Aide beside your own team every night.

Updated November 16, 2026

Everything you check before going live

- Check trigger
- Replay a past conversation
- Agent preview
- Compared with your team
- Watch only
- Test senders
- A reason for each near miss
- Edits stay a draft until Set live
- Drafting hint per topic level
- Times run this month, per procedure
- Sent, drafted or neither per inbox
- Previews never billed
- 50 Of your real conversations: Check trigger tries a procedure on your last 50 conversations before it goes live.
- Watch only Live traffic, nothing sent: Aide decides and drafts on incoming tickets inside Aide, and writes nothing to your helpdesk.
- Every night Evals after go-live: Test conversations replay against the agent's current setup, each one passing or failing.
- Trusted by customers on: [Zendesk★★★★★5.0 rating](https://www.zendesk.com/marketplace/apps/support/421826/aide/) [Shopify★★★★★5.0 rating](https://apps.shopify.com/aide)

Really great approach and strong answers out of the box. Ability to grow into your AI journey at your own pace starting with suggested responses, macro suggestions or precanned messages and then slowly automate once you're confident.

Zendesk Marketplace reviewZendesk

We currently use Zendesk and as such needed an AI solution that works with our existing CX platform and allows us to fully leverage our Shopify data to cut down support times and costs. Aide was a perfect fit as it integrates with both and seamlessly helped our agents get through tickets faster and automatically populates ticket fields on Zendesk!

Patchstore24Shopify

We have all the student and program information right at our fingertips, plus intelligent draft responses that we can personalize. Parents get faster, more accurate answers that still feel completely human and caring.

The team from LandoCustomer story

Test before deploy

## Watch it work before customers do

Every example below is a preview or a staged rollout, not a reply to a customer. This is what your team sees first.

### See which conversations a procedure fits

Check trigger tries a procedure's When to use on your last 50 conversations, lists the matches and declines, and gives a reason for each near miss. A misjudged one becomes an example.

### Replay a past conversation with the draft

Test replays a past conversation with the draft procedure, or starts a live chat inside it, so you see what Aide would ask, do and reply. Preview conversations never reach customers.

### Chat with the agent yourself

Write as a customer and push on the cases that worry you: a seat change mid-contract, a disputed charge, a course moved to a new date. The preview says what the agent is answering from.

### Compare with what your team sent

Connect a helpdesk and Aide drafts replies to recent tickets inside Aide. On a solved one, its draft sits above the reply your team actually sent, so you judge it by your own standard.

### Watch only on live traffic

Put one inbox, say billing or enrollments, in Watch only and Aide decides and drafts on real incoming messages inside Aide. Your team sees exactly what it would have done, and nothing is sent.

### Test senders first, then a share

Only the addresses you list get Aide's replies, so your team can write in as customers. Each inbox's Activity shows every write as applied or skipped, with the reason. Then a share of conversations, then everyone.

Aide is the AI-first customer service platform. New topics start at Drafts, and you choose when an inbox goes past Watch only. Read how [procedures](https://aide.app/platform/procedures) are written, how [every action is recorded](https://aide.app/platform/action-trace) and how [Aide improves every night](https://aide.app/platform/self-improving-ai-agents), or about [agent governance](https://aide.app/solutions/agent-governance) and [security at Aide](https://aide.app/security).

After you go live

## Tested before launch. Graded every night after.

Testing decides what goes live. The Quality page keeps it accountable once it does.

Picture: Graded beside your team

Graded beside your teamA pass or fail scorecard against a written quality bar, with Aide beside your team's median and top quarter.

Picture: Evals every night

Evals every nightEvals replay against the agent every night, and each change to how Aide writes is listed with an Undo.

Picture: Ready, topic by topic

Ready, topic by topicQuality says, topic by topic, when Aide's drafts are good enough to send, and you turn sending on.

Picture: On the record

On the recordThe topic, sources and actions behind every reply are kept, with two years of audit log.

Getting started

## Test on the conversations you already have

Connect your helpdesk or add Aide Chat, and testing runs on your own conversations. No migration and no engineering.

- [Zendesk](https://aide.app/integrations/zendesk): Drafts on recent tickets as soon as you connect.
- [Gorgias](https://aide.app/integrations/gorgias): Watch only, then test senders, per inbox.
- [Front](https://aide.app/integrations/front): Per-inbox rollout from Watch only to everyone.
- [Shopify](https://aide.app/integrations/shopify): Order and shipment conditions in every preview.
- [Salesforce](https://aide.app/integrations/salesforce): Contact fields in a procedure's Only if conditions.
- [Claude](https://aide.app/integrations/claude): Test the agent with Claude during setup.

Connect your own systems through an API or an MCP server, and the procedures that use them are checked the same way.

## Frequently asked questions

### What is AI agent testing?

AI agent testing shows you what an AI support agent would do before your customers see it. In Aide, testing runs on your own conversations: Check trigger tries a procedure's When to use on your last 50 conversations, Test replays a past conversation with the draft procedure, and the agent preview lets you write in as a customer.

### How do you test an AI support agent before it goes live?

In Aide you test on your own conversations. Check a procedure's When to use on your last 50 conversations, replay a past conversation with the draft, and chat with the agent in its preview. Then keep an inbox in Watch only before widening it to test senders, a share of conversations and everyone.

### Can we test Aide on our real conversations?

Aide tests on your real conversations. Check trigger tries a procedure on your last 50 conversations, and Test replays a past conversation with the draft. When you connect Zendesk, Gorgias or Front, Aide also drafts replies to up to 10 recent open tickets and 3 recently solved ones inside Aide.

### How do we judge Aide's quality before launch?

Before launch, Aide is judged procedure by procedure on your own conversations. Check trigger shows which recent conversations a procedure would pick up and why it passed on the near misses, and Test replays a past conversation with the draft. On recently solved tickets, Aide's draft sits beside the reply your team sent, so you grade it against your own standard.

### Does testing send anything to customers or our helpdesk?

Testing in Aide sends nothing to customers. A preview shows the reply Aide would write, or why it would hand off or hold back, and nothing is sent or written. An inbox in Watch only drafts inside Aide and writes nothing to your helpdesk.

### Can we let Aide answer a subset of tickets first?

Each helpdesk inbox in Aide moves from Watch only to test senders, where only the addresses you list get Aide's replies, then to a share of conversations and everyone. You can also put one topic at AI Agent and keep the rest at Drafts or None.

### How is Aide tested after it goes live?

After go-live, Aide's Quality page shows a pass or fail scorecard against a written quality bar. Evals replay against the agent every night, and Aide's replies are graded beside your team's median and top quarter on the same scorecard. Quality also says, topic by topic, when Aide's drafts are good enough to send.

### What happens when a teammate marks a draft wrong?

In a conversation in Aide, a teammate marks a draft Good or Wrong with a reason, and that verdict overrides Aide's own review. A Wrong draft becomes an eval case that replays against the agent every night, so every later change is checked against it.

### Who can check and set procedures live?

Owners and admins run Check trigger in Aide and set a procedure live with a note. Edits to a live procedure stay a draft while the live version keeps running, and version history restores an earlier one.

### Do previews count as AI replies?

Previews and spam are never counted as AI replies in Aide. Replies to test senders count like any reply, and drafts kept inside Aide during Watch only count once per customer message, like any draft.

## Set it up

- [Test your agent before going live](https://docs.aide.app/aide-ai-agent/test-agent)

## Governance and trust, defined

The vocabulary of testing and trusting AI agents, defined.

- [Action trace](https://aide.app/glossary/what-is-an-action-trace)
- [Agent simulation testing](https://aide.app/glossary/what-is-agent-simulation-testing)
- [Agent Simulator](https://aide.app/glossary/what-is-an-agent-simulator)
- [AI agent evaluation](https://aide.app/glossary/what-is-ai-agent-evaluation)
- [AI agent sandbox](https://aide.app/glossary/what-is-an-ai-agent-sandbox)
- [AI audit log](https://aide.app/glossary/what-is-an-ai-audit-log)
- [AI escalation policy](https://aide.app/glossary/what-is-an-ai-escalation-policy)
- [AI explainability](https://aide.app/glossary/what-is-ai-explainability)
- [AI governance for customer service](https://aide.app/glossary/what-is-ai-governance-for-customer-service)
- [AI grounding](https://aide.app/glossary/what-is-ai-grounding)
- [AI guardrails](https://aide.app/glossary/what-are-ai-guardrails)
- [AI hallucination](https://aide.app/glossary/what-is-an-ai-hallucination)

## Test Aide on your own conversations, free

Connect your helpdesk and see Aide's drafts on your recent tickets in minutes.

Not ready yet? Get product updates by email
