How can GenAI be optimized for people and processes?

This article is brought to you thanks to the collaboration of The European Sting with the World Economic Forum.

Author: May Habib, Co-Founder and Chief Executive Officer, Writer AI

• Rapid innovations in GenAI are enabling widespread adoption of the technology.
• But organizations are also having to make tradeoffs between cost and scalability.
• We highlight three kinds of organizations that are successfully optimizing GenAI.

Generative AI (GenAI) innovations are coming fast and furious. Progress is happening at such a rapid pace that the barriers to entry — and adoption — are falling away, and work that once required specialized teams or outside suppliers is now in reach. Enterprise employees are completing a variety of tasks — writing, coding, analyzing, making predictions — more quickly and with fewer in-house skills than ever before.

But using GenAI across the business — the elusive holy grail of this technology epoch — is a different story. Organizations envision building solutions to make workflows across business teams and functions more productive, but the journey is proving longer and more expensive than expected. The dirty little secret of generative AI is that executives are struggling to integrate the technology piece-parts and get consistently accurate output from their solutions. They’re also making hard tradeoffs between cost and scalability.

Copilot-type solutions targeted at individuals do an amazing job at discrete personal tasks: bootstrapping sales emails, getting a blog started, and doing research. But they’re one-off in nature, and the generated output is untethered from the business. They don’t scale in a setting that relies on consistent data, knowledge, messaging, and other important business markers, so their value is limited. On the other end, there are fantastic AI solutions that streamline enterprise processes (think auditing expense reports or automating supply chains). They massively reduce tedious work and errors, but they also tap out in value because they’re honed for a single process and lack broad applicability.

But some companies have figured out how to optimize for both people and process. Here are a few examples.

What kind of companies are successfully optimizing GenAI?

A global cosmetics company automates content creation from its product database to the “digital shelf” across hundreds of products and dozens of e-retail distribution channels. Imagine how often product descriptions (and products themselves) change, driven by new positioning, competitive jockeying, and events such as holidays or season changes. Now imagine delivering content for each change to dozens of countries in the local language, accounting for legal, regulatory, and cultural requirements. With GenAI, the organization gets new product materials in market weeks earlier and uplifts online conversions, but it also empowers people — from brand owners to regional managers — by producing content that’s nearly market-ready the first time around, eliminating the back-and-forth that hinders productivity.

A CRM (customer relationship management) platform leader has built a variety of GenAI applications to satisfy dozens of complex cross-company workflows. The organization executes product launches across each of its brands multiple times per year, as well as hosts or attends hundreds of global events annually. Each of these moments has a “bill of materials” that touches some combination of product, UX, sales, marketing, customer support, and the executive suite and takes weeks to months to produce.

With its GenAI initiative, the organization taps into knowledge sources, style guides for each of its businesses, and automates the creation of materials from product support articles to customer stories to executive commentary. By generating custom content for the moment, and facilitating the workflows around it, the company not only reduces time and organizational thrash, but elevates stakeholders’ contributions from tactical to strategic.

A data protection innovator recently launched a new product, and used GenAI to create an internal sales enablement app that taps into the company’s knowledge, messaging, and brand sources to answer questions about customer personas, use cases, success stories, product and implementation details, benefits, and key messages by audience. Now, hundreds of sales professionals around the globe can get accurate answers to their questions in real time. They’re empowered to have intelligent, helpful conversations with their prospects, customers, and partners about the platform, all while cutting out the friction it usually takes to get questions answered.

What can we learn from these examples?

These projects have three important things in common that helped each company deliver GenAI value quickly with minimal expense and effort. First, they shared a set of trained LLMs (large language models), retrieval capabilities, and company-specific guardrails across the business, while customizing the last-mile application to the function or workflow. This gives them stamp-and-repeat rollout speed, output accuracy, and brand consistency across their AI deliverables.

Second, each took a collaborative approach to solution deployment, involving stakeholders early, iterating quickly with prototypes, and promoting knowledge sharing and skill-building across teams. As a result, teams are self-sufficient and can run fast with new use cases and solution advancements. And finally, they relentlessly pursued “uglies” in the system — sussing out bugs, hallucinations, and data indexing problems — and fixing them in the shared technology. This early work makes each company’s AI output more accurate and higher-quality for all stakeholders.