Adobe announces Firefly AI Assistant to orchestrate multi-step workflows across Creative Cloud apps
Adobe announced Firefly AI Assistant, an AI agent designed to handle multi-step workflows across Creative Cloud applications including Photoshop, Premiere, Illustrator, and Lightroom from a single interface. The assistant will enter public beta in the coming weeks, with Claude integration via Anthropic partnership.
Adobe announces Firefly AI Assistant to orchestrate multi-step workflows across Creative Cloud apps
Adobe announced Firefly AI Assistant, an AI agent that will orchestrate multi-step tasks across Creative Cloud applications from a unified interface. The assistant will enter public beta in the coming weeks, with no specific launch date disclosed.
Core capabilities
Firefly AI Assistant operates across multiple Creative Cloud apps including Photoshop, Premiere, Express, Lightroom, and Illustrator. Users provide prompts, and the assistant executes actions across these platforms without requiring detailed knowledge of individual applications.
The system maintains context across sessions and learns user preferences over time, according to Adobe, including aesthetic choices and preferred workflows. The assistant presents suggestions and asks contextual questions based on user requests, allowing users to intervene at any point to adjust outputs.
Pre-built and custom workflows
Adobe will provide a library of pre-built "Creative Skills" for common tasks such as retouching portrait photos with consistent presets or generating content across social channels. Users can also create custom skills to streamline their specific workflows.
The assistant includes context-aware features. For example, when editing a product photo in a forest setting, the system may present a slider to adjust surrounding trees and foliage density without complex manual edits.
Frame.io and third-party integration
Adobe's workflow platform Frame.io integrates with Firefly AI Assistant, enabling users to package materials for presentations, share with collaborators, collect feedback, and apply requested changes automatically.
Adobe partnered with Anthropic to bring Firefly AI Assistant compatibility to Claude, allowing creators to access Adobe functionality within Claude's interface. Additional third-party integrations are in development.
Immediate updates to existing tools
Alongside the Firefly AI Assistant announcement, Adobe released immediate updates to existing products:
Firefly Video Editor additions:
- Enhance Speech feature from Premiere and Adobe Podcast, with noise reduction, reverb removal, and audio balancing
- Color adjustment controls for exposure, contrast, saturation, and temperature
- Adobe Stock integration providing access to 800 million licensed assets
New third-party video models: Adobe added Kling 3.0 and Kling 3.0 Omni to its roster of over 30 third-party models, joining Google's Nano Banana 2 and Veo 3.1, Runway Gen-4.5, Luma AI's Ray3.14, Black Forest Labs' FLUX.2 [pro], ElevenLabs' Multilingual v2, and Topaz Lab's Topaz Astra.
Firefly image editing updates:
- Precision Flow: Generates multiple variations from a single prompt with an adjustable slider
- AI Markup: Brush and rectangle tools for direct image control, object placement, and lighting refinement
What this means
Firefly AI Assistant represents Adobe's entry into cross-application AI orchestration, competing with general-purpose AI assistants by specializing in creative workflows. The Frame.io integration and Claude partnership indicate Adobe's strategy to embed its tools both within its ecosystem and external platforms. The success of this approach depends on execution quality and whether the context-aware features deliver consistent results across complex creative tasks. Pricing has not been disclosed, leaving questions about accessibility for individual creators versus enterprise customers.
Related Articles
Anthropic adds dynamic workflows to Claude Managed Agents, allowing up to 1,000 parallel sub-agents per execution
Anthropic has added dynamic workflows to Claude Managed Agents, letting a lead agent plan a task, distribute it to up to 1,000 parallel sub-agents per execution, and merge the results. Anthropic claims the approach found 66 of 70 hidden bugs in a 116,000-line codebase, versus 14 to 27 for a single agent. Pricing and token costs were not disclosed.
Anthropic adds Dashboards and Motion betas to Claude; Docs, Slides and Design exit beta for all plans
Anthropic launched two beta features for Claude: Dashboards, which builds auto-updating dashboards from connected data sources, and Motion, which generates animated explainer videos exportable as MP4. Docs, Slides and Design leave beta and now work on every plan, including free accounts.
Anthropic launches OSS Scanner: free, model-generated security scans for opt-in open-source projects
Anthropic has launched OSS Scanner, an opt-in service that gives open-source projects periodic vulnerability scans at no cost, run by its strongest models including Claude Mythos. Reports are fully model-generated with no human review or triage, so Anthropic warns some may be incorrect or invalid.
Google Cloud launches Gemini agent, a multi-model enterprise agent that runs Gemini and Claude, in private preview
Google Cloud announced the Gemini agent at Gemini at Work 2026 on October 8. It is an enterprise agent that routes each job to either Gemini or Claude models. It is in private preview, with wider availability for select Workspace Business and Enterprise plans coming soon.
Comments
Loading...