Scaling the Unknown in AstraZeneca - Scaling Process and Analytics Services with AI Agents

Executive Summary

Balaji presented on how AstraZeneca’s Process and Analytics Services within GBS is integrating AI to scale services, improve efficiency, and reduce cost per insight.

Context & Strategy

1. GBS supports multiple AstraZeneca functions (Finance, HR, R&D, Compliance, etc.).
2. Focus: embedding AI into enterprise services while balancing automation with human oversight.
3. AI adoption is guided by risk assessment (probability & severity of harm), ensuring the right level of human involvement (“in the loop,” “over the loop,” or fully automated).
4. Transformation pillars: People, Value, Capabilities—developing both AI “virtual workforces” and human decision-making skills.

Key Service Areas

1. Business Intelligence – shifting from static dashboards to conversational data access.
2. Content Creation & Management – scaling knowledge sharing across the enterprise.
3. Helpdesk & Support – enabling employees to “talk to their knowledge.”

Process & Analytics Services

1. Current offerings: Visual Analytics, Business Intelligence, Decision Intelligence, Process Insights, and Automation.
2. Goal: embed AI into these services (“drink our own champagne”) to improve speed, reduce costs, and standardize outcomes.

Practical Application – Requirements Gathering Tool

1. Challenge: inconsistent requirements gathering across teams.
2. Solution: AI tool that transcribes calls, summarizes context, drafts business/analytical problem statements, generates process maps, and ensures consistency in language.
3. Outcomes:

• Common language across teams.
• Automated process map generation with ~90% accuracy.
• Similarity search to avoid duplicate work.
• Automated milestone, roadmap, and user story generation.
• Faster project delivery and reduced iterations with stakeholders.

Results & Impact

1. Early pilots: ~170 unique users, 600+ problem statements logged.
2. Efficiency gains: ~77% faster in milestone delivery; customer iterations reduced from 7–8 to 2–3.
3. Stronger measurability: Jira tracking, A/B testing, time-in-motion studies.
4. Financial impact: significant cost savings through reuse of assets and reduced rework.

Next Steps

1. Extend AI assistants beyond requirements gathering to data prep, modeling, dashboard development, testing, and deployment.
2. Continue academic partnerships to measure value via design of experiments.
3. Maintain feedback loops to ensure AI learns and adapts to human needs.

Conclusion

AI is being embedded carefully, with a process-first, measurable, human-in-the-loop approach, enabling AstraZeneca GBS to scale services efficiently, reduce costs, and empower people to focus on higher-value decision-making.

Video of Presentation