Reference
Feature flags
Some HelpStack capabilities are switched on per organization rather than shipped to everyone at once. This page lists every one of them, what it does, and how you get it — so that "is this available to me?" has an answer you can look up instead of asking.
Three things decide whether a capability is on for you:
- On for everyone. Nothing to ask for.
- Included from a plan up. It arrives when you are on that plan or higher.
- Available on request. Off until we switch it on for your organization. Some of these are in beta; some are held back deliberately because of what they can do. Ask us and we will tell you which applies to you.
If you are on Growth or above and want something in the request list, ask. That is what the list is for. We would rather switch a flag on for a business that needs it than have you conclude HelpStack cannot do it.
An explicit setting for your organization always wins, in both directions — a capability included with your plan can still be switched off for you if you ask.
On for everyone#
| Capability | What it does |
|---|---|
| Shopify integration | The one-click "Connect Shopify" option, and Shopify order data made available to the AI and the agent sidebar. |
| WooCommerce integration | The WooCommerce connect option, and WooCommerce order data made available to the AI and the agent sidebar, through the HelpStack WordPress plugin. |
Both still need the corresponding connection to be configured before anything appears, so the button never shows up somewhere it could not work.
Included with a paid plan#
| Capability | Plan | What it does |
|---|---|---|
| MCP (set up HelpStack by asking an AI assistant) | Starter | Issue an API token and connect Claude or another assistant to configure your knowledge base, FAQs and agent tools by describing what you want. See MCP. |
| MCP conversation access (read the inbox, draft replies) | Growth | Lets an MCP token be granted the conversation scopes, so an assistant can read customer conversations and write drafts for a person to approve. It can never send. An admin still chooses per token whether to grant it. See MCP. |
| Advanced analytics | Growth | Response and resolution times, the AI-versus-human split, per-channel and per-agent breakdowns, and usage trends. |
| Conversation tagging | Growth | A curated tag taxonomy, manual tagging, AI classification of each conversation, and tag filtering in the inbox. The classification cost is covered by the plan price rather than metered against your credits. |
Each arrives when you reach that plan or higher, and can still be switched off for your organization if you ask.
Available on request#
Channels and connections
| Capability | What it does | Notes |
|---|---|---|
| Voice channel | The real-time AI voice agent, as a standalone voice widget. | Beta, and working — it handles real conversations today. |
| Gmail one-click email | The "Connect with Google" option when adding an email channel, instead of entering IMAP and SMTP settings by hand. | Also needs Google OAuth configured on our side. |
Widget behaviour
| Capability | What it does | Notes |
|---|---|---|
| Proactive engagement | Lets the website widget open itself with a page-aware message when a visitor lingers on a high-intent page without asking for help. | Off by default on purpose: it interrupts the visitor, so it is something you choose rather than something that appears one day. |
Social
| Capability | What it does | Notes |
|---|---|---|
| Social comments | Pulls comments on your Facebook Page and Instagram posts into a Comments inbox, with public reply, "Continue in DM", hide, and ignore. | Needs Meta comment permissions approved. |
| Social publishing | Compose and publish posts to your connected Facebook Page and Instagram professional account. | Restricted, not merely beta — see below. |
Acting outside the conversation
These three are the reason this page distinguishes "on request" from "on a bigger plan". Each one lets HelpStack act outside the conversation it was asked about, using your identity. None of them is granted by upgrading, at any plan level, because paying more is not the same as deciding you want this:
| Capability | What it does | Why it is held back |
|---|---|---|
| Deferred agent tools | The AI emails a carrier or other third party from your own domain, on its own initiative, and picks the customer's conversation back up when the reply arrives. | It sends real mail to a real outside party, from your mailbox, without a person approving that message. |
| Email composer | Lets an agent start an email thread with someone who has never written in. | Mail to arbitrary addresses from your domain carries a deliverability and reputation cost that is yours, not ours. |
| Social publishing | Posts publicly as your brand. | A public post is not retractable in the way an unsent draft is. |
Ask for any of them and we will switch it on. The point is that you decide, knowing what it does.
Related#
- Billing & tokens — what each plan includes.
- Agent tools — including the deferred and async types.
- Asking a counterparty — how the email-inquiry flow behaves in practice.
- Glossary — definitions of the terms used here.