Ally’s generative AI experiment reduces marketing busywork

Dive Brief:

• Ally Financial said that a proprietary large language model (LLM) helped make its marketing operations more efficient in a monthlong test, per details shared with Marketing Dive.
• The Ally.ai chat solution, which launched in June, delivered an average of 34% time savings on marketing tasks compared to when they were done without artificial intelligence (AI). Ally estimates this level of productivity would save almost 3,000 working hours annually.  
• A sample of 80 prompts inputted into the LLM during the pilot carried an 87% usefulness rate and delivered accurate responses 81% of the time. The company will continue to experiment with the technology as generative AI continues to take hold heading into 2024.

Dive Insight:

Ally is the latest brand to offer evidence that the hype around generative AI is translating into concrete products that are impacting marketing functions. The company’s bets on the tech take the form of an LLM, a sophisticated chat interface in the vein of OpenAI’s ChatGPT that can toggle between sharing prompts to guide the creative process and answering user queries. 

Ally is focused on building out the feature in-house by leveraging its own data and cloud-computing infrastructure, an approach that executives said will help scale the tool across the enterprise while better monitoring for risk. Investing in proprietary tech could be of greater importance to categories like banking and finance that deal in sensitive personal information consumers wouldn’t want scraped by a third-party vendor.  

For now, Ally.ai appears most useful in the early stages of marketing initiatives. In one example of Ally.ai in action, a transcript of a podcast featuring an Ally executive was input into the LLM for the purpose of producing a short article that could be posted on the bank’s blog. The software was specifically asked to call out the most informative and entertaining bits of the discussion for consumers. A first draft materialized in 15 minutes but required a careful edit and human review. Ally said that the work overall took an hour when it would usually take four without help from AI. 

“We look at Ally.ai as an assistant that will be able to help our teammates with faster and smarter project delivery, which leads to a better experience for our customers,” said Andrea Brimmer, Ally’s marketing chief, in a statement. “In several different instances, our content writers were able to use thoughtfully crafted prompts to either start up the creative process or help extend the distribution of content to new channels. Anytime we can reduce the time to publish while also letting our creatives do what they do best, that means a lot in our fast-moving, competitive environment.”

Other use cases called out by Ally include generating first drafts of ad copy, video scripts, and social posts, along with improving search engine optimization, quality control and data analysis. The company said that it is currently evaluating potential applications of the solution in over 100 different areas, though acknowledged that not all of them may pan out. Ally.ai was previously piloted by Ally’s customer service department, helping summarize calls and ease workflows.

The bank views the LLM as a way to familiarize its marketers in a tech field that will likely be of growing importance in 2024. Ally has been hosting half-day sessions called AI Days to let employees explore the technology and share their findings. 

Eighty-seven percent of marketing and communications professionals are dipping their toes into the AI space, according to a recent benchmark survey from The Conference Board, with summarizing content and aiding with creative ideation some of the most popular applications.