Establishing the right mindset
"How can we act like an enterprise startup?" asks Pinak. "The focus is on becoming customer-led rather than business- and IT-led.
"The approach requires fast delegated decision making rather than top-down governance. At the macro level, how do we release new products when there's a value to customer rather than releasing only when something is perfect? How do we enable VC-style funding rather than doing project specific business cases? And how do we get a single multifunctional end-to-end team rather than just IT and business teams being co- located to each other?
"When it comes to the actual execution, we're looking beyond the fixed triangle of time, cost and scope—we believe in a fundamental mindset shift focusing on speed, quality and value."
Generative AI for the banking enterprise
Publicis Sapient's approach to Generative Al concentrates on nine categories where businesses can make progress:
- Creating new offerings
- Personalizing and creating comms
- Personalizing and creating content
- Focusing on conversations with the customer across channels
- Co-piloting for colleagues to improve the quality and productivity of tasks
- Automation of business processes and ways of working
- Product engineering and the software development life cycle
- Controls and protection for governance, fraud protection and guardrails
- Insights and decisioning to support strategy formulation
Pinak explains that Publicis Sapient takes an enterprise diagnostic approach to developing an Al suitability score to look at the drivers and barriers before identifying the critical use cases an organization should focus on.
"In parallel, you need to start creating a Gen Al platform by selecting your public LLM models and prompt engineering tools. And then, after getting hold of all your data, you build an enterprise conversational platform to interact with the LLM models.
"We then work on fine-tuning those models with plug-ins, and applying guardrails and content filters, before creating reusable libraries and prompt completion histories. Finally, we apply user analytics, operational analytics, model testing and experimentation, not forgetting there's always a human in the loop."
Continuous learning with AI
Pinak urges the need for centers of excellence to build Al capabilities across financial institutions: "We fundamentally believe in a continuous learning mindset. And we have a specific program called Espresso delivering exactly that.
The goal is to ignite curiosity for new topics and share what's happening across the industry. We focus on unlearning some of our bad habits to make sure our people are up to date and ready to support and enable our clients."
Modernization and simplification: A dual process
"It's easy to create something greenfield," says Pinak. "The issue is how to integrate that back into your legacy. How do you do your ledger reconciliations, risk or treasury integration or integrate back into identity and access management?
"How do you engineer genuine integration with existing systems? These are the nuances of modernization. And once you modernize, how do you go that last mile with the decommissioning of legacy systems?