Walking the walk: Zennify’s AI transformation

The Wake of Agentforce 

Salesforce’s recent launch of Agentforce is set to transform how businesses operate, empowering organizations to build and manage autonomous agents that can perform tasks across departments—from sales to marketing, service, and platform management. The possibilities are vast. Agentforce holds the potential to revolutionize how financial Institutions leverage AI, enabling them to work faster, smarter, and more efficiently.

But how can you unlock this potential? While having the right data is essential, many institutions struggle to fully capitalize on it. Imagine if we told you that we’ve already been leveraging our own AI platform to enhance internal efficiency and operational enablement.

Here’s how we’re blending human expertise and AI to transform our business from the inside out.

Zennify’s Leap Forward in AI

Since inception, we've always embraced new technology. When ChatGPT debuted in late 2022, we eagerly experimented with its potential. Although our clients weren't ready for generative AI, we decided to focus inward, demonstrating the value within Zennify first.

Our AI and Data Engineering team, led by VP of Technology Philip Paz and Principal Architect Sam Kerr, quickly created a custom environment on AWS. They fine-tuned open source models using proprietary data and hosted them on AWS GPU clusters alongside a custom web app for running prompts. We extracted common threads from Slack channels and Zoom meetings and refined our prompt engineering expertise for our AI to better address emerging questions and issues.

Initially, we faced challenges with response times and scalability. However, through rapid iteration, we migrated to Azure Cognitive Services and Azure OpenAI. This transition significantly improved speed, scalability, and reliability. Our AI grew smarter, faster, and more reliable. We even gave it a name: Arti.

“It has been exciting to be on the forefront of the expanding generative AI capabilities and use it to transform how our teams access and apply knowledge. The immediate impact on productivity was incredible. We are now able to use the knowledge and experience we have gained to demonstrate and enable our clients for AI.”

– Sam Kerr, Principal Architect.

Arti now leverages advanced features like Chain of Thought and vectorized search indexes to enhance accuracy and reduce errors. The integration of speech processing enabled interactive capabilities and additionally, Arti gained multi-modal abilities, to handle image and video prompts with ease. As demand for contextual inferencing of unstructured data like video and images increased, we integrated Databricks into our processes to manage heavy data payloads. Additionally, we are utilizing the Data Intelligence Platform to ask questions of our data without having to write SQL queries, further streamlining our approach.

Arti’s Evolution and Automating the Future

Arti wasn’t just a technical experiment—it quickly became an integral part of our day-to-day operations, driving a 10% productivity increase across critical components of our consulting value chain. More than just answering questions, Arti is transforming how we solve client challenges, enabling faster, more efficient solutions. Here’s how Arti is delivering value across our teams:

• HR Team: Janet Larsen, our VP of People Operations, immediately recognized Arti’s potential. Instead of responding to 20-30 HR inquiries weekly, Arti has become the go-to virtual assistant for policy and benefits questions, freeing up HR resources for more strategic initiatives.
• Project Teams: By mapping thousands of past solutions into a comprehensive data lake, our AI team grounded Arti in Zennify’s extensive consulting history. Project teams were early adopters, utilizing Arti to streamline workflows from discovery through testing, significantly reducing manual effort and increasing project efficiency.
• Consultants: Arti revolutionized how our consultants approach their work. From summarizing client meetings to generating key project documents like user stories, Arti became an essential tool. Our Technical Architects use it to produce first-pass reference architectures based on simple photos of whiteboard sketches, accelerating the design process.
• Developers: Zennify has a vision to automate development and platform configuration on Salesforce and Databricks. We’ve started to explore the next phase—accelerating development cycles with AI-driven automation. We implemented an internally hosted All Hands AI instance in our Arti app, allowing users to prompt an AI Engineer to generate and execute code in an environment with tools like SFDX CLI for developing and deploying Salesforce components and data sets. Developers and Technical Architects also use Arti to generate technical designs and solution diagrams, significantly speeding up delivery timelines.
• Business Analysts and QA Teams: Arti played a pivotal role in testing processes, assisting our business analysts and QA teams with more accurate, efficient validations. Internal AI office hours helped our teams share best practices, unlocking even more opportunities to boost productivity and optimize workflows.
• Cross-Tool Integration: Arti’s capabilities weren’t limited to just one area. We extended its virtual assistant functionality to the tools we use daily, like Slack and Salesforce. Now, instead of asking colleagues if we’ve ever integrated a Jack Henry Silverlake banking core with Salesforce, Arti instantly provides the answer with client references, saving us time and mental energy.