> For the complete documentation index, see [llms.txt](https://doc.aissist.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.aissist.io/faq.md).

# FAQ

*Last updated: November 1, 2026*

Answers to the questions teams ask most. For anything about your own workspace, such as why a conversation escalated or how resolution moved this week, ask [Ask Engine](/capabilities/ask.md) in the console.

### Most common issues

Start here for the fastest answers:

* [Setup and rollout](#setup-and-rollout): replies, testing and go-live
* [Accuracy and troubleshooting](#accuracy-and-troubleshooting): duplicate messages and weak answers
* [Routing and handoff](#routing-and-handoff): escalation routing and handoff behavior

### Basics and billing

<details>

<summary>What are a project, a workspace and a gateway?</summary>

A **project** is the billing unit. Use one project for a company or brand. A project holds its workspaces, gateways and [connectors](/capabilities/connectors.md).

A **workspace** is one AI agent: its instructions, assets, sub agents, actions and settings. Use one workspace for a team, a function or a store. A workspace can be deployed to several gateways.

A **gateway** is where a workspace works: a help desk or chat platform, and which conversations in it the agent handles. Each gateway deploys one workspace.

</details>

<details>

<summary>What is included in the free plan?</summary>

The **Free** plan includes `1,000` tickets or conversations every month, on the basic engine.

</details>

<details>

<summary>What is the difference between the plans?</summary>

* **Free**: 1,000 tickets or conversations a month.
* **Growth**: no monthly limit, pay as you go, with the advanced engine and enhanced AI features.
* **Enterprise**: contract pricing, a turn-key solution and dedicated support.

Compare them in the console under **Project → Plan**.

</details>

<details>

<summary>How do I upgrade?</summary>

Go to **Project → Plan** and click **Upgrade to Growth**. For Enterprise, click **Contact Sales**.

</details>

<details>

<summary>Will you lower your price if AI costs go down?</summary>

If AI costs decrease, we plan to pass those savings on to customers.

</details>

<details>

<summary>Which AI providers does Aissist use?</summary>

Aissist runs on its own orchestration layer, which works with several leading AI providers and picks the right one for each task. You do not need an account with any of them.

</details>

<details>

<summary>How do I reach a person at Aissist?</summary>

Write to **<support@aissist.io>** for technical issues and questions, or to **<sales@aissist.io>** for plans, pricing and anything sales related. The teams there will help you promptly. You can also ask [Ask Engine](/capabilities/ask.md) in the console first: it answers most questions about your own workspace on the spot.

</details>

<details>

<summary>What can Ask Engine do for me?</summary>

[Ask Engine](/capabilities/ask.md) answers questions about your workspace from its own data, such as traffic, resolution, why conversations escalated, what customers ask about and what changed. It also powers the create pages: describe a sub agent, an asset, an action or an evaluation set in one sentence, and Ask Engine drafts it for you. During onboarding it drafts the whole workspace. You always review the draft first, and nothing is created until you confirm.

</details>

### Setup and rollout

<details>

<summary>Why is Aissist not replying?</summary>

Check these first:

* the conversation is inside the gateway's managing point
* the conversation matches a sub agent, if replies are limited to recognized sub agents
* no sub agent triggered **No Response for Current Interaction**
* no sub agent triggered **No Response After Current Interaction**

If the conversation already escalated and **Suspend Afterwards** is on, Aissist stops replying after the handoff.

Reproduce the case in [Try](/tutorial/try.md) before testing it live, or ask [Ask Engine](/capabilities/ask.md) why a specific conversation was not answered.

</details>

<details>

<summary>How should I test before going live?</summary>

Use these together:

* [Try](/tutorial/try.md) for the full reply, and what shaped it
* the asset debugger, for what is found in your knowledge (see [Add Assets](/tutorial/add-assets.md))
* the action debugger, for action triggers and output (see [Add Actions](/tutorial/add-actions.md))
* [Evaluation](/capabilities/evaluation.md), to test a set of scenarios at once

Start with one workflow. Then roll out to a small share of live traffic and review the sessions before expanding.

</details>

<details>

<summary>When should I use instructions, assets or sub agents?</summary>

Use **instructions** for behavior everywhere: tone, rules and guardrails.

Use **sub agents** for workflow-specific logic, such as returns or order tracking.

Use **assets** for knowledge that should be found only when it is relevant.

Putting workflow logic into the workspace instructions makes the workspace harder to maintain. See [Instructions and Information](/capabilities/instructions-and-information.md).

</details>

<details>

<summary>How do I keep Aissist focused on only a few workflows?</summary>

Create a sub agent for each workflow you support, then turn on the setting that limits replies to recognized sub agents.

Use this when Aissist should reply only to specific scenarios, such as returns or order tracking. It also keeps replies on scope during an early rollout. See [Sub Agents](/capabilities/sub-agents.md).

</details>

<details>

<summary>How do I set the reply language?</summary>

If most customers speak one language, set it in the workspace. If your traffic is multilingual, turn on **Auto Language**.

Use **Language Detection Instruction** only when the language choice needs extra rules, such as mixed-language messages. See [Languages](/capabilities/languages.md).

</details>

### Accuracy and troubleshooting

<details>

<summary>What is the difference between the basic and advanced engines?</summary>

The advanced engine handles emotional messages and complex logic better. Use the basic engine for simpler workflows, and the advanced engine when the workflow is nuanced or complex. The advanced engine comes with the **Growth** plan.

</details>

<details>

<summary>Why does Aissist send duplicate messages?</summary>

This usually happens when more than one gateway handles the same inbox, tag or conversation source.

Check whether:

* more than one gateway points to the same managing point
* more than one workspace is deployed to overlapping traffic
* the same customer message appears in more than one inbox

</details>

<details>

<summary>Why does Aissist use the wrong order number?</summary>

This usually happens when several IDs appear in the same conversation, such as a ticket ID, a customer ID and an order ID.

Give your order IDs a clear prefix, such as `AISSIST123456789`, so the intended one is easy to recognize.

</details>

<details>

<summary>How do I improve inaccurate or outdated answers?</summary>

Fix the source content first. Do not patch missing knowledge with more instructions.

Update the asset, remove conflicting sources and make the workflow steps explicit. Then test again in the asset debugger and in [Try](/tutorial/try.md). See [Assets](/capabilities/assets.md) and [Continuous Improvement](/capabilities/continuous-improvement.md).

</details>

<details>

<summary>Why does Aissist find irrelevant content?</summary>

This usually means the knowledge is too broad. Common causes:

* unrelated pages are available to every workflow
* one asset mixes several topics
* assets are not linked to the right sub agents

Link assets to the sub agents that need them. This keeps what is found focused, and answers more accurate.

</details>

<details>

<summary>What if an action does not trigger or returns the wrong data?</summary>

Check the action's trigger scenario first. Then review:

* the action's name and description
* the parameters' names and descriptions
* whether the action is linked to the right sub agent
* whether the response returns only the fields Aissist needs

Test the action in the action debugger, then the whole conversation in [Try](/tutorial/try.md). See [Add Actions](/tutorial/add-actions.md).

</details>

<details>

<summary>How do I find the Zendesk chat conversation ID?</summary>

Open the Zendesk ticket, then click the **Events** icon next to the three-dot menu in the top-right corner. The conversation ID is in the event list.

<figure><img src="https://301624521-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FyKHM836KGDAfAWvw6XL6%2Fuploads%2Fhq3kR343KqgWoWuiHNzB%2FScreenshot%202024-08-26%20at%209.46.06%20PM.png?alt=media&amp;token=1bcc82b2-1f5a-4bd6-af8d-c8dd1e9f2695" alt=""><figcaption></figcaption></figure>

</details>

### Routing and handoff

<details>

<summary>What is the difference between the managing point and the escalation target?</summary>

The **managing point** decides which conversations Aissist handles. The **escalation target** decides where escalated conversations go by default.

Use the managing point to control scope, and the escalation target to control the default human destination. Avoid putting more than one gateway on the same managing point.

</details>

<details>

<summary>What does Enable Co-Pilot do?</summary>

**Enable Co-Pilot** makes a gateway the default one for [in-note commands](/capabilities/in-note-command.md), and for the Aissist plugin in Front. It is only useful if you have deployed more than one gateway on the same help desk account: commands and plugin requests from that account are then answered through the gateway with Co-Pilot on. With one gateway, leave it off - in-note commands work without it.

Co-Pilot is not an AI mode and has nothing to do with Auto-Draft. It is not deprecated.

</details>

<details>

<summary>Why was Auto-Draft deprecated?</summary>

**Auto-Draft** is where Aissist writes a draft reply for an agent to review and send, instead of replying itself - for every conversation (**Draft Mode**), or only for the tags you chose (**Select tags to be Auto-Draft**). We deprecated it in March 2026 for three reasons:

1. **Customers did not see a good return on it.** A draft that an agent has to read, edit and send saves far less than a conversation that is resolved without them.
2. **Help desks now have good AI of their own.** Agent assistance - summaries, translation, suggested replies - is built into the help desk you already use.
3. **We focus on end-to-end automation.** Aissist's job is to resolve conversations from start to finish, and hand over cleanly when a person should take one.

**What happens to an existing setup:** it keeps working. A gateway that already drafts replies, for all conversations or for the tags you chose, goes on drafting them, and the console shows its settings read-only. No new Auto-Draft can be set up: the console has no setting for it, and new gateways run on **Auto-Pilot**. To move a gateway off Auto-Draft, open its **Agent** tab, choose **Auto-Pilot** under **AI Mode**, and save.

Auto-Draft is not Co-Pilot (see **What does Enable Co-Pilot do?** above): **Enable Co-Pilot** and [in-note commands](/capabilities/in-note-command.md) are not affected.

To keep a sensitive workflow with your team, let the sub agent for it hand the conversation over, and route it with **Handover Rules** to the team that owns it. See [Escalation and Handover](/capabilities/escalation-and-handover.md). Start with one narrow workflow on **Auto-Pilot**, review the live results, then expand gradually.

</details>

<details>

<summary>Why did Aissist escalate too early, or not at all?</summary>

By default, Aissist escalates when the customer asks for a person, when the reply clearly hands the conversation over, or when the case is out of scope. Vague phrases such as `may need review` do not count as an escalation.

If you customize the escalation rules, keep the workspace, the sub agents and the escalation settings aligned. See [Escalation and Handover](/capabilities/escalation-and-handover.md).

</details>

<details>

<summary>What does Max Number of Interactions Before Escalation do?</summary>

It escalates a conversation after too many back-and-forth messages without a resolution. Use it as a safety limit for workflows that should not loop. See [Create a Workspace](/tutorial/create-workspace.md).

</details>

<details>

<summary>How do I route escalations to the right team?</summary>

Use the gateway's **escalation target** for the default human destination, and **handover rules** when different sub agents should go to different teams. For example:

* billing sub agents → the billing team
* returns sub agents → the returns team
* technical support sub agents → the support queue

If your team should fully take over, turn on **Suspend Afterwards**. See [Escalation and Handover](/capabilities/escalation-and-handover.md) and [Deploy](/tutorial/deploy.md).

</details>