AEM Content Fragments Bulk Operations and Co-author for Coworker – early adopters

One video, two skills, one full authoring loop.

• Bulk Operations for Content Fragments — search, update, tag, and create fragments from natural language. Field edits stage into an AEM Launch that's reused across the conversation, so reviewers get one clean diff to promote instead of a scattered mess.
• Co-Author Advisors — pre-flight your content before it ships: is it ready to publish? is it ready for semantic search? Two lenses on the same content, non-destructive reports authors can act on.

Demo: bulk-ops+co-author (1).mp4

Metadata support for Content Fragments

Content Fragment metadata is now fully supported in the Admin UI. Authors can view and edit fragment metadata directly where they work with their content, with full parity in the new editors as well, and manage their metadata schemas entirely in the new experience. And clients using Touch UI schemas today can keep using them — now in the Admin UI for Content Fragments too.

Demo: Content Fragments Metadata support in Admin View (1).mov

Content Fragments Folder Actions

Now Content Fragments Admin UI supports folder actions like Rename, Copy, Move and Delete. This creates alignment with Touch UI and Assets UI: same actions available on all UI surfaces.

Demo: FoldersActions (1).mov

Alt-text Autofix Pre-verify

Follow us through the extensive journey which allows the alt-text autofix to reach our customers.

Pre-verify flow + bugs→fixes — how we run fixes against a writable RV clone and verify they actually render (JCR + DOM) before a customer ever sees them, then mine the “failed to verify” buckets for bugs.

V2 migration status (all CS customers) — the report numbers: 56% autofix-generated × 63% verify-success = 35% visible; of 419 failed-to-verify, we proved only 10% are genuine content gaps and 90% were validation failures we could fix.

Live DEV apply/revert — deploy a fix from the Sites Optimizer UI, confirm it landed on the author clone, and revert it — the exact loop that surfaced several of the revert bugs. (DEV writable clones only; customer prod stays read-only.)

Demo: Demo

Audit Policy Backend

The goal is to ensure Mysticat audits the pages that matter most to each customer. Pages are automatically discovered through the sitemap, while customers can control which pages are included or excluded from auditing via an API.

Demo: audit-policy-manual-urls-backend.mp4

WebMCP in Experience Workspace

This PoC embeds WebMCP generation in Experience Workspace — making website functionality instantly invocable by AI agents. Through recommendations and the EW assistant, the experience owner enables WebMCP actions.

Demo: web_mcp_demo_with_voice_and_sound.mp4

AEM Co-Author for Coworker

This project brings 3 new plugins into your Coworker, letting authors know their content before something breaks:

• Publish Readiness Plugin - A pre-flight check for every publish. Tells you if your content is healthy enough for being published
• Reference Health Plugin - Know what you’ll break before you break it — invisible debt becomes an actionable reference tree
• Semantic Readiness Plugin - Spot the fragments invisible to AI search — and exactly why.

Demo: Summer GW 2026 - co-author for Coworker.mov

Content Lineage Skills

• This project contains a few AEM MSM-related skills, integrated via Coworker for a smooth in-conversation experience:
• aem-msm-launches: audit live-copy drift and run rollouts, sync, and live-copy creation, all in chat.
• aem-rollout-assistant: preview a rollout's exact impact (overwrites, protected overrides, additions) and approve before anything is written.
• attach-livecopy: adopt an existing page tree into MSM without recreating or losing its local edits.
• msm-doctor: diagnose a live copy's health and catch the broken inheritance and stale rollouts no one spots by hand.

Demo: GW_project_msm.mp4

Site2AEM

Site2AEM is an AI agent that reads most existing website and rebuilds its content inside a local AEM instance - in minutes, on a regular laptop. The freemium on-ramp that brings small businesses into AEM, with a natural upgrade path to Cloud. Human stays in the loop at every phase gate - the agent proposes, you approve.

Demo: Video

Web MCP

This PoC automatically generates WebMCP tool definitions for AEM sites — making any customer's interactive functionality instantly discoverable and invocable by AI agents, without the customer writing a single line of WebMCP code.

Demo: Demo Video

Tweak

Tweak - A live edit of published Adobe React Spectrum (S2) apps.

Pick elements on a live dev page, adjust styles with token-aware controls, preview instantly, then let Tweak apply the changes back to real source code mapping raw values to Spectrum tokens.
The first GW project where we partnered with the UI/UX team for tool development.

Demo: Tweak Demo

Groundwork - AI-Native Design & Scoping

Groundwork compresses the slowest, most senior-dependent phase of engineering — design, scoping, sequencing — from days or weeks to hours, while raising consistency and rigor (full requirement traceability + multi-persona review) and keeping humans in control at every gate. It democratizes senior-architect-level design judgment, and it produces exactly the artifact a downstream agent workflow needs to start building the right thing.

Demo: Video

FlowCast - AI Marketing Consultant [PoC]

A replacement for the traditional marketing agency. No more sending data out and waiting weeks for a strategy deck, FlowCast runs the whole workflow itself, end to end:

• Data Intelligence — Semrush, LLM Optimizer, Google Trends, and Sites Optimizer feed a unified view of your brand
• AI Strategy — Adobe CoWorker + Marketing Agent turn that intelligence into a prioritized strategic brief
• AI Execution — Firefly, Express, and Premiere Pro produce the actual campaign assets, ready to publish

One connected Adobe Experience — from insight to impact.

Demo: Video

TestyBot

Coverage tools tell you if code ran. TestyBot tells you if your tests would actually catch a bug , it reviews your repo and scores how well it's actually tested, not just how much of it executes:

• Quality Scoring — every test gets a score for meaningful assertions vs. boilerplate, duplicates, and happy-path-only coverage
• Maintenance Waste — flags dead tests, over-mocking, and duplicate scenarios inflating your test suite for no real benefit
• Missing Scenarios — analyzes production code and control flow to surface untested edge cases, then generates real tests to prove whether the bug exists

Demo: Video