CACTUS accelerates prototyping using generative AI on AWS
Academic research services company CACTUS Communications built more than 20 generative AI use cases, including the Ask R Discovery research assistant and Paperpal's ChatPDF feature, on Amazon Bedrock with Anthropic Claude, cutting idea-to-prototype cycles by 65 percent and growing daily active users by 40 percent.
Overview
Academic research services company CACTUS Communications built more than 20 generative AI use cases, including the Ask R Discovery research assistant and Paperpal's ChatPDF feature, on Amazon Bedrock with Anthropic Claude, cutting idea-to-prototype cycles by 65 percent and growing daily active users by 40 percent.
This entry has 13 published fields tied to exact passages in an immutable source capture.
Inspect the highlighted sourceThe challenge
CACTUS wanted to build generative AI-powered research and publishing services using the latest LLMs without the cost and burden of provisioning and managing its own training and inference infrastructure.
The solution
CACTUS built more than 20 generative AI use cases on Amazon Bedrock (running Anthropic Claude) integrated with its existing AWS stack (Amazon EKS, AWS Lambda, Amazon Athena, Amazon S3, Amazon OpenSearch Service), using Terraform and AWS CDK to automate provisioning and Amazon Bedrock Guardrails for safety, producing products like Ask R Discovery and Paperpal's ChatPDF.
Reported business value
CACTUS cut idea-to-prototype cycles for new AI features by 65%, doubled its annual growth in AI products for three straight years (over 100% annually), grew daily active users 40%, reduced token usage costs 5-10% via prompt caching, and maintains a net promoter score of 50-70 versus an industry average of 20-30.
Sources
Open any source and check the claim yourself — that is the point of the register.
This record was researched and written with AI assistance, and its claims were checked against the sources above. (EU AI Act art. 50 transparency notice.)
Other education entries in the register.
Fontys ICT runs a six-month, 300-user pilot of an EU-first multi-provider AI gateway platform
Fontys ICT, a Dutch university of applied sciences, designed and operated an institutional AI platform in a six-month, 300-user pilot. It combines an OpenWebUI frontend with single sign-on, a Portkey gateway that enforces access, budgets and EU-first routing, and a model catalog spanning Azure AI Foundry EU regions, GreenPT on French datacenters, Fontys-operated hardware in the Netherlands, and US providers behind explicit consent. The authors report the platform ran reliably with no privacy incidents and strong user adoption.
North Dakota University System builds a GenAI 'Policy Assistant' on Databricks for instant policy search
The North Dakota University System (NDUS) used the Databricks Data + AI Platform with Llama 2/DBRX, Foundation Model APIs, Unity Catalog and AI Search to build 'Policy Assistant,' a GenAI application that synthesizes over 3,000 public policy PDFs so staff can query regulations in plain English and get answers with page references and links in seconds instead of hours. The system cut NDUS's time to bring new data insights to market from one year to six months, 10-20x faster than manual policy search.
Open Universities Australia triples lead qualification with generative AI agents
Open Universities Australia (OUA) used LivePerson's AI Studio to build an LLM-powered generative AI agent that handles initial prospective-student inquiries, replicating the conversational style of human advisors. The agent achieved 3x lead qualification rates compared to self-searching students and 2x compared to OUA's prior scripted chatbot, with an average response time of 6.3 seconds versus roughly 2 minutes for human advisors.
Ativion Powers Generative AI Classroom-Management Solution StudentKeeper with Amazon Nova
EdTech company Ativion built StudentKeeper, a generative AI classroom-management solution powered by its ARIA assistant, using Amazon Nova Pro and Amazon Nova Lite foundation models on Amazon Bedrock to filter content, calculate near real-time student attentiveness scores, flag safety concerns, and automatically draft teacher progress notes from the 200-400 million daily data events generated by its ContentKeeper filtering service.
Was this helpful?
Your feedback helps us improve our use case database
