Showpad Revolutionizes Sales Enablement Using Anthropic’s Claude Models in Amazon Bedrock

Founded in Belgium in 2011, Showpad operates in more than 50 countries and serves large enterprise customers. Its products help marketing and revenue teams come together to empower sellers with content and training that engage buyers.

To better serve customers, Showpad formed a team in 2021 to enhance its data engineering, analytics, and AI offerings. Since 2015, the company has been using AWS throughout the business and wanted to place all its structured and unstructured data—amounting to multiple petabytes—in a data lake built on Amazon Simple Storage Service (Amazon S3), an object storage service built to retrieve virtually any amount of data from anywhere. “Using AWS, we built a foundational data layer that’s simple to access and is compatible with other services,” says Jeroen Minnaert, chief architect at Showpad. “Analysts, data scientists, and engineers can now build data-driven applications without the complexity of connecting to multiple data sources. It’s all in one place.”

Showpad began experimenting with AI technology in 2016 but faced challenges with deploying AI at scale. In June 2023, Showpad gained early access to Amazon Bedrock. The company worked closely alongside AWS and provided feedback to improve the service. By September 2023, when Amazon Bedrock was generally available to the public, Showpad was already primed and prepared to release Showpad AI, a comprehensive suite of AI-powered features that helps marketing, sales, and enablement teams drive better buyer conversations and close more deals through powerful insights, recommendations, and actions.

Showpad chose to use Anthropic’s Claude models in Amazon Bedrock because AWS and Anthropic share a commitment to AI integrity and responsible AI. Serving customers in sensitive industries like healthcare and financial services, Showpad appreciated this alignment of values because it adheres to Showpad’s four responsible AI principles—human agency, transparency, inclusivity, and integrity. For example, Showpad’s user interface clearly highlights all AI-generated content so that users can make more informed decisions. The company has thoughtfully introduced AI functionality, carefully incorporating it into the existing user experience to build trust and give customers time to adapt.

One of Showpad’s first experimental AI projects aimed to generate asset summaries so that sales representatives could see key learnings from a document during conversations with buyers. Before using Amazon Bedrock, the company used a single endpoint that didn’t support concurrent requests, which meant that data summarization happened asynchronously. When Showpad transitioned this use case to Amazon Bedrock, it could generate summaries in near real time using APIs in a more cost-effective way instead of waiting for batch processing. “Using Amazon Bedrock, we can scale up linearly and pay for the amount of work we actually need,” says Minnaert. Showpad also now achieves high performance with a time to inference of 2–3 seconds; this took up to 30 seconds previously. Sales representatives benefit from a seamless experience because summaries are generated in near real time when users engage with content or training in the software.

Because Showpad built a strong data foundation on AWS, its developers can focus on features rather than managing infrastructure. “We’ve removed code because AWS handles maintenance, pipelining, and plumbing that we previously needed to account for,” says Minnaert. “Our code has gotten more intuitive. Using Amazon Bedrock, we can scale more and release more features, helping sellers improve productivity and build buyer trust.”

In 1 year, Showpad quickly scaled from early experimentation with AI to releasing 12 new AI features, including asset summaries, search, and message composing. Showpad adopts new models in Anthropic’s Claude model family as soon as they’re available in Amazon Bedrock. Showpad also migrated eight AI features to Anthropic’s Claude 3 Haiku foundation model, which is 30 percent faster and one-third the cost of Claude Instant, just 1 day after its launch in the Europe (Frankfurt) AWS Region. The company has advanced from text-to-text transformation features using retrieval-augmented generation to more conversational features. For example, Showpad uses Claude 3 Haiku for object-to-text transformation to automatically annotate images with detailed tags, like the specific model of a photographed product, which helps sellers find the search results they need faster.

Showpad can now better support its global user base. Previously, asset summaries were available only in English. Using Anthropic’s Claude models in Amazon Bedrock, users can translate asset summaries into 31 languages, helping customers in non-English-speaking regions or bilingual countries operate at scale and consistently serve global revenue teams. Showpad also developed PitchAI—sales coaching software that gives actionable feedback on recorded sales pitches based on factors like tone of voice and body language. In its analysis, PitchAI considers the expected speaking pace in each language.