🤖 U.S. Doctors Are Quietly Using This AI Search System

🥼 U.S. Doctors Are Quietly Using This AI Search System

🏆 Amazon’s Employees Gamify AI Usage Metrics

📍 6 AI Stock Sectors

🛠️ 5 Trending Tools

🥪 4 Brief Bites

💰 3 Funding Frontlines

💼 4 Job Opportunities

It’s estimated that global medical knowledge now doubles every 73 days. “There are about a million new articles added to PubMed every year,” noted Daniel Parente, M.D., Ph.D., assistant professor of family medicine and community health at KU Med. So, what’s the best way to stay current on all this scientific literature?

In the previous two years, U.S. healthcare professionals have been quietly embracing OpenEvidence, an AI-powered medical search system, to streamline clinical decisions at the “point of care.”

For example, a primary care physician struggling with a patient’s resistant hypertension might consult OpenEvidence to find the most up-to-date treatment options, such as prescribing spironolactone, with clickable citations explaining how it helps the kidneys excrete excess salt and water.

In the past month alone, 65% of U.S. doctors consulted this AI-powered medical search system across almost 27 million clinical encounters. For context, more than two-thirds of licensed U.S. physicians, or roughly 650,000 U.S. doctors, actively consult OpenEvidence every day.

On Jan. 21st, 2026, OpenEvidence closed a $250 million Series D funding round at a $12 billion post-money valuation. It’s become the AI era equivalent of consulting a colleague for their expert opinion.

Amazon created “MeshClaw,” an internal AI automation tool designed for technical teams to build, train, and run AI Agents that execute multi-step workflows on their behalf. To enforce this, the online retail giant set up AI leaderboards to monitor the number of tokens employees process.

A token is a unit of text that AI Agents process to understand inputs and generate outputs. So, more tokens being utilized means more inputs and outputs being processed.

The technical teams realized they could exploit “MeshClaw” to triage empty emails, scan Slack channels, and run pointless loops of code to burn millions of tokens, a tactic known as “tokenmaxxing.”

Instead of pretending to be really busy when the boss walks by, employees these days are trying to make their AI Agents look completely swamped. One OpenAI engineer burned through 210 billion tokens in just one week. That’s enough tokens to fill Wikipedia 33 times over. One Anthropic engineer used roughly $150,000 worth of tokens on Claude Code in a single month.

This free-wheeling “tokenmaxxing” era won’t last forever. It’s a classic case of Goodhart’s Law: when a measure becomes a target, it ceases to be a good measure. Eventually, Big Tech companies will realize it doesn’t tie cleanly to outcomes. They’ll start treating tokens as a constrained resource, not an unlimited one. The goal isn’t just to use more AI. It’s about getting more business value out of every token processed.

Vapi raised a $50M Series B to speak human to every customer.

Exaforce closed a $125M Series B to detect AI-era cyber threats.

Forus landed a $160M Series B to build modern medicine’s AI engine.

Salesforce (New York, NY): Software Engineer Intern, Summer 2027

NVIDIA (Santa Clara, CA): DFT Engineer, Silicone Lifecycle, Entry-Level

OpenAI (San Francisco, CA): Pricing Strategist, Monetization, Mid-Level

Anthropic (Sydney, AU): Data Center Hardware Ops. Lead, Senior-Level