Customer Agent building in Composer
Composer lets teams build, test, and operate Customer Agent in the same workspace already used for marketing campaigns.
Entries on this page are generated by AI from vendor sources. How AI is used
Tools to create, configure and coordinate custom agents.
| Month | Changes |
|---|---|
| 2026-01 | 4 |
| 2026-02 | 1 |
| 2026-03 | 3 |
| 2026-04 | 3 |
| 2026-05 | 3 |
| 2026-06 | 6 |
| 2026-07 | 9 |
| 2026-08 | 9 |
| 2026-09 | 13 |
| 2026-10 | 8 |
Select a month to view the whole-record timeline. This clears timeline filters.
Platforms with at least one feature state classified in this capability (primary or secondary, minor updates included), in the site's fixed order. Select one to filter the timeline. Counts are not shown side by side: they reflect how finely each vendor documents, not how much it offers.
Filters apply to the timeline. The chart and platform totals describe the whole record. Selecting a chart month clears timeline filters to show that same whole-record scope.
Matches the Release Matrix evidence: latest state reached in the selected period, entries under review excluded, primary and secondary capabilities included.
No recorded changes match these filters.
Within each period: major updates as cards, then notable ones as rows; newest first in each group. Minor updates below.
Composer lets teams build, test, and operate Customer Agent in the same workspace already used for marketing campaigns.
Will let marketers create AI agents that run as a smart campaign flow step and act on a lead record from plain-language instructions, such as normalizing data or flagging bots.
Admins can batch-test Customer Agent against sample questions and coach it, using no credits.
A reimagined Breeze Assistant will take a direction from a user and assign the right agents to carry it out.
Treasure AI Studio is available as a native Android app on Google Play, syncing chats, work folders, skills, and data access with the Web and Desktop apps via server-side execution.
Customer Agent knowledge, actions, tone and handoffs can be configured per visitor audience segment, in private beta.
Auto-Setup creates, tests and deploys a Customer Agent using step-by-step guidance built from a portal's own Help Desk and Inbox conversations.
Lets users save a recurring multi-step prompt as a reusable action in a work folder, rerun with one click or a slash command.
Lets a Custom Agent delegate to other agents within one session, running sub-agents in parallel and consolidating their output.
Marketers can build custom Merlin AI agents with their own guardrails and full visibility of each step; Custom Agents are live on all MoEngage data centres.
Saved AI Studio actions can now run automatically on a repeating schedule or cron expression, with run history and an enable switch.
A searchable catalog of pre-built CRM, Workflow and Shopify actions replaces single-form setup, with Action Gaps surfacing missing config.
Adds scheduling with notifications, selectable model tiers, workspace-level memory, and new tool integrations to Custom Agents.
Agent Builder lets users describe custom agents and agentic workflows in plain language to Breeze Assistant, built directly on existing CRM data.
Agent Hub brings every HubSpot AI agent into one place, showing live status and recent results by outcome, with a single action to activate idle agents.
Treasure AI Studio ships as a native Windows desktop app, bringing the existing web and Mac experience to Windows for the first time.
Account admins can set an org-wide custom instruction added to every agent's system prompt in AI Studio, applied automatically to new chats.
Custom skills built in an AI Studio project can now be shared account-wide as a private plugin, installed by teammates without a public GitHub repo.
A "Run Agent" workflow action triggers any Breeze AI agent and feeds its output back into automation for record updates or downstream actions.
AI Studio adds a Plugin Marketplace where users install skills from public GitHub-hosted marketplaces, with per-user and admin governance controls.
Now GA (Confirmed by Aug 2026). Current summary: Marketers can build custom Merlin AI agents with their own guardrails and full visibility of each step; Custom Agents are live on all MoEngage data centres.
Agent Console adds Operator-driven use-case setup, agent duplication, draft saving, fallback outputs for Canvas agents, required Catalog agent inputs, and re-running only empty cells.
In Agent Console, Operator can apply pre-configured starting templates for new agents, setting instructions, output fields, and context.
Admins write custom, testable trigger instructions in a rich text editor to control exactly when Customer Agent initiates lead qualification.
Admins write plain-language guidelines for when Customer Agent hands off to a human instead of maintaining keyword lists, using ready-made templates.
Campaign Creation and Journey Decisioning agent templates are available in the new Agentforce Builder low-code canvas using Agent Script; existing deployments must be redeployed manually.
New Agentforce and flow templates accelerate building and managing marketing campaigns.
Lets teams build custom AI agents inside Braze that generate content, make real-time decisions or update catalog data within set guardrails.
Agent configuration adds separate read/write toggles per CRM tool, with an optional human approval step before a write action executes.
AI Foundry agents now execute with the permissions of the end user running them rather than the agent's original creator, improving access governance.
Select customers get complimentary trial access to Adobe Experience Platform Agents via the AI Assistant interface, respecting existing access controls.