Kipu Health automates compliance checks to improve patient care using Amazon Nova

Kipu has been providing support for patient care through its software solutions since 2012.

Behavioral health providers are expected to do more with less, delivering exceptional care while juggling complex documentation, shifting regulations, and tight operational margins. Kipu’s solution is designed to empower providers to overcome these challenges and deliver efficient and effective care. But although the business had already developed AI-powered features to streamline clinical documentation workflows, it quickly realized the need to integrate an analytics and intelligence layer to support its products.

Kipu expanded its AI offering, Kipu Intelligence, to transcribe and summarize clinical sessions into structured notes. However, compliance checks were still performed manually—a process that was time-consuming, error-prone, and struggled to keep pace with evolving compliance policies. Clinical notes could get rejected for insurance claims or fail to meet legal, regulatory, and best practice standards. This challenge was amplified by the fact that compliance requirements vary by provider, state, and discipline. Scoring the completeness or accuracy of a clinical note wasn’t a one-size-fits-all process—it needed the nuance and flexibility of something smarter. “We were struggling to find a way to score a range of different clinicians’ notes to determine whether they met the required criteria,” says Prabhat Bhattarai, chief architect at Kipu.

To accelerate its ambitions, Kipu called on the expertise of the AWS Generative AI Innovation Center—a team of AWS science and strategy experts with comprehensive expertise spanning the generative AI journey—to help the company develop a new capability for Kipu Intelligence.

Working alongside the AWS Generative AI Innovation Center, Kipu saw an opportunity to use Amazon Nova Foundation Models to deliver frontier intelligence for its product. Specifically, Kipu wanted to explore using the Amazon Nova Pro multimodal model to automate compliance scoring with precision and speed. Kipu benefited from the model’s industry-leading price performance while keeping patient data secure within AWS. “We chose Amazon Nova Pro for several reasons: the price performance, the availability on AWS, and the powerful outcomes that it drives,” says Russell Olsen, chief technology officer at Kipu.

The multimodal model analyzes clinical notes against rules and regulations that are defined through prompt engineering. It then calculates a compliance score in near real time that’s tailored to each provider’s requirements. Providers can add in context that can be applied to every note. “We can then inform the clinician on what they are missing or how they can improve documentation before they submit it for insurance and payment,” says Olsen.

Using Amazon Nova Pro, Kipu achieves rapid response times without compromising quality. The company boosts efficiency by reducing the staff time and effort needed for manual paperwork and by detecting issues in real time. With the help of Amazon Nova Pro, clinicians can instantly locate relevant data summaries on patient history, medications, outcomes, and treatments.

By using Amazon Bedrock to fast-track its generative AI applications and agents from prototype to production, Kipu can innovate while protecting customer data. “We’re dealing with really sensitive information, so keeping this within our secure AWS environment is really important,” says Olsen. Kipu protects personally identifiable information by using AI as a guide to strengthen documentation and keep clinicians on the right track, rather than rewriting notes on their behalf.

The business impact of using Amazon Nova Pro has been significant—clinicians save an estimated 40 percent of the time previously spent on documentation, and they generate compliance reports 80 percent faster. Using a generative AI–powered solution also means that Kipu can scale compliance checks across multiple facilities and states.

Another important benefit of using Amazon Nova Pro is that Kipu gained a 50 percent bigger context window compared with its previous solution—meaning that it can process more information at a time. Instead of needing to truncate large charts, it can now analyze the full picture of patient data for greater accuracy.

The business has also seen improved price performance using Amazon Nova Pro, resulting in 75 percent cost savings versus the previous model that it used. “The cost is so much more effective. We’ve been evaluating the model’s results and we’re very happy with the accuracy and output,” says Olsen.

Kipu’s solution not only helps optimize operational and cost efficiency but also assists providers in delivering better patient care. “For clinicians, the efficiencies gained means that they get to give quality time back to patients and focus on care delivery, not documentation,” says Bhattarai.

Looking ahead, Kipu is exploring how other models such as Amazon Nova Lite can help deliver even more cost-effective, scalable solutions to the behavioral health community. The business is bringing AI into other parts of the care journey, including treatment planning, mental status exams, and even agentic scheduling. “Our plan is to use AI throughout all the different workflows that our product supports and help optimize our customers’ experience and improve patient care,” says Olsen.

Using Amazon Nova Foundation Models will continue to play a pivotal role in how Kipu shapes the behavioral health landscape, helping the company provide capabilities such as hyperpersonalization. With all its core products running on AWS, Kipu is prepared to amplify its results. “Using Amazon Nova Foundation Models is letting us grow faster and support our customers better,” says Olsen. “It’s been a huge partnership for us, and we’re excited to scale what’s possible.”