Feature Request

Use case: The AI Agent to detect when an outbound message has failed so it can take the right next step automatically. Why this matters: When a message fails, the conversation may need human intervention. If the AI Agent can see the failed status, it can: assign the conversation to a human add an internal comment with context trigger a workflow for follow-up retry after a few minutes Expected outcome: This would help reduce missed replies, speed up resolution, and ensure failed outbound messages are handled consistently.

Desired outcome Have a direct integration with Google Calendar for appointment management: Scheduling, inquiries, modifications, and cancellations. It would be ideal for AI Agent to be able to collect the info and perform the action directly with the calendar.

Business Problem: The current role-based permissions structure (Agent, Manager, Admin) does not provide enough flexibility for organizations with complex operational workflows. Many businesses — especially those with multiple internal departments — require fine-grained control over what each user can view, edit, or do inside the platform. This leads to risks such as accidental customer replies, unauthorized data edits, or exposing sensitive information to users who should not have access. Desired Outcome Introduce a granular permission system that allows workspace admins to customize access levels per user or per role. This should include the ability to enable or disable specific actions, modules, or permissions, such as: Access to Dashboard, Contacts, Messages, Snippets, Users & Teams (but not Workflows or advanced settings) Read-only access to conversations Restricting ability to reply to contacts Restricting ability to send files, surveys, or voice notes Allowing comments only for inter-team collaboration Use Cases Multi-Team Operational Workflows Some customers (e.g., online stores) have an operations team that reviews cases internally. The desired flow: Customer care assigns a contact to the operations team Operations team reviews history Leaves internal comments Contact is returned to the agent However: The operations team should not reply directly to the customer Replies must be text-only for some agents Operations team should have limited permissions (no surveys, no attachments, no voice notes) Restricting Access to Sensitive Contact Fields Currently, contact fields marked as hidden are only collapsed behind a dropdown — but are still accessible and editable. Organizations need stronger control over field visibility to protect internal or confidential data. View-only fields: Users (at least agents) should be able to see certain fields but not edit them. Fully hidden fields: Some fields should be completely invisible to certain roles (not displayed anywhere in the interface). This is important for companies storing sensitive customer data (financial info, internal IDs, CRM-synced fields, etc.), where only a subset of users should have access or edit rights.

Problem Statement: Currently, our AI features can’t send images or files to customers. This is a major roadblock for both sales and support teams, who often need to share product images, brochures, guides, or troubleshooting documents to give customers the full context or help them solve problems. Without this, customer interactions are less effective and require manual intervention for something as simple as sending a file. Desired Outcome: AI should be able to recognize when an image or file is needed and send the appropriate material automatically, based on the conversation. This would allow customers to get complete answers and resources instantly, boosting satisfaction, speeding up resolutions, and letting teams focus on more complex tasks—without having to jump in just to attach a file.

Business Problem Right now, AI Agent keeps up to 20 messages as context—even after a conversation has been closed. This can lead to responses feel outdated, irrelevant, or overly influenced by past conversations the user didn’t intent the AI to recall. For example, a Contact previously asked about refund. When they return later to ask about pricing, AI Agent references refund-related context, causing confusion and irrelevant responses. Desired Outcome Provide an option that allows users to disable historical context. When enabled, AI Agent should only reference the messages in the current conversation, ensuring responses stay fresh, relevant, and scoped to the active interaction.

Business Problem: Currently, each workspace can only create up to 10 Developer API Access Tokens, which limits how teams can operate: Forces multiple team members to share a single token, making it difficult to attribute actions to specific users or audit API usage effectively. Restricts teams that require individual tokens per member for access control, security, or role-based integrations. Blocks growing teams that rely on API access for automation or AI-powered workflows from scaling their usage within a single workspace. Desired Outcome: Increase the limit and allow workspaces to create more than 10 Developer API Access Tokens to support larger teams, enable proper per-user access management, and reduce reliance on shared tokens.

Business Problem Customers are currently unable to increase their AI Credits once they reach their limit, forcing them to wait until the next billing cycle. This creates a significant disruption in critical business work flows, such as automated customer support and lead qualification, which rely on uninterrupted AI availability. Desired Outcome To enable customers to top up their AI Credits both manually and or automatically (e.g. similar to MAC overcharges).

Currently, when a user replies to an Instagram story, the AI Agent usually receives only the message sent by the user, but not the content of the original story. This creates a significant limitation for those of us who use AI as the first level of care. For example, many users respond to stories with messages such as: * How much is it? * Is it today? * How do I sign up? * Is there still room? * What time? * Where is it? For a human person, the context is obvious because they can see the story to which they responded. However, the AI Agent does not have access to that information and must respond generically or derive the conversation unnecessarily. This especially affects businesses that use stories to promote: * events * restaurants * promotions * reservations * shows * commercial campaigns It would be very useful if Respond.io could expose to Agent AI all the information allowed by the Instagram API, for example: * text of the story * image or preview * associated link * sticker used * any metadata available Even if it were not possible to access the full content, it would be very helpful to have a specific variable or field that clearly identifies that the message comes from a response to a story and, if possible, to include a summary or reference of the original content. This would allow for much more precise answers, reduce unnecessary referrals to human agents and significantly improve the experience for both companies and users.

Business Problem: Custom Inboxes and Lifecycle stage folders currently show only the total count of open conversations next to the folder name. There is no visibility into how many of those conversations are still awaiting a response. The Unreplied toggle can filter within a folder to surface those conversations, but it doesn't show a count at the folder level — so teams have to click into each folder and toggle the filter to gauge their unreplied backlog. For high-volume teams, this makes it hard to spot which queues need attention without navigating folder by folder. Desired Outcome: Display both a total conversation count and a pending (unreplied) count next to each folder name in the inbox sidebar. For example: 100 / 90 — where 100 is the total number of conversations in the folder, and 90 is the number still waiting for a response. This should apply to Custom Inboxes and Lifecycle stage folders. A dual-count display at the folder level would let teams monitor workload distribution, spot response backlogs at a glance, and prioritize queue management without navigating into each folder individually.

Business Problem: Our broadcast analytics cover metrics such as Sent, Delivered, Read, and Failed statuses. However, we also need the capability to monitor the number of customers who respond to the broadcasts we send. Desired Outcome: To have the broadcast analytics for channels that support "Read" or "Replied" status.