ChatGPT Plugin

Aide

AI customer service platform

Brand: Aide

Category: Customer Support
Primary Subcategory: AI Chat & Messaging Agents

Integration details

Description

Aide is an AI customer service platform. Its AI agent answers your customers in a chat widget on your website and by email, or inside Zendesk, Front and Gorgias, from your knowledge base and Shopify orders, and hands what it can't resolve to your team with a summary. This plugin connects ChatGPT to your Aide workspace. You sign in as yourself, and ChatGPT can do what your role in Aide allows. Set up Aide from ChatGPT: - Check what's left before Aide answers your customers, and do the next step together: company details, knowledge, a test of the agent, a channel, then going live. - Teach Aide your website or help center, import a site again, and write articles for what the site doesn't say. - Change the agent's name, instructions, greeting and signature, and test it with customer questions in a private preview no customer sees. - Get the chat widget snippet to paste on your site, add your support email address and test that it forwards to Aide, or get a link to connect Gmail, Shopify, Zendesk, Front or Gorgias in your browser. Choose where Aide's drafts appear in a connected helpdesk. - Deploy the agent, choose which inboxes it answers, and pause it at any time. Work your support queue: - Find conversations by keyword, status, priority, assignee, team, topic, channel or helpdesk ticket id, and read their messages, Aide's reply draft, the customer's orders and CRM details. - Add internal notes, fill in a macro, change a conversation's status, priority, assignee or team, and reply to the customer on the conversation's channel. Knowledge and orders: - Search your help center articles, website pages, products and macros, and read them in full with a link to the source. ChatGPT can also search and cite them as company knowledge. - Look up a customer's Shopify orders by email or order number, with status, items, fulfillment and tracking, and check what on an order can be returned. - Refund or cancel an order on your Shopify store, and approve or reject order changes waiting for approval. A change above your approval limit waits for an admin. Analytics: conversation volume, AI resolution, handoffs, response times and customer satisfaction. Who it is for: owners, admins and agents of an Aide workspace. Setting up, changing, testing and deploying the agent, and analytics, are for owners and admins. How access works: you sign in with your Aide account and approve the connection on Aide's consent screen, which shows what ChatGPT asked to do: read only, or read and act. ChatGPT sees only what you can see in Aide, and every action is recorded as you. Tools that message a customer, change an order, deploy the agent or turn on AI replies for an inbox are marked destructive, so ChatGPT can ask you to confirm before using them. You can disconnect ChatGPT at any time in Aide under Settings > Developers. Limitations: - Requires an Aide account. The plugin works only in the workspace you approve when you sign in. - Order lookups and order actions need a Shopify store connected to Aide. - You finish connecting Gmail, Shopify or a helpdesk on that service's sign-in page in your browser, and you paste the chat snippet on your site and turn on forwarding in your email provider yourself. - A reply sent to a customer cannot be unsent, and a refund cannot be reversed. - The plugin cannot delete conversations or manage teammates.

Integration type: Plugin
Verification status: Not applicable
Platform: ChatGPT
Category: Customer Support
Primary Subcategory: AI Chat & Messaging Agents
Secondary Subcategories: None listed
Brand: Aide
Access: Account required
First tracked: 2026-11-04
Tool count: 91
Market: US

Agent DiscoverabilityChatGPT Plugin Discovery Score

ChatGPT organic discovery is live

Your score is coming

ChatGPT now suggests Plugins on its own when they match a user's request.Your Plugin Discovery Score measures how often yours appears, and it will show here as soon as it’s ready.

ChatGPT recommends Airtable when a user asks it to update a CRM. — ChatGPT surfaces Airtable because it matches the user's request.

Primary Subcategory

Competing in ChatGPT AI Chat & Messaging Agents

View Category

Listed in alphabetical order

Primary Subcategory

Actually AskedComing soonGet notified

AideComing soonGet notified

Alhena AIComing soonGet notified

AsyntaiComing soonGet notified

BedonesComing soonGet notified

Tools

91 tools agents can invoke

Write knowledge Aide answers customers from (a policy, shipping times, an FAQ…) in the workspace's articles. An `article` (the default) is customer-facing: published at once to Aide Chat's public help section unless saved as a draft. A `snippet` is internal: never shown to customers, used by Aide as soon as it is saved, with no draft. Use a snippet for notes only Aide should read, an article for help customers should be able to read. Use it for what the website doesn't say, or what the user tells you directly. Returns the document with its `id`; once published it is used in answers.
add_knowledge_article
Set up Aide on the address customers email (such as support@example.com). For a Gmail or Google Workspace inbox, prefer connect_integration with `gmail`: no forwarding. Returns the mailbox with its `id` and `inboundAddress` for the user to forward the address to, then call send_forwarding_test. Gmail: Settings, See all settings, Forwarding and POP/IMAP, Add a forwarding address (Gmail emails Aide a confirmation code: get_support_email), then Forward a copy of incoming mail, Save Changes. Outlook or Microsoft 365: Settings, Mail, Forwarding, Enable forwarding, Save; a Microsoft 365 admin may have to allow external forwarding.
add_support_email
Add an instruction at the end of the list: a rule or fact for every conversation, not one kind of message (that is a procedure) and not tone. It can't run actions, tag or hand off. Active instructions apply to the next AI reply on every conversation they cover; replies read them in order, up to 30 and 4,000 characters. Returns the instruction.
create_agent_instruction
Add an internal note to a conversation, as the signed-in teammate. Customers never see notes.
add_note
The workspace's support report over a date window: conversation volume by day, AI resolution rate (confirmed, assumed, reopened), automation, handoffs by reason, response and resolution times, CSAT for Aide and for teammates, channels, topics and sales from chats, each metric with the previous window of the same length. The last 30 days when no window is given; optionally one channel or topic.
analytics_overview
Count the workspace's conversations over a date window: all of them (`conversations`), those with a detected topic (`conversationsWithTopic`), those a scenario ran on (`conversationsWithScenario`) and those where a teammate used Aide's draft (`conversationsWithDraftUsed`), each as `{ value, previous }` with the previous window of the same length. The last 30 days when no window is given. For resolution, handoffs and response times use analytics_overview; for satisfaction, analytics_satisfaction.
analytics_summary
Approve an order action waiting for approval (status `pending_approval`), as the signed-in teammate: Aide runs it on the Shopify store now, so a refund is paid and a cancellation happens. Agents can approve only refunds within the workspace's approval limit. The result is the action with its new status. Calling it again does not run the action twice: once it is no longer waiting for approval, the call is refused with `not_pending_approval`.
approve_order_action
Cancel an order from a conversation on the Shopify store, as the signed-in teammate. The workspace's approval rules apply: a cancellation above their limits waits for an admin, and the result says whether it ran or waits. Repeated calls are not deduplicated: an order already cancelled is refused, but calling it again while a cancellation waits for approval adds another request.
cancel_order
Change who gets sent replies on a support mailbox and its quiet period; only the fields given change. A wider rollout emails Aide's replies to those customers from then on, and sent emails can't be recalled.
update_support_email_settings
Try a When to use (saved or not) on the last 50 conversations with a customer message on the procedure's channels: which it would match and decline, with a reason for near misses. Uses AI on each conversation, at most 20 checks an hour. Returns an `id` and `status`; poll get_procedure_trigger_check until `ready` or `failed`.
check_procedure_trigger
Check live with Shopify what on an order from a conversation can be returned: each line's returnable quantity, whether the return rules block it and the return reasons to choose from, the policy applied and whether a return is already open. `permission: missing` means the store has not granted Aide the returns permission. Changes nothing.
check_return_eligibility
Start connecting Gmail, Shopify, Zendesk, Front or Gorgias. Returns a dashboard `url` that opens the sign-in for that service, and `tellUser`: give the user both, then check get_setup_status once they are done. Pass Zendesk `subdomain`, Gorgias `account` or Shopify `shop` when known; otherwise the page asks. set_up_helpdesk finishes a helpdesk (Zendesk, Front, Gorgias) once connected.
connect_integration
Send a follow-up message in a preview conversation test_agent started, and wait (up to about 25 seconds) for the agent's answer. With pending: true, read the answer with list_messages after a few seconds; don't ask again. When `stillLearning` is set, Aide is still reading the website and found nothing yet: tell the user, ask them meanwhile for the facts customers ask most and save each with add_knowledge_article, and test again after the minutes it gives. When `sources` is empty otherwise, the agent found nothing to answer from: add it with add_knowledge_article and test again.
continue_agent_test
Customer satisfaction (CSAT) over a date window, for Aide's conversations and teammates' separately: the score with the previous window's, positive ratings, the rating distribution, and customers' comments with their conversation ids. The last 30 days when no window is given; optionally one channel or topic.
analytics_satisfaction
Delete a procedure; it stops applying at once. To pause it, update_procedure.
delete_procedure
Delete an instruction; replies stop following it at once. To keep it but stop applying it, update_agent_instruction with isActive false.
delete_agent_instruction
Put an AI agent live: from now on it answers real customers on the channels it is assigned to (the chat widget, and the inboxes set with set_inbox_ai). Confirm with the user before calling it. Returns the agent with status `deployed`; pause_agent stops it.
deploy_agent
Throw away a procedure's draft; what is live stays as it is. Returns the procedure.
discard_procedure_draft
Dismiss an open knowledge gap whose fix the user doesn't want, with the reason, so Aide stops proposing it. Returns the gap.
dismiss_knowledge_gap
Count how many recent conversations some Only if conditions (saved or not) would let through. Returns an `id` and `status`; poll get_procedure_estimate until `ready` for `result.estimate` and a sample.
estimate_procedure
Read one knowledge document in full by the id search returned: its `title`, full `text`, `url` and `metadata` (source, collection, kind, status, last update).
fetch
Fill a macro's reply for a conversation, its placeholders (customer, order, teammate…) resolved, without running the macro or sending anything. Returns the text as `body` and the placeholders that could not be filled as `unresolved`. Send it with reply_to_conversation.
render_macro
Find conversations by a helpdesk's own id (externalId), keywords (q) or filters, which combine. Each has its id, subject, status, priority, channel, requester, assignee, team, a preview and timestamps. Next page: `metadata.nextCursor` as `cursor`, same filters and sort.
find_conversation
Read one conversation: its subject, status, priority, customer, assignee, team, topics, channel, handoff state and custom fields. Read its messages with list_messages.
get_conversation
Read the CRM records (Salesforce contacts, accounts or leads) matched to a conversation's customer, with the fields the workspace chose to show and how they matched. Empty when no CRM is connected or nothing matched.
get_customer_context
Read the store customer matched to a conversation (how they matched and whether the match is verified) and their Shopify orders, newest first, each with its id, number, status, totals, what is still refundable, items, fulfillments, tracking and shipping address. For chats, also the product or collection the shopper is viewing and their cart. Use an order's id with refund_order, cancel_order and check_return_eligibility.
get_conversation_orders
Read one knowledge document in full: its title, link, source, collection, status and body.
get_knowledge_document
Read one knowledge source, such as a website add_website started: its `status`, import progress, `documentCount` and the error when an import failed. Use it to tell the user when Aide has finished learning a site.
get_knowledge_source
Read one macro: its name, description and actions (the reply it posts and the status, tag, assignment and other changes it makes).
get_macro
Read how Aide answers a support mailbox: rollout (who gets sent replies) and quiet period, with their `sources`.
get_support_email_settings
Read one procedure in full: its When to use and examples, Only if conditions, steps, finish, what it does when the customer changes subject, `status`, and `draft` (the edits not live yet, in the same shape, or null). Read it before changing it.
get_procedure
Read a procedure backfill: its `status`, `preview` counts (matching, skipped actions), `expiresAt`, and once run its `result`.
get_procedure_backfill
Read an estimate estimate_procedure started: its `status`, then `result` (matched, estimate, sample conversations).
get_procedure_estimate
Read a preview preview_procedure started: its `status`, then `result` (the action, the draft, whether the procedure applied).
get_procedure_preview
Read a check check_procedure_trigger started: its `status`, `error` when failed, then `result` (the threshold, how many matched, the tokens used, and each conversation with its score, whether it matched and the reason).
get_procedure_trigger_check
Read one synced Shopify product in full: its description, variants with prices and stock, images, collections, tags, metafields and storefront and admin links.
get_product
Read one support mailbox: its `supportAddress`, `inboundAddress` and forwarding status, with the confirmation code Gmail sent Aide when forwarding was added (`forwarding.confirmationCode`) for the user to enter in Gmail's settings.
get_support_email
Read one AI agent in full: its name, status, instructions, greeting, signature, contact information and the knowledge it uses. Read it before update_agent to change the instructions rather than replace them.
get_agent
Check how far the workspace is from Aide answering its customers: whether it is `live` (Aide has answered or drafted for a real customer), the `nextStep`, and each step (`profile`, `knowledge`, `agent_tested`, `channel`, `agent_deployed`, `live`) with its `status` (`done`, `current`, `todo`, `skipped`), a `detail` saying what to do (for an unfinished `profile`, the questions still to ask), the `tool` that does it and a dashboard `url`, for when the user prefers the dashboard; prefer the tool. Call it first when the workspace is new or not live, then do the steps in order, past any that waits on an import or the user, calling it again after each.
get_setup_status
Get the workspace's Aide chat page: a public link (`url`) where anyone chats with Aide, with nothing to install, and `tellUser`. Offer it right after test_agent: give the user the link and `tellUser`, so they send it a message and see Aide answer live, then share it with customers (a bio, an email signature, order emails, a help center). `answering` false: the agent isn't deployed yet; ask for the OK and call deploy_agent in the same turn. `installSnippet` is the same chat for their own website, second: the script tag to paste before `</body>` on every page, or in the site builder's custom code setting.
get_chat_page
Read Aide's current reply draft for a conversation (the Copilot draft teammates see), with its sources. Answers `data: null` when the conversation has no draft.
get_draft
Who this connection is: the workspace, the app and signed-in teammate, their role, whether it may only read or also make changes, and what needs the user's attention (a connection to sign in to again, a support email to finish, a paused agent or AI replies). Call it when a call is refused or before explaining what you can do.
get_connection
Import a website or help center source again, such as one whose import failed, in the background: its pages at and under `url` (the source address when omitted, on the same site; another site is add_website). Returns the source with `status` `importing`; check it with get_knowledge_source.
reimport_knowledge_source
Teach Aide a website or help center: Aide crawls the pages at and under `baseUrl` in the background and answers customers from them once imported. Use kind `website` for the company's site and `help_center` for a Zendesk, Intercom or similar help center. Returns the knowledge source with its `id` and `status`; it imports in the background for a few minutes: say so and carry on, checking it later with get_knowledge_source. A site already added (saving the workspace's website adds it) is refused (`duplicate`): find its id in get_setup_status or list_knowledge_sources; retry it with reimport_knowledge_source.
add_website
Turn on AI replies on an inbox (a support mailbox, a Gmail or helpdesk inbox) by assigning it to this AI agent, replacing the agent that answered it before. While the agent is deployed, it answers that inbox's customers (or drafts replies, where the inbox's `replies` is `draft`). Confirm with the user first. Answers `ai_unavailable` for an inbox Aide can't answer (`answerable: false`). Returns the inbox.
set_inbox_ai
List the workspace's AI agents (every workspace starts with one) with their ids, names and status (`draft`, `deployed`, `paused`). Use an agent's id with get_agent, update_agent, test_agent and deploy_agent.
list_agents
List the order actions (refunds, cancellations, address changes, returns and the like) taken or requested on a conversation's orders, newest first, each with its status (`pending_approval` waits for an admin), amount, order, reason, result and who asked.
list_conversation_order_actions
List the workspace's Instructions: short rules and facts AI agents follow in every conversation, as policy they can state to customers (tone and style are the agent's settings instead). Each has its `id`, `body`, `channels` (empty: all), `agentId` (null: every agent), `position`, `isActive` and `source`, in the order replies read them.
list_agent_instructions
List the inboxes an AI agent can answer across every channel (the chat widget, support mailboxes, connected helpdesk and Gmail inboxes), each with its `id`, `name`, `medium`, whether this agent is `assigned`, whether Aide can answer it (`answerable`) and whether replies are sent or drafted (`replies`). Use an id with set_inbox_ai.
list_agent_channels
List what AI apps changed in the workspace, newest first: each change's action, what it changed and the teammate or API key they acted as. It does not tell apart which AI app made a change. Pass `metadata.nextCursor` as cursor for the next page.
list_ai_app_changes

FAQs

Aide ChatGPT Plugin FAQ

How the directory, categories and Discoverability Score work.

Read the methodology

How do I improve Aide's ChatGPT Plugin discoverability?

The levers are the listing surface agents actually read: names, descriptions, keywords, tool metadata, and registry health. Which lever matters depends on where discovery breaks, which is what continuous measurement shows.

What are Aide alternatives on ChatGPT?

As of 2026-11-04, Aide competes with Actually Asked, Alhena AI, Asyntai, Bedones, Bravos AI, BubblaV AI Chatbot, call4me, Capi Agent for Merchants and 43 more in ChatGPT AI Chat & Messaging Agents, ranked by public Discoverability Score.

Where is this profile measured?

This profile uses the geography attached to the latest public registry snapshot: US. Locale tags are intentionally omitted.