Transforming the Industry with Innovations and Scalable Frameworks

IBM is the hybrid cloud and AI technology and services leading provider, focused on delivering client value through a combination of technology and business expertise. IBM solutions draw from an industry-leading portfolio of capabilities in software, consulting services and a deep incumbency in mission-critical systems, all bolstered by one of the world's leading research organizations.

Briefly describe the nominated organization: history and past performance (up to 1,250 characters):

The Cognitive Tooling Team, a dynamic group of 24 professionals within the IBM CIO organization, is dedicated to transforming the IBM enterprise through AI-driven solutions that improve sales, enhance experiences for external clients, partners, and IBMers, while delivering measurable business value. Team's mission centers on building innovative AI and automation capabilities that strengthen IBM's sales and customer support operations within the Salesforce enterprise ecosystem.

Over the past year, the team delivered 37 AI-powered offerings, including enterprise-wide self-service chat solutions for IBM.com, generative AI answering capabilities, advanced search tools, and recommendation engines. These solutions are designed with embedded analytics to track usage and effectiveness, ensuring clear visibility into business impact. For example, a successful AI chat session marked with positive user feedback is counted as a value-driven transaction.

Business Impact:

In 2025 alone, these offerings generated an estimated $247 million in business value, driven by work avoidance and productivity gains. By enabling customers to resolve technical issues autonomously, IBM significantly reduced reliance on human support, saving time and cost.

• 2024: 1.78M hours saved $213M value
• 2025: 2.00M hours saved $247M value

Beyond individual solutions, the team developed scalable AI frameworks that empower hundreds of IBM teams to customize and deploy AI capabilities seamlessly. One example is the AI-enabled Chat Framework, which allows teams to tailor content and generative Q&A features while leveraging Cognitive Tooling's technical expertise. In 2025, these frameworks supported 153 million transactions, amplifying IBM's enterprise-wide AI adoption.

Outline the organization's achievements since July 1, 2023 that you wish to bring to the judges' attention (up to 1,500 characters):

The Cognitive Tooling team's accomplishments over the last 2 years include innovative software solutions that utilize the latest and greatest technologies that deliver value to the business and technical teams. In creating these, team also have solved some complex problems that have been major issues in today's AI world. These exist as full standalone applications, frameworks built to enable scalability across multiple teams and applications, as well as supplemental applications to enterprise platforms.

Key Accomplishments include (See supporting materials):

AI Chat Framework for Sales and Marketing

• In 2025, this framework executed 516,350 chats on ibm.com with an estimated business value of $43 million in work/effort avoidance.

AI Chat Framework for IBM Customer Support

• Enabled 93 different product teams to offer customer self-help and automated help options directly in the chat experience. Over 80,000 chats are executed with IBM customers to date.

AI Hallucination Detection

• Created a reusable AI Hallucination Detection system, automatically checks the answers to confirm they are grounded in the documentation that is used by GenAI to provide the answer.
• To date, the Hallucination Detection system has been called 1,310,777 times. Of those, 112,903 (9%) were detected as hallucinations; therefore, they were NOT shown to users.

AskSales

• Enabled an AI powered Agentic chat solution that provides insights and actions from within IBM enterprise CRM system.
• Over 48,000 chats were executed to help sellers become more productive with tasks on the CRM platform, delivering insights on potential clients and prioritizing opportunities.

Retrieval Augmented Routing

While unifying many different agents, how do we effectively route the user query to the correct agent to provide the answer?

• Devised Retrieval-Augmented routing, achieving over 90% accuracy in routing.

LLM as a Judge (LLMaaJ)

• Built an LLMaaJ system, utilizing the power of LLMs to confirm if the GenAI outputs of the system were high quality, saving an estimated ~1000 hours of data scientist time.

Application Modernization:

• Devised a new way to modernize old offerings, achieving a 93% reduction in the amount of technical maintenance required from the delivery team prior to its modernization.

MAGMA

• Introduced MAGMA a "build once, use anywhere" framework to accelerate the creation of AI applications powered by LLMs.

• Over 3.47 million transactions executed in the systems, contributed to over 600K hours saved across IBM support agents.

Explain why the achievements you have highlighted are unique or significant. If possible compare the achievements to the performance of other players in your industry and/or to the organization's past performance (up to 1,500 characters):

Over the past two years, the Cognitive Tooling team has delivered transformative AI solutions that address critical challenges in enterprise-scale adoption of Generative AI. These achievements stand out for their innovation, measurable impact, and industry relevance.

The team developed AI Chat Frameworks for IBM Sales, Marketing, and Customer Support, enabling personalized, scalable experiences that processed over 596,000 chats and generated $43M in estimated business value last year. They tackled reliability validation issues devising a Hallucination Detection system, preventing over 112,000 incorrect responses from reaching users. To unify fragmented AI capabilities, the team introduced AskSales and pioneered Retrieval-Augmented Routing, boosting routing accuracy from 60% to over 90%, a breakthrough for multi-agent environments.

Further, team created a Supervisor Agent for orchestrating complex, multi-turn agenting conversations and an LLM-as-a-Judge system to automate LLM output quality evaluation, saving ~1,000 hours of data scientist time. Their application modernization strategy, powered by LLMs, reduced codebases by 84%, complexity by 92%, and infrastructure costs significantly, while strengthening security. Finally, the MAGMA framework enables rapid, production-ready deployment of LLM-powered applications, adopted by 16 IBM teams and saving 600K hours across operations.

These innovations are significant because they solve industry-wide problems, deliver quantifiable business outcomes, and establish future-ready architectures that position IBM at the forefront of AI-driven enterprise transformation.

Reference any attachments of supporting materials throughout this nomination and how they provide evidence of the claims you have made in this nomination (up to 1,500 characters):

PLEASE review the attached Outcomes Brief, which provides short summaries on the accomplishments of the Cognitive Tooling Team. Thank You!