Doximity, Inc. (NYSE: DOCS) has been named a founding partner in Fireworks’ newly launched Specialized Intelligence Index, a cross-industry benchmark for leading AI systems. The digital health platform joins Sierra and Harvey as category leaders spanning healthcare, customer support, and law.

As part of this launch, Doximity released Bedside Bench, an open-source benchmark designed to measure clinical-grade AI. Unlike traditional evaluations that rely on multiple-choice questions to test recall, Bedside Bench utilizes complex clinical scenarios to assess real-life performance and safety at the point of care.

Benchmark methodology and focus areas

The new benchmark targets areas identified by physicians and developers as having the highest impact on medical practice. It moves beyond simple fact retrieval to test an AI’s ability to handle nuanced, high-stakes decisions.

Key test areas include:

• Drug-safety questions where minor errors pose major health risks
• Correct application of evidence from drug guidelines and trials
• Open-ended questions with multiple reasonable answers, differing in completeness and safety
• Prompts containing fabricated drugs or nonexistent studies
• Questions designed to uncover health-equity mistakes and bias
• Ability to select appropriate actions from long lists while avoiding inappropriate ones

Performance against frontier models

When independently graded by Fireworks, Doximity Ask ranked first compared to frontier models. This result follows a previous achievement in July, where Doximity Ask outranked all other U.S.-based models in the independent NOHARM study conducted by researchers at Stanford and Harvard. These outcomes suggest that AI systems built specifically for medicine can exceed general-purpose frontier models on clinically relevant tasks.

Louis-Antoine Mullie, Head of Medical AI at Doximity, stated that clinicians need assurance regarding AI trustworthiness on practical decisions. He noted that Bedside Bench was built around physician-validated, real-world clinical tasks, addressing risks such as drug safety and incorrect use of evidence.

What the numbers show

The dual validation from Fireworks’ independent grading and the July NOHARM study highlights a consistent pattern: specialized medical AI outperforms general-purpose frontier models on domain-specific safety metrics. This divergence suggests that general-purpose scale does not automatically translate to clinical reliability, reinforcing the value of domain-specific benchmarks like Bedside Bench.

Doximity, Inc. (NYSE: DOCS) released its 2026 Physician Compensation Report, revealing that nearly a quarter of US physicians expect artificial intelligence to increase their total compensation within the next year.

The digital platform for medical professionals found that 23% of surveyed doctors anticipate higher earnings from AI adoption in the next 12 months. A larger majority, 67%, believe that staying current with AI tools will provide a meaningful earnings advantage over peers who do not adopt the technology.

What the Numbers Show

The data highlights a divergence between workforce strain and technological optimism. While 82% of physicians reported being overworked in a June 2026 poll, a significant portion views AI as leverage rather than a threat. Specifically, 46% of overworked physicians are eyeing early retirement, up from 34% previously, yet 39% report that AI proficiency is now a factor in hiring and promotion decisions for their specialty.

Workforce Strain and Compensation Trends

The physician shortage continues to impact clinical practice, with 85% of doctors stating it has affected their work. 76% reported that systemic pressures compromised the quality of care provided in the past 12 months. Among those feeling overworked, 66% are considering a career change.

Broader compensation metrics show moderating growth:

The gender pay gap held steady at 26% for the second consecutive year. The disparity between surgical specialists and primary care physicians widened, with specialists now earning 90.1% more than primary care providers, compared to 87.3% in 2024.

AI as Professional Currency

AI proficiency is emerging as a critical career asset. At least 39% of physicians indicated that AI skills influence hiring and promotion decisions, including 15% who described it as a "major" or "moderate" factor. Doximity, which has over 85% of US physicians as members, bases these findings on salary data from more than 23,000 responses collected in 2025 and new survey findings from June 2026.