Adobe's Journey to Generative AI Leadership Using AWS
Adobe leveraged AWS services to train its Firefly family of generative AI models, integrating them into flagship creative software products like Photoshop, Illustrator, and Adobe Express. This enabled them to achieve generative AI leadership, launching new AI-powered features and products within nine months and scaling model training by 20x.
The Business
Adobe leveraged AWS services to train its Firefly family of generative AI models, integrating them into flagship creative software products like Photoshop, Illustrator, and Adobe Express. This enabled them to achieve generative AI leadership, launching new AI-powered features and products within nine months and scaling model training by 20x.
Before AI
Adobe was a leader in creative and digital experiences, offering traditional software products, but recognized the need to embrace generative AI to help its customers navigate technological disruption.
What Changed
Adobe trained its own Firefly family of generative AI models, integrating them into flagship creative software products like Photoshop and Illustrator. This involved building an AI training platform and data pipelines on AWS, and using Reinforcement Learning with Human Feedback (RLHF) to continuously refine the models based on user interactions.
The Result
Adobe launched its Firefly family of generative AI models in nine months and integrated them into core products, achieving a 20x scale-up in model training, empowering creativity on a global scale.
Why this matters
Training Custom AI Models for Core Business Integration
For companies with unique data assets or specific domain expertise, training proprietary AI models can be a powerful differentiator. Rather than relying on generic off-the-shelf solutions, developing a custom model allows for fine-tuning to specific use cases, ensuring higher quality, greater control over capabilities, and the ability to integrate AI deeply into existing products and workflows. This approach can lead to a significant competitive advantage and unlock new revenue streams by transforming core offerings.
Furthermore, the case demonstrates the value of iterative development and continuous feedback loops in AI. By incorporating Reinforcement Learning with Human Feedback (RLHF) and user feedback mechanisms, businesses can ensure their AI models evolve to meet customer preferences and address potential biases in real-time. This user-centric approach not only refines the AI's output but also builds trust and transparency, setting a new standard for ethical and responsible AI development in any industry.
Technology Used
Used for:
Source
Results are reported by the identified source/company and have not necessarily been independently verified by AI YAI YAI.
Could this work in your market?
Explore this opportunity in South Africa.
Related Cases
Software & TechnologyEngine accelerates AI agent development on the Salesforce Platform
Engine, a $2.1 billion travel platform, leveraged the Salesforce Platform to dramatically accelerate the development and deployment of AI agents. They reduced the time to build their first AI agent, Eva, from months to just 12 days, and significantly improved testing and integration processes for subsequent agents.
SoftwareSix Products, One Support Team, Zero Wasted Searches
Digital.ai, a company grown through acquisitions, faced challenges with its support team navigating six disparate product knowledge bases. They adopted Maven AGI to unify search capabilities internally, enabling engineers to find answers efficiently without context-switching. This successful internal deployment led to a customer-facing AI agent, improving service delivery by allowing customers direct access to a comprehensive, AI-powered knowledge base.
SoftwareDigital.ai Unifies Fragmented Support with AI, Empowering Engineers and Customers
Digital.ai, a company grown through acquisitions, faced challenges with support engineers searching across six disparate knowledge bases. By integrating Maven AGI, they unified their support knowledge, first internally then for customers, streamlining answer retrieval and improving service.
Media & EntertainmentLuka Dončić's AI version powered by ElevenLabs voice technology
Luka Dončić's production company, 77x World, partnered with ElevenLabs to create an AI-powered voice clone, Luk.AI. This AI voice accurately reproduces Luka's speaking style and can speak multiple languages, enabling the creation of new media and fan engagement experiences globally, exemplified by a campaign with Nike's Jordan brand.