Fueling creativity and driving impact at scale with gen AI

Creative Assistant was created to be more than a front-end interface. It was designed as a purpose-built, AI-powered tool to supercharge content creation with precision, speed and brand integrity.

The process began with a user prompt. With that prompt, a sophisticated network of microservices, APIs and intelligent agents sprang into action. The system uses retrieval augmented generation (RAG) with Elasticsearch to enrich the prompt to draw context from both user-provided data and IBM’s trusted knowledge sources.

A reranking service was then added to prioritize the most relevant information for generation to help every response be grounded in relevance and accuracy. The enriched input was passed to a large language model (LLM) trained and hosted on IBM watsonx.ai™, which offered flexibility in model selection and training based on the task at hand. While IBM® Granite® foundation models are commonly used for its enterprise-grade performance, other models are also employed, when appropriate, to help drive optimal results. The chosen model transforms raw data into high-quality, tailored content delivered in preset templates for emails, presentations, blog posts, client stories and product pages. The system is optimized on a regular basis to improve the quality of the responses.

Behind the scenes, AI agents break down tasks, coordinating searches and generation. The final output is handed off to a custom-built React user interface, which formats the response into clean, readable templates ready for review.

The strength of Creative Assistant lies in its foundation—a rich ecosystem of IBM trusted data sources, including IBM.com, product documentation, brand guidelines, competitive research, IBM RedBooks® and private repositories such as Seismic. An automated process using IBM Cloud® Object Storage helps ensure these data sources are periodically indexed and stored in IBM Cloud Databases for Elasticsearch and Milvus on IBM watsonx.data® data store. Data indexing and storage helps drive precise, vector-based querying and fast, scalable data retrieval across both public and private datasets.

Building on this foundation, IBM’s Creative and Content team has developed the next generation of Creative Assistant since the initial rollout. The upgraded version includes an agent‑driven chat experience, real‑time monitoring with IBM Instana® Observability, quality and style‑guide review capabilities, and chat‑history functionality.

By simplifying parts of the creative process and streamlining workflows, Creative Assistant can help IBM’s marketing teams focus on what matters most—delivering sharp, on-brand creative assets while paving the way for what’s next.

While adoption is still evolving and other tools are also being leveraged to enhance the workflow, Creative Assistant represents a significant step forward in addressing challenges around speed, consistency and scale. With prebuilt, brand-compliant templates in place, Creative Assistant can significantly reduce the time marketers spend on formatting and structure, freeing them to expand their creative purview. In the first year of launch, the tool onboarded over 1010 active users who generated over 7000 drafts of assets across 10 different content formats.

Creative Assistant has been especially transformative for the client stories workstream. A small content team, focused on elevating client and partner voices and responding rapidly to market opportunities with compelling stories, was often slowed down by fluctuating bandwidth and competing priorities. With the tool now embedded into their workflow, team members can quickly generate high-quality, brand-aligned initial story drafts without dependency on project managers or copywriters. In the first year of launch in 2025, the workstream produced over 2,400 client story drafts and versions to help iterate during the initial creative process. Initial outlines are created in roughly four hours. And, review-ready client story drafts, refined by writers and editors for compliance and IBM Brand standards, can now be completed in just about five days, down from the previous 10-day timeline—an estimated 50% improvement. The team also streamlined the creation of client story one-pagers—vital resources for Sales, Partners, and Executives that quickly showcase impact and drive meaningful engagement. What once took over roughly an hour now takes just about 12 minutes, delivering about an 80% efficiency boost. This shift enables the team to respond to market opportunities with greater speed and consistency, without compromising quality or brand voice.

Beyond speed and scale, Creative Assistant also offers built-in editorial support for grammar, clarity and tone. An integrated product assistant democratizes access to product insights, empowering every team member to create accurate, informed content without having to depend on siloed expertise or conduct extensive research.

Looking ahead, IBM’s Creative and Content team plans to further streamline the content lifecycle by adding additional functions, such as translation capabilities and workflow tools designed to introduce more intelligent automation into the creative process. When coupled with an expansive set of AI agentic solutions and innovative automations currently deployed and under development, Creative Assistant will help advance IBM’s vision for a more strategic marketing division.

IBM is a global technology company that offers a wide array of products and services in the fields of cloud computing, AI, data analytics and consulting. They’re committed to technological innovation, with a significant focus on research and development.