← Back to all sparks
A

Appcues

ANALYTICS
Velocity0.0

Appcues drops Embeds — in-product experiences that live inside the UI rather than overlay it.

product adoptionin-app experiencesai assistantmcpembedsworkflows
◆Current state
Appcues is a product-adoption platform whose recent quarter has run two parallel storylines. Captain AI, the in-product assistant, has gone from a chat helper to something that drafts segments, analyzes funnels, diagnoses display problems, and explains performance — adding capability essentially every monthly release. Alongside that, the team has expanded the experience surface itself: an MCP Server that exposes Appcues data to ChatGPT and Claude, and Embeds — a new experience type that lives inside the product UI rather than as an overlay.
◆Where it's heading
Appcues is reframing what an 'in-product experience' tool covers. Embeds break the long-standing overlay-only model that defines the category (Pendo, Userpilot, Chameleon all anchor on overlays). MCP exposes the same data surface to external AI tools, which makes Appcues a source as well as a destination. Captain AI keeps absorbing operator tasks — segmentation, funnel analysis, install diagnostics — turning the product manager's in-tool workflow into more of a conversation than a configuration session.
◆Prediction
Expect Captain AI to start fully building things autonomously rather than drafting (the team teased this in the January notes), and for Embeds to gain a bigger pattern library now that the underlying primitive is shipped. The MCP server integration line will likely grow with more bidirectional actions exposed to external AI tools.

◆Recent moves

  1. 5mo ago

    Embeds: in-product experiences that live inside the UI

    ⚡ SPARK

    Embeds break the overlay-only orthodoxy of in-product engagement tools — letting Appcues live inline where overlays would have felt interruptive opens a different conversation with PMs about what surfaces they can personalize.

    View source ↗
  2. 6mo ago

    February 2026 update: Captain AI drafts segments and diagnoses display issues

    Captain AI gains a noticeable batch of operator tasks: drafting segments end-to-end, analyzing 2-10 step funnels with drop-off user lists, comparing draft and published content, and diagnosing display issues by reading frequency and throttling rules.

    View source ↗
  3. 7mo ago

    January 2026 update: segmentation planner and @ mentions

    Captain AI's segmentation planner closes one of the more painful authoring loops — figuring out which property to use when integrations have created near-duplicate names. Mentions via @ handle the same friction in the navigation surface.

    View source ↗
  4. 8mo ago

    December 2025 update: MCP Server exposes Appcues to ChatGPT and Claude

    ⚡ SPARK

    The MCP Server move makes Appcues addressable from external AI tools — it's the connector that lets a PM ask ChatGPT or Claude about Appcues content without context-switching, and combined with Captain AI's bulk publish/unpublish actions, opens automation patterns from outside the product.

    View source ↗
  5. 9mo ago

    November 2025 update: Captain AI comparisons and Banner rich text editor

    Captain AI gets broader analytical questions — comparing flow performance, surfacing top performers, account-level engagement snapshots — alongside a Banner rich-text editor and drag-drop reorganization in Workflows. Steady, not directional.

    View source ↗
  6. 9mo ago

    October 2025 update: multi-path Workflows and Goals expansion

    Multi-path Workflow branching is real new capability for personalization, expanded Goals coverage rounds out measurement, and Captain AI now diagnoses display issues using the same checklist support uses.

    View source ↗