Adobe has introduced Firefly AI Assistant, a conversational tool designed to direct tasks across its creative software applications. The announcement includes new video and image editing features, as well as an expanded selection of third-party AI models available within the Firefly service.
AI Assistant Capabilities
The Firefly AI Assistant is built on what Adobe describes as a creative agent. According to Adobe, the assistant can perform multi-step work across applications including Photoshop, Premiere, Lightroom, Express, and Illustrator. Users describe desired results in plain language, and the assistant handles task sequences within a single interface.
The assistant is designed for both less experienced users and professional creators. It can track context, decisions, and progress across sessions and carry that context into individual Adobe applications.
Key functions include:
- Pre-built creative skills for repeat workflows
- Option to create customized skills
- Ability for the system to learn user preferences over time
- Capability to work with images, video, design files, and brand assets
- Organization of review and sharing tasks in Frame.io before applying changes based on feedback
Firefly AI Assistant follows Adobe's existing AI assistants in Photoshop, Express, and Acrobat. Adobe has stated it plans to extend this approach to third-party AI models, including Anthropic's Claude.
New Editing Features
Video Editor Additions
Adobe added the following functions to Firefly Video Editor:
- Speech clean-up through Enhance Speech
- Noise and reverb reduction
- Controls to balance speech, music, and ambient sound
- Color controls for exposure, contrast, saturation, and temperature
- One-click visual looks
- Adobe Stock integration, providing access to over 800 million licensed assets
Image Editing Tools
Adobe added two new tools:
- Precision Flow: Generates image variations from a single prompt with slider navigation
- AI Markup: Allows users to specify where edits should apply using a brush, rectangle tool, or reference images
Model Expansion
Firefly now includes more than 30 AI models from Adobe and third-party providers.
The latest additions are Kling 3.0 and Kling 3.0 Omni for video generation. Other third-party models include those from Google, Runway, Luma AI, Black Forest Labs, ElevenLabs, and Topaz Labs, alongside Adobe's own Firefly models.
Adobe stated that the broader selection is intended to give users more content creation options within a single product.
Availability
- Firefly AI Assistant is scheduled to enter public beta in the coming weeks
- New video and image editing functions are available now to customers with a Firefly plan
- Access to added partner models is available now to customers with a Firefly plan
Additional Updates
Adobe also introduced a feature called Quick Cut into its Firefly video editor. This tool processes B-roll, long-form clips, and newly generated footage to produce a structured first cut within Firefly's multi-track video editor. Users can upload footage, describe the video content, and provide optional details such as a shot list or script.
Adobe has also expanded its Creative Agent to third-party chatbots including ChatGPT, Claude, and Copilot, with plans to add support for Google Gemini and Slack. New capabilities include brand kit generation, where the assistant helps design logos and brand identity, and organizational features such as automatically labeling and grouping folders.