actAVA AI

🚀 Introducing actAVA We launched the first AI orchestration platform built for healthcare, by healthcare experts. With Kora, we bridge the gap between AI potential & real-world implementation—secure, scalable, compliant. 👉 Learn more: actava.ai #HealthcareAI

Introducing CHI-Bench on @huggingface: the world’s first long-horizon healthcare benchmark for AI agents. 75 real healthcare workflows + 20 apps + 200+ MCP tools + 1,290 skills + process / outcome rewards huggingface.co/datasets/actav… Any questions, lmk!

The best AI agent (Claude Code + Claude Opus 4.6) passes only 28% of real healthcare workflow tasks. CHI-Bench by @actAVAai @iscreamnearby @HaolinChen11, built with Johns Hopkins, Yale, Stanford, CMU, Oxford and 20+ institutions, was designed to find out exactly how far we are. 🏥 Try it yourself 👉 modelscope.ai/datasets/actav… Three long-horizon domains tested: 🏥 Prior Authorization: provider intake and PA preparation for new referrals 📋 Utilization Management: full payer review cycle from intake to peer-to-peer 👥 Care Management: chronic disease follow-up, outreach, assessment, care planning 75 tasks + 3 marathon tasks + 23 end-to-end dual-agent scenarios. 20 medical apps via MCP, 1,279-document handbook. 💻 Git: github.com/actava-ai/chi-… 🔗 Leaderboard: actava.ai/benchmarks

CHI-Bench is the world's 1st long-horizon healthcare benchmark for AI agents. If you're building or buying AI for healthcare, this is the test that actually matters — real clinical workflows, not toy demos. U.S. healthcare needs this. 🏥🔬

healthcare benchmark, built on harbor!

🚨 Historic moment for @actAVAai ! 📷Just one day after launch, our benchmark dataset is already #10 most popular on Hugging Face — out of 1 million+ datasets! Huge thanks to @iscreamnearby , @HaolinChen11 , Deon Metelski, Leon Qi, Tao Xia, Joon Lee, Steve Brown, Kevin Riley, T. Y. Alvin Liu, M.D., Zhiwei Liu, Qingsong Wen, @CaimingXiong , Sanmi Koyejo, Eric Xing & all our collaborators. 📷📷

actAVA AI integrates CHI-Bench with @huggingface and @harborframework today. Users can run the CHI-Bench evaluation and RL training from both platforms.

🚀 CHI-Bench !!! (with @actAVAai et al.) CHI = Clinical Healthcare In-Situ 📄 Paper: huggingface.co/papers/2605.16… 🤗 HF: huggingface.co/papers/2605.16… ⭐ GitHub: github.com/actava-ai/chi-… 🏆 Leaderboard: actava.ai/benchmarks/lea… 🔗 Project page: actava.ai/benchmarks/chi…

In real healthcare operations, agents must do far more than answer medical questions. They need to read charts, interpret clinical and operational policies, verify coverage, route referrals, draft P2P scripts, and finalize care plans — where a single policy violation can mean a denied claim or missed patient outcome. @actAVAai @iscreamnearby led and developed CHI-Bench (Clinical Healthcare In-situ Benchmark), the first long-horizon, policy-rich benchmark for AI agents operating across end-to-end U.S. healthcare workflows. Key highlights: ▶️ High-fidelity simulators for Provider Prior Authorization, Payer Utilization Management, and Population Health Care Management, all exposed as MCP servers over patient, clinician, and insurer records. 🧪 Each trial runs 60–80 agent steps across 4–6 clinical stages, with access to 21 healthcare apps, 200+ MCP tools, and a 1,279-document operations handbook. Leaderboard results across 30 frontier agents: • Claude Code + Opus 4.6: 28% pass@1 • Codex + GPT-5.5: 21% • Utilization review: 41% • Care management: 32% • Prior authorization: 29% Reliability remains a major challenge: no agent exceeds 20% when the same case is repeated three times.

(1/n) After a few months of work with multiple hospitals, universities and research facilities, today we're open-sourcing CHI-Bench: the first long-horizon benchmark for healthcare AI agents on real clinical and healthcare workflows. Best frontier agent overall: 28% pass@1. End-to-end prior authorization: 0%. A thread on what we found 🧵

1/ Introducing CHI-Bench 🧵 Can AI agents automate U.S. healthcare workflows end to end — given only clinician & insurer apps, operations, and a medical policy library? 75 long-horizon workflows × 30 frontier agents. Best agent solves just 28%. #AIinHealthcare 👇

1/🧵Can AI agents automate U.S. healthcare workflows end to end given just clinician & insurer apps and operations, medical policy library? Introducing CHI-Bench: 75 long-horizon realistic healthcare workflows × 30 frontier agents. Best agent solves only 28% #AIinHealthcare 👇

One thing I’d emphasize, though, is that the real value of the FDE model comes when the actual product developers work directly with customers to solve problems and ship features, without relying on third-party implementation firms. Otherwise, it starts to look more like a traditional professional services model rather than a true product engineering feedback loop.

Product engineers beating implementation firms is the right call. But it has a prerequisite that's easy to miss. Those engineers need enterprise scar tissue. If your FDEs have never shipped to a Kaiser or an ADP, fast iteration just means iterating fast on the wrong things. Speed without judgment. Our team built agents for Kaiser, ADP, RBC, and Disney back at Salesforce, through Agentforce and the AI research labs. We saw what enterprise breaks on before we touched actAVA. That's what makes the FDE loop actually compound. Not just product engineers in the room. Product engineers who already know where things break.

The Problem: AI capability is racing ahead, but healthcare adoption is stuck. Modern AI can answer clinical questions, automate workflows, and coordinate care—yet most healthcare orgs can’t safely deploy it. Why? Three reasons: 1️⃣ No playbook. There’s no incumbent AI agent product. Every org is building from scratch—often learning (and failing) inside live clinical environments. 2️⃣ Healthcare complexity. You can’t just plug in a frontier model. You need HIPAA compliance, clinical accuracy, role-based guardrails, and deep integration with FHIR, claims, and dozens of EMRs. One hallucination isn’t a bug—it’s a patient safety incident. 3️⃣ Generic models fall short. Frontier models are built for everyone, which means they’re optimized for no one. Healthcare demands specialized intelligence, not general-purpose AI. Our unfair advantage: We build AI that beats frontier models in healthcare. We come from Salesforce AI Research—the team behind Agentforce and advanced post-training techniques. We’ve shown that small, specialized models can outperform frontier giants when purpose-built. Our edge is data synthesis + post-training. We simulate real healthcare workflows—CDI, prior auth, patient conversations—to train models that follow clinical protocols and respect healthcare constraints in ways GPT-4 or Claude don’t. That’s why we built actAVA, a healthcare AI platform with compounding value: BLUE: Role-aware, healthcare-native agents with FHIR connectors and HIPAA compliance. RED: Automated safety testing for hallucinations, PHI leakage, and bias. GREEN: A learning flywheel that post-trains specialized models from real workflows—cheaper and better than frontier APIs. GOLD: Monetize institutional knowledge by licensing private, post-trained models as revenue-grade APIs. The result: AI that gets smarter, cheaper, and more profitable over time. We don’t sell software—we create compounding advantage. Every deployment generates data. Data trains better models. Better models reduce costs and unlock new revenue. This is the flywheel healthcare has been waiting for. Ready to turn workflows into competitive advantage? Let’s talk. #HealthcareAI #AIAgents #AICompliance #EnterpriseAI

👤 Patients Are Demanding to Know: How Did AI Make That Decision About My Care? A patient advocacy group just filed complaints with HHS demanding that hospitals disclose when AI influences treatment decisions. Their argument is simple but powerful: If an algorithm affects your care, you have a right to know—and to question it. The problem? Most healthcare AI operates as a "black box." Even the physicians using these tools often can't explain how the AI reached its conclusions. This creates an ethical and legal nightmare: How can you provide informed consent when neither the patient nor the doctor fully understands the decision-making process? The "right to explanation" is already enshrined in GDPR for European patients. American patients are now demanding the same transparency. Several states are considering legislation that would require disclosure of AI in clinical settings. Healthcare organizations are caught unprepared. Many don't even have complete inventories of the AI tools used across their systems, let alone the ability to explain how they work. Transparency isn't just ethically right—it's becoming legally required. The AI systems we deploy today must be explainable tomorrow. How does your organization document and communicate AI use to patients? Are you prepared for transparency requirements? Let's discuss strategies in the comments below! #PatientRights #AITransparency #InformedConsent #HealthcareAI #ExplainableAI #PatientAdvocacy #DigitalHealth #HealthcareEthics #MedicalAI #AIAccountability

If AI Misdiagnoses a Patient, Who Goes to Court? The Answer Is Terrifying. A landmark case currently in federal court could reshape healthcare AI forever: A patient is suing a hospital, a physician, AND an AI vendor after an algorithm missed a cancer diagnosis. The question before the court: Where does liability begin and end? The hospital claims it relied on FDA-cleared technology. The physician argues they followed the AI's recommendation as trained. The AI vendor says its tool is "decision support," not diagnostic. Everyone is pointing fingers, and the legal framework to sort this out doesn't exist. This ambiguity is freezing innovation. Healthcare organizations are hesitant to fully integrate AI because liability insurance policies haven't caught up. Meanwhile, AI vendors are burying themselves in indemnity clauses and limiting language that leaves providers exposed. We're operating in a regulatory vacuum where 19th-century malpractice law is being applied to 21st-century technology. The result? Defensive medicine meets defensive AI deployment, and patients lose. We need clear liability frameworks before another patient becomes a test case. Healthcare leaders and legal experts: How are you addressing AI liability in your contracts and policies? Drop a comment or reach out—this conversation is urgent. #HealthcareLaw #MedicalMalpractice #AIliability #HealthcareAI #LegalTech #DigitalHealth #HealthPolicy #RiskManagement #MedTech #HealthcareCompliance

AI Promised Equal Care. Instead, It's Amplifying Disparities. New research published this month reveals that widely-used AI diagnostic tools show significantly lower accuracy for Black and Hispanic patients compared to white patients. The reason? Training data that predominantly represents white patient populations. This isn't just a technical problem—it's a civil rights issue that's gaining traction in federal courts. Several healthcare systems are now facing discrimination lawsuits alleging that AI-driven care recommendations systematically disadvantage minority patients. The healthcare AI industry has a diversity problem that goes beyond representation in boardrooms. It's baked into the algorithms themselves. When your AI learns from historical data that already reflects systemic bias, it doesn't just replicate inequality—it automates and scales it. OCR (Office for Civil Rights) has started investigating whether the use of biased AI violates Title VI of the Civil Rights Act. The message is clear: "We didn't know" is no longer an acceptable defense. Healthcare leaders must audit their AI systems for bias before regulators—or lawyers—do it for them. Is your organization conducting bias audits on AI tools? What metrics are you using? Share your approach—we need to learn from each other. #HealthEquity #AIbias #HealthcareDisparities #AlgorithmicJustice #InclusiveAI #HealthcareAI #DiversityInTech #PatientCare #HealthJustice #ResponsibleAI

🚨 Patient data is training AI & most people have no idea. A new lawsuit says a major EHR vendor used PHI without consent. HIPAA wasn’t built for AI that can re-identify “anonymous” data. Innovation now collides with legal risk. Transparency will decide who survives. #AIethics

Palantir vs. Percepta AI 👀 When your engineers are the algorithm, how do you protect your IP? The healthcare AI talent war just got legal. #AI #Healthcare #TechLaw

AI in healthcare is leveling up: ⚡️ Stroke diagnosis in 7 mins 💰 $8.5M for chronic care AI 🏛️ Policy momentum from AHA 👩‍⚕️ Women leading the future of AI in medicine From pilot to proof. The real shift is here. #AIHealthcare #DigitalHealth #HealthTech