Platform · Dashboard

Where humans and AI meet.

Shared inbox, contacts, AI sandbox. The three operator surfaces your team lives in every day, built so a human can step in on any thread and step back out when the AI picks it up.

  • Unified cross-channel inbox
  • Searchable contacts with bulk ops
  • AI playground for prompt iteration
Try MessageMind free Jump to reference

Watch the walkthrough

Dashboard surfaces in a few minutes

What's inside

The headlines before you dive in.

  • Shared inbox: unified view of every conversation on every connected channel, filterable by channel, assigned teammate, state (unread / open / closed) and date range.

  • Shared inbox: live socket-driven updates so new messages, status changes and AI replies appear without a refresh.

  • Shared inbox: per-conversation detail pane with full message history, contact context and the ability for a teammate to take over from the AI and reply directly.

  • Shared inbox: inline call surface for channels that carry voice, with a modal to accept, place or review calls tied to the same contact record.

  • Shared inbox: campaign view for outbound sends, letting operators see which contacts a broadcast reached and how each thread evolved after delivery.

  • Contacts: searchable, filterable contact table with saved filters for channel, tags, source, lifecycle state and custom fields.

Documentation Platform fundamentals Dashboard surfaces

Platform integration

Dashboard surfaces

  • CategoryPlatform

Overview

The MessageMind dashboard is the operator cockpit where humans and AI meet. It is organised around three primary surfaces: the Shared Inbox, a unified cross-channel conversation stream that mirrors every connected channel in real time; Contacts, a people-centric database with filtering, segmentation, imports, exports, bulk labels and per-contact profiles; and the AI Sandbox plus AI Playground, a two-pane environment for drafting, testing and iterating on the prompts and behaviours that drive your automations. Together they give teams a single place to triage conversations, maintain the customer record, and refine the AI that is answering on their behalf.

What MessageMind can do with it

  • Shared inbox: unified view of every conversation on every connected channel, filterable by channel, assigned teammate, state (unread / open / closed) and date range.
  • Shared inbox: live socket-driven updates so new messages, status changes and AI replies appear without a refresh.
  • Shared inbox: per-conversation detail pane with full message history, contact context and the ability for a teammate to take over from the AI and reply directly.
  • Shared inbox: inline call surface for channels that carry voice, with a modal to accept, place or review calls tied to the same contact record.
  • Shared inbox: campaign view for outbound sends, letting operators see which contacts a broadcast reached and how each thread evolved after delivery.
  • Contacts: searchable, filterable contact table with saved filters for channel, tags, source, lifecycle state and custom fields.
  • Contacts: per-contact profile combining identity fields, channel handles, tags, timeline of conversations and the latest message bubbles.
  • Contacts: bulk operations, including bulk label assignment, bulk template sends and bulk export.
  • Contacts: CSV import with a mapping table, review-import step and identity-conflict confirmation so duplicates are merged rather than silently created.
  • Contacts: export popout to pull a filtered segment to CSV for reporting or ad-platform audience building.
  • Contacts: template send modal to message a filtered segment through an approved channel template.
  • AI sandbox: dedicated page to draft and store prompt revisions for your agents without touching live traffic.
  • AI sandbox: AI Playground pane for running prompts against sample inputs, inspecting the model response and comparing variations side by side.
  • AI sandbox: ability to replay realistic conversation turns so a prompt is tested against the kind of messages it will actually meet in production.
  • AI sandbox: path from a validated sandbox prompt into the automations that drive the Shared Inbox, so iteration happens in a safe surface before anything ships.

Available data and actions

Reads

  • Live message stream from every connected channel (WhatsApp, Instagram, Messenger, SMS, web chat, voice and any other configured integration).
  • Contact records, including identity fields, channel handles, tags, lifecycle state and custom fields.
  • Conversation history and assignment / status metadata for each thread.
  • Prompt revisions, sample inputs and model responses stored for the AI sandbox and playground.
  • Campaign and template metadata used by the inbox campaign view and the bulk template send flow.

Writes

  • Outbound replies and template sends posted back to the originating channel.
  • Conversation state changes (assign, unassign, open, close, mark unread, hand off from AI to human).
  • Contact edits: field updates, tag and label changes, merges on identity conflict, and newly imported rows.
  • Bulk operations: bulk label writes, bulk template sends and CSV import commits.
  • Prompt and configuration changes made inside the AI sandbox before they are promoted to live automations.

AI agent use cases

  • A support lead opens the Shared Inbox each morning, filters to unassigned open conversations across WhatsApp and Instagram, and triages escalations the AI flagged overnight.
  • A teammate takes over an AI-handled thread mid-conversation, replies directly from the inbox for a sensitive question, then hands the thread back to the AI once the issue is resolved.
  • A marketer filters Contacts by tag and lifecycle state, exports the resulting segment to CSV, and uses it to build a lookalike audience in an ad platform.
  • An ops teammate imports a CSV of new leads from an event, walks through the mapping table and review-import step, and resolves identity conflicts so existing contacts are merged rather than duplicated.
  • A product owner iterates on a new agent prompt in the AI Sandbox, replays it in the Playground against representative customer messages, and only promotes it to the live automation once the responses look right.
  • A manager opens a campaign in the inbox campaign view after a broadcast, scans how recipients replied, and reassigns the hottest threads to specific teammates for follow-up.

Configuration

  • Shared inbox: filter presets by channel, assignee, status and date range, plus the default view a teammate lands on.
  • Shared inbox: per-conversation assignment, state (open / closed / unread) and AI handoff toggle.
  • Contacts: saved filters, visible columns and the field set exposed on the contact profile.
  • Contacts: tag and label taxonomy used by bulk operations and segmentation.
  • Contacts: import mapping (CSV column to contact field) and identity-conflict resolution rules.
  • AI sandbox: active prompt revision, sample input sets, and which playground result becomes the candidate for promotion to live automations.

Example workflows

Review an escalation in the shared inbox

  1. Open the Shared Inbox and filter to open conversations where the AI has requested human help or the contact has replied with a negative sentiment signal.
  2. Select the thread to open the detail pane and read the full message history, including the AI turns that led to the escalation.
  3. Check the contact profile in the side pane for prior tickets, tags and lifecycle state so the reply has context.
  4. Toggle the thread from AI to human, send a tailored reply (or place a call from the inline call modal if the channel supports voice), and update any contact fields that changed during the exchange.
  5. Resolve the thread: set its state to closed, reassign it back to the AI if follow-up is automatable, and add a tag that will let you find similar escalations later.

Iterate on a prompt in the AI sandbox

  1. Open the AI Sandbox page and duplicate the current live prompt into a new revision so production traffic is untouched.
  2. Edit the prompt (instructions, tone, guardrails, tool access) in the sandbox editor.
  3. Switch to the AI Playground pane, load a set of representative sample inputs (ideally real anonymised turns pulled from the Shared Inbox), and run them against the new revision.
  4. Compare the responses against the previous revision, note regressions and refine the prompt until the sample set reads cleanly.
  5. Promote the validated revision into the live automation that powers the Shared Inbox so the next real conversation uses the updated behaviour.

Limitations

  • The dashboard assumes at least one channel integration is connected. Surfaces render but stay empty until messages and contacts start flowing in.
  • Bulk operations and exports respect the current filter: a mis-set filter will act on a wider or narrower segment than intended, so review the match count before confirming.
  • CSV imports rely on the mapping table and identity-conflict step. Skipping conflict review can create duplicate contact records that are expensive to reconcile later.
  • The AI Sandbox tests prompts against sample inputs you provide; it cannot guarantee behaviour against every possible live message, so monitor the Shared Inbox after promoting a revision.
  • Voice features in the inbox require a channel integration that carries voice; text-only channels will not expose the call modal.