SoftBank Corp. is transforming sales with AI agents in Dataiku, capturing every conversation as insight, boosting data quality, and delivering various insights while working to reclaim a quarter-million hours a year for selling.
90%
of sellers report higher quality and efficiency in data-driven selling
80%
of customer conversations now link directly to opportunities
~20
hours saved per seller per month, projected 250,000 hours saved annually at scale
The Challenge
SoftBank Corp.’s sales organization requires effective information-sharing among sales representatives and diverse internal teams for customer proposals. However, key customer intelligence (such as records of conversations and insights) was fragmented across minutes, memos, and various other formats and locations, and therefore could not be centrally consolidated. Information obtained through sales activities was too vast to be reliably aggregated and shared via CRM, and data granularity was often compromised during the tool input process.
As a result, the organization was unable to fully leverage all available information for pipeline reviews, sales forecasting and reporting, or as training data for AI. An internal survey further revealed that sellers spent only about 20% of their time with customers, spending much time on internal communication and coordination.
The Vision
The breakthrough came when sales leadership reimagined their organizational sales model with AI at its core, where every conversation becomes structured, accessible data and AI agents deliver the right insights, instantly.
"We are working to automatically capture every conversation, without requiring sales input as a trigger, and have an AI agent deliver the right outputs to the sales team at the right time."
Shintaroh Imano, SoftBank
The Execution
Using Dataiku, The Universal AI Platform™, SoftBank built a suite of AI agents that automated insight capture, analysis, and communication across the sales cycle. An insight-extraction agent structured meeting data using the team’s sales metrics, a standardization agent unified opportunity stages for consistent forecasting, and research/chat agents generated account-plan materials and let sellers instantly query deal status, meeting outcomes, and next steps. Within one month, a working prototype launched and adoption quickly scaled across regions.
The results were transformative:
- 90% of sellers report higher data-driven selling quality
- 80% of conversations link directly to CRM opportunities
- ~20 hours saved per seller per month, 250,000+ hours annually
Pipeline reviews are becoming more objective, the effort involved in information sharing is being reduced, forecasts are trending toward greater reliability, and sellers are getting more time for what matters: customers.
What's next? SoftBank is extending its AI-agent model across the enterprise, using Dataiku to build predictive agents that flag risk, surface insights, and recommend the next best action.
