At AWS, Jeffrey Hammond helps software companies accelerate product delivery, create new revenue streams and reduce technical debt.
Investment in generative artificial intelligence (GenAI) is ramping up at an impressive rate, with spending projected to grow at an annual rate of 36% through 2030.
A Forrester study (commissioned on behalf of AWS) of 657 software companies with production GenAI capabilities found that 51% are projecting a potential increase in revenue from their efforts and 46% are using cloud-based generative AI platforms to accelerate their efforts.
With higher investment comes higher expectations, but success is not guaranteed. Improve your odds of driving profitable growth with a four-step approach to defining GenAI use cases that drive business value.
1. Start with your customer and work backward.
As I work with software company CPOs and help them prioritize potential GenAI capabilities, the first question I ask is, “Where are there opportunities to reduce customer toil in your products?” Toil-filled processes with repetitive, low-value tasks are great candidates for GenAI automation.
Xero, a New Zealand-based accounting software company and AWS partner, launched a generative AI-powered assistant called Just Ask Xero (JAX) that automates important—but time-consuming—tasks like payroll and issuing payment invoices. Wendy’s FreshAI improves customer experience in the drive-through lane by making it easier for customers to order from complicated menus and even in multiple languages. FreshAI also reduces employee toil by allowing crew members to focus on food prep and order completion instead of issues with orders.
Other questions to ask include:
• “Where can we generate content (e.g. text, video, audio) that was time-consuming or uneconomic to produce manually?
• “Where can we light up data that was hard to access or hard to understand?”
One CPO mentioned to me that some of his company’s customers have decades of design data stored away in file cabinets where it’s hard for current employees to access and use. Another CPO shared that by summarizing a complex set of test results, they’re able to save practitioners five to eight minutes every time they interpret a test result by directing attention to the most important readings.
• “Where can we understand a customer’s intent through natural language and act on it for them?”
Every time we start from the customer’s perspective, we can find areas that are good candidates for GenAI. We take these as starting points for Step 2.
2. Set goals for accuracy and performance.
With customer-focused use cases in hand, it’s important to define how accurate a response from a large language model (LLM) needs to be and how fast it needs to return a response to drive business value. Higher accuracy and faster performance often have a direct impact on the cost of executing a specific business case.
For example, one of our prototyping teams collaborated with a software company on a mapping use case. Currently, customers manually match thousands of items between two lists. The CPO set a 90% accuracy threshold as the point at which automation would outperform the manual process. Through testing different LLMs and mapping strategies, the team identified a low-cost, high-performing solution that met the business requirements.
If you can’t balance the triangle of accuracy, performance and cost to serve, then it’s time to pause use case development, revisiting it on a regular basis. The rapidly dropping cost of inference means that use cases that are uneconomic today may not be six to 12 months in the future.
Once use cases meet accuracy, performance and cost requirements, you’re ready for Step 3.
3. Establish pricing power and define pricing strategy.
Developing an effective pricing strategy is crucial to driving profitable growth from GenAI use case that pass the business value test. And according to the Forrester survey, software companies are struggling with it. One in five of the independent software vendors (ISVs) Forrester surveyed indicated pricing is their biggest business challenge, and 42% of respondents rank pricing it as a top three challenge.
When matching your GenAI use cases to appropriate pricing models, it’s important to assess your pricing power. The Boston Consulting Group has excellent guidance for software company CPOs. Some key questions to ask when assessing your pricing power:
• What’s the unique data we apply in solving the customer need? The ability to use proprietary data to differentiate and improve use case accuracy creates differentiated business value.
• How do we deliver our user experience? When generative use cases are delivered as separate user experiences that are distinct from a software company’s “Hero product,” it can reduce the pricing power of the experience and create user confusion.
• Do we create user leverage? Use cases that automate and replace humans (such as contact-center optimization) may benefit from a consumption-based pricing model where use cases that create user leverage (e.g., product design) are better candidates for per-user or outcome-based pricing.
4. Make sure you’ve got the ‘-ilities’ covered.
Data privacy and ethical considerations are consistently key concerns in the development of generative AI products. Make sure you understand the most common risks and vulnerabilities for generative AI applications and the best ways to mitigate them.
Using cloud-based generative AI platforms can also help software companies remain secure and keep their customers’ data private. For example, Amazon Bedrock keeps customer data secure by design and supports configurable guardrails, while Google Vertex AI implements Google Cloud security controls to help secure your models and training data.
Expectations Met, Not Opportunities Missed
The points above can serve as valuable resources for software companies when defining and prioritizing GenAI use cases. If you want to further accelerate results, consider working with a cloud partner that can provide cost-effective access to frontier models and cutting-edge AI tools, as well as expert support and go-to-market opportunities.
Every investment carries risk, but following these four steps can help you select the ideal GenAI use cases with the highest business value and price them for profit.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
