Pronto Channel

Pronto has no bot accounts, so your agent works through a dedicated Pronto user. People message that user directly, or @mention it in groups, and the agent replies as that user.

Before You Start

• Integrations must be enabled for your Pronto organization. If you don't see API Tokens in Pronto's settings as an admin, ask your Pronto Customer Success Manager to enable integrations.
• You need a Pronto organization admin to create the agent's user.

Connecting Pronto

Step 1: Create a Dedicated User for the Agent

1. In Pronto, create a new user for the agent (for example "Communa Agent")
2. Give it the admin role — Pronto only lets admins receive webhooks
3. Make sure the user has an email address — Pronto requires one before webhooks can be enabled

Step 2: Create an API Token

1. Sign in to app.pronto.io as the agent's user
2. Open Settings → API Tokens and create a token with these scopes:

org-users-read is the only scope that shows other people's emails. If you'd rather not grant it, use profile-read (and users-read for starting conversations) instead — the agent then treats every Pronto user as an external contact.

No other org-* scopes are needed — leave them off.

Step 3: Connect in Communa

1. Go to your agent → Channels tab
2. Click Connect Channel → Pronto
3. Paste the API token and click Connect Pronto

Communa checks the token's scopes, confirms the user is an admin, and creates a webhook endpoint on that user in Pronto. You don't need to configure anything in Pronto's Webhooks screen.

Step 4: Add the Agent to Chats

Pronto only sends the agent messages from chats its user is a member of, never your whole organization. Add the agent's user to every group where people should be able to reach it; threads inside those groups are included. Removing the user from a chat stops its messages from reaching the agent. Anyone in your organization can also message the agent directly.

---

How the Agent Responds

• Each group reply chain is its own conversation, so follow-ups in the same chain keep their context. Mention the agent again for each follow-up.
• @everyone does not trigger the agent.
• Group messages that don't @mention the agent are discarded on arrival. They aren't stored, logged, or read by the agent.
• Messages sent from the agent's own user are ignored, so the agent never answers itself. To test it, message the agent from a different Pronto account.
• Send /stop to cancel the agent's current task in that conversation.
• The agent shows a typing indicator while it works.

Context isolation — Under Settings → Pronto you can choose whether different Pronto users can see each other's conversations with the agent, just like the other channels.

Starting Conversations

By default the agent only messages people who contacted it first: it replies to them, and can follow up later in the same DM or reply chain (for example when a scheduled task finishes).

To let the agent message anyone in your organization — say, "tell Dana the report is ready" — turn on Start Conversations under Settings → Pronto. This needs chats-write plus org-users-read (or users-read); if the token was created without them, replace the API token.

• The agent finds the person by name. A full name that matches someone it already knows on Pronto (including linked workspace members) is used directly; otherwise it also searches your whole Pronto organization. If several people match — say you only gave a first name and there are two Eitans — it asks which one you mean instead of guessing.
• It opens a direct message from the agent's user, or reuses the existing one.
• When they reply, the agent sees what it sent them, plus any note it kept for follow-ups (for example which report it meant and where it's saved) — so "which one?" gets a real answer.
• The person appears under People right away, with 0 messages until they reply. You can link them to a workspace member before they answer.
• It can only start direct messages — not post in group chats nobody has mentioned it in.
• Messages appear under the agent's name, so only turn this on for agents you trust to reach out on their own.

Workspace Members

Your teammates on Pronto are usually also members of this Communa workspace. When someone's Pronto email matches a workspace member's email, the agent treats them as that member: it uses their role (owner, admin or member) to decide what they may ask for, and — with User History context — remembers them across the dashboard and Pronto as one person. Everyone else is an external contact.

• Needs org-users-read on the token, and Recognize Workspace Members under Settings → Pronto (on by default).
• Matching happens when you connect, when you replace the token, whenever you run Test Connection, and the first time a new person messages the agent or the agent messages them.
• A match needs the exact email (case doesn't matter) on exactly one active Pronto account. The agent's own Pronto user is never matched.
• If you linked or merged someone by hand under People, that choice is kept.
• Someone removed from the workspace is treated as an external contact again.
• Every automatic match is recorded in the agent's change history.

Attachments

Photos, videos, files and audio sent to the agent are stored in its files. Voice notes and audio files are automatically transcribed, so the agent reads what was said. The agent can also send files back into the conversation (up to 50MB each).

Limits

• Pronto messages are limited to 5,000 bytes. Longer replies are split into several messages at paragraph or sentence boundaries. Non-Latin scripts such as Hebrew use about two bytes per character.
• Pronto rate-limits API calls per organization. The agent backs off briefly when Pronto asks it to.

---

Managing the Connection

Don't edit the webhook endpoint in Pronto. Communa creates it on the agent's user (its description says it is managed by Communa). In particular, don't rotate its signing secret — Communa can no longer verify deliveries after a rotation, and you'll need to disconnect and reconnect the channel.

Changing the token's scopes? Create a new token as the agent's user and use Replace API token in the connection's settings. Don't disconnect: disconnecting deletes the channel's conversation history.

Troubleshooting