Paul Tidwell

Integrating large language models and powerful algorithms into your workflows is a massive undertaking. All that hard work can be undermined if the outputs fall short, quickly eroding stakeholder interest and trust. That's why effectively evaluating LLMs and algorithms is crucial. From initial prototypes to production, and through continuous experimentation and benchmarking, mastering evaluation ensures these advanced tools truly deliver on their promise. #LLM #Algorithms #AI

8

Signal's co-founder just launched an AI that actually keeps your secrets. Moxie Marlinspike has launched Confer, a privacy-focused AI assistant that promises something most competitors cannot: conversations that remain truly private. Unlike ChatGPT or Claude, Confer encrypts all interactions using WebAuthn passkeys and processes everything within Trusted Execution Environments, ensuring the host company never accesses your data. Marlinspike's motivation stems from recognizing AI chat as an inherently confessional technology that knows more about users than any previous platform. The architecture is deliberately complex, combining encryption, remote attestation, and open-weight models to deliver on its privacy promise. While the free tier offers 20 daily messages, unlimited access costs $35 monthly - significantly more than ChatGPT Plus. The premium reflects the reality that robust privacy infrastructure comes with higher operational costs. For users concerned about data collection and advertising targeting, Confer represents an intriguing alternative that prioritizes user privacy over profit maximization. Check out the article here: https://lnkd.in/e-3DTjGS If you like this content, consider subscribing to my weekly newsletter where I share 3 key events in the data and AI space, 2 BigQuery tips, 1 thing that piqued my curiosity - completely FREE. Follow the link below to sign up.☟ www.beardeddata.com/signup #Privacy #AI #ChatGPT #DataSecurity

4

1 Comment

Big Step, Daniel! SAMA’s 2025 momentum is clear: Version 1.0 of the use cases, risk evaluation framework, and implementation guide is now adopted. Participation includes 170+ participants across 35 states. Next up is a model state Medicaid AI policy aligned to NIST and built for real program operations. If your team wants a walkthrough or to learn more about SAMA, message me or Daniel and we'll set it up. #Medicaid #AI #ResponsibleAI #NIST #SAMA #AIinHealthcare

11

1 Comment

Anthropic released #Claude for #Healthcare today at #JPMHealthcare. Looks like a thoughtful set of agents with connected, commonly used data assets. Here's a summary from all four corners of the healthcare game board. **Providers** Claude offers HIPAA-ready infrastructure that enables healthcare organizations to deploy AI for workflows involving protected health information. - **Administrative efficiency**: Speeding up prior authorization requests and helping with patient care coordination to reduce pressures on clinicians’ time - **Database access**: Connections to ICD-10 codes for medical coding and billing accuracy, the National Provider Identifier Registry for credentialing and network management, and PubMed’s 35+ million biomedical articles for literature reviews - **Interoperability tools**: FHIR development agent skills to help connect healthcare systems faster with fewer errors - **Claims support**: Accelerating insurance appeals with structured reasoning and evidence synthesis **Patients** Claude can summarize patient-uploaded health records, explain test results in plain language, detect patterns across fitness and health metrics, and prepare questions for appointments . Additional features include: - Integration with Apple Health and Android Health Connect (rolling out in beta for Pro and Max subscribers) - Private by design with explicit opt-in required, and health data excluded from model training - Help navigating complex medical reports and understanding patterns over time **Pharma & Life Sciences** New capabilities for life sciences extend from preclinical R&D into clinical operations and regulatory affairs , including: - **Clinical trials**: Drafting clinical trial protocols and tracking trial indicators - **Regulatory submissions**: Helping with regulatory submissions so that more life saving drugs can come to market faster - **Scientific research**: Deep integrations with scientific databases and lab tools, with biosafety guardrails built in - Improved performance on spatial biology analysis and other life sciences benchmarks with the Claude Opus 4.5 model **Payers** Connectors to the Centers for Medicare & Medicaid Services Coverage Database enable Claude to verify locally-accurate coverage requirements and support prior authorization checks . Additional capabilities include: - Building stronger claims appeals - Support for revenue cycle and compliance teams - Prior authorization review agent skills that can be customized to organizational policies All standard #AI disclaimers apply. Always protect your #Data. https://lnkd.in/gZxQHd7h

12

This is a great article on the importance of focusing on outcomes of implementing AI not the technology itself in healthcare. The pattern is true across industries. https://lnkd.in/gEHRV5AK

5

Looking to turn AI investments into real business impact? The AI Center of Excellence (AI CoE) approach empowers teams to start experimenting with AI in practical, everyday ways — even without a technical background. Nice blog by Aubrey Northrup who is driving the AI-augmented marketing capabilities at Concord in discussion with AI CoE leaders Plamen Petrov, Hristo Aladjov, and Mudit Jindal 

17

2 Comments

To fix the EHR, we should focus on simplicity and stability. In the below insightful article, David Lareau argues that EHR consolidation and "innovation by marketing press release" are putting the industry at risk. While I agree that the shrinking vendor landscape is a challenge, I see a massive opportunity to break the stagnation by simplifying how we think about EHRs. Let's start by boiling the system down to its essence: Input/Storage: We are already seeing ambient AI streamline this, turning data entry into a background process rather than a burden. Retrieval/Output: This is the next frontier—generative UI/UX that adapts to the clinician's immediate needs rather than forcing them through static screens. However, innovation requires patience. David notes that constant ownership turnover kills progress. That is why I value the "buy and hold forever" philosophy at Harris/Constellation. It gives us the runway to build these long-term, trust-based AI solutions without the disruption of resetting roadmaps every few years. Read David's full article here. It's a great and thought-provoking read: https://lnkd.in/g3nTSMXj #ArtificialIntelligence #HealthcareIT #Workflow #AmbientAI #HarrisComputer

36