Discover the key takeaways from our recent interview with Dave Broering, President of Integrated Logistics at NFI Industries
This past Tuesday, we took a trip to King of Prussia, Pennsylvania, a short drive from NFI headquarters in Camden, New Jersey, to have an informative conversation about NFI’s recent experience building a data warehouse and, ultimately, implementing generative AI tools. This fireside chat led by zeb CEO Mal Vivek, was the highlight of the live event, hosted by Databricks. As innovators in the strategic implementation of AWS technology, zeb partnered with NFI to help them rethink their platform approach to better align with business objectives.
Vivek’s guest, Dave Broering, has over two decades of experience within the non-asset-based logistics space. As president of Integrated Logistics, he brings a blend of technological focus and operational execution to NFI, one of the largest privately held third-party logistics providers in the U.S. His non-asset division is currently one of the fastest-growing divisions within the company.
The Conversation
Attendees learned that NFI initially started down the Gen AI path with OpenAI but quickly transitioned to Databricks. “The reasons for this transition were quite simple,” Broering explained. “It was not only faster, but also cheaper as we looked at scaling. Projections showed that even with 550 users representing our entire organization, the actual cost of the AI would still be under $50.”
One year into its data journey with zeb and Databricks, NFI has integrated Nudge and SuperInsight to create reporting-friendly models and gain a clearer understanding of its data than ever before. The Power BI reports and drilldowns NFI is getting today would not be possible without its data infrastructure having been built the right way.
Data Infrastructure
Before NFI could do anything in Gen AI, it needed a way to identify, canonize, catalog and govern its data. Enter Databricks Unity Catalog. Next, the organization’s leaders identified that in addition to transactional TMS data, there was key data in sources like Hubspot and Slack. Partnering with zeb, NFI rolled all its data into a lakehouse + Unity Catalog strategy.
Business Insights
Working backwards, zeb started with the insights NFI was looking to gather. We were able to determine key reporting needs, confirm how they wanted to slice the data, and design visuals for the reports. By building the data model around the way NFI was going to look at the data, this approach was also more conducive to maximizing Gen AI when it came time to implement it.
Machine Learning
One unique approach at the onset of AI implementation was that NFI leadership asked users try to stump the AI to collect additional testing data. This feedback was collected through SuperInsights and inspired additional feature development such as threaded replies for additional reporting asks, scheduled report delivery, and a whole new implementation in predictive shipper trends with Nudge AI.
Trial, Error, and Success
That said, Gen AI still requires extensive testing and trial and error. We began SuperInsights with a very small group at NFI and used their feedback in Slack to make multiple tweaks to the implementation – not only for it to be scalable, but also for it to be capable of handling more complex requests.
That’s because compound AI systems are the real path to ROI on Gen AI implementations. We used a compound AI model to first process user intent and then combine that with intelligent contextual SQL query generation leveraging Databricks. We also leveraged Mosaic as part of our overall SuperInsight architecture to fine-tune capabilities using industry adapters developed on large industry tribal knowledge datasets containing terminology and term equivalents.
ROI Achieved
In conclusion, the live event underscored the transformative impact of combining advanced data strategies with cutting-edge AI tools. Dave Broering’s insights into NFI’s journey highlighted how leveraging Databricks and Generative AI, in conjunction with a well-structured data infrastructure, can lead to significant efficiencies to meet evolving business needs. By refining its data models and embracing a strategic partnership with zeb, NFI has set a benchmark for how organizations can achieve meaningful ROI through thoughtful implementation of AI technologies. Contact zeb today to discuss how we can provide valuable guidance and unleash the full potential of your data assets.
+1 732 737 9188
sales@zeb.co