Before the Doctor Says a Word, AI Has Already Finished the Questions

Before the visit, AI has already finished asking the patient all the questions.

A clinical team reports: manual patient data-entry time reduced by more than 80%. Concurrence’s AI agent collects medical history, symptoms, and medication records by voice or text before appointments. By the time the doctor sees the patient, a summary is already in the medical record. For patients recovering from cancer, the agent can also provide support between visits, with emergencies escalated to human clinicians.

One commenter said that, rather than repeating a medication list in the exam room, they would rather use that time to discuss treatment and ask questions. This is also the opening Concurrence chose: solve the most tedious part first, rather than starting with high-risk scenarios like diagnosis. Medical AI keeps talking about disruption, but what actually gets deployed first is often the most unglamorous data-entry work.

Concurrence was formerly named Amigo. Founder Ali Khokhar mentioned in the launch tweet that his mother died when he was 13 from a cancer that could have been detected earlier. He said what they are launching today is “the biggest leap toward superintelligent healthcare”—a system that can see a patient’s future in time and change it before outcomes occur.

In the promotional video, a person whose face is blurred stands indoors, with health data floating beside them: total cholesterol 195 mg/dL, blood pressure 128/82 mmHg, and blood oxygen saturation 97%. This is the information the agent collects before the visit.

The system is called Superintelligent Care, and its research framework has three parts: see the whole patient, act on what is seen, and learn from outcomes.

The whole-patient view is achieved through digital twins. Concurrence integrates clinical data, behavioral patterns, and social and economic factors into a continuously updated computational model to predict the patient’s health trajectory. The underlying infrastructure is called the Health World Model.

Disease processes have recognizable patterns; symptoms often appear in a fixed order, and patients drop out of treatment at specific points. The team calls these abstract patterns Archetypes. Each patient’s treatment outcome becomes the basis for treating the next similar patient.

For an AI agent to qualify to act, it must go through Digital Residency—a simulation engine that trains the agent through millions of simulated patient interactions, following the same logic as physician residency training. Official figures: 100% pass rate on clinical safety metrics, and 98% of patient interactions require no human intervention.

Concurrence says it operates at trillion-token scale, supports more than 10 million patient interactions, serves more than 30 healthcare organizations across 10 countries, and partners with Databricks, Snowflake, Amazon, and Microsoft. It has passed SOC 2 Type II audit and complies with HIPAA standards.

Some of the responses are worth pondering. The real test is whether doctors trust a summary they did not collect themselves. One commenter cited a 2016 study: for every hour U.S. doctors spend face-to-face with patients, they spend nearly two hours on screens and paperwork. Will the time saved be given back to patients, or turn into more appointments? Another commenter said that a 78-year-old reading an entire bottle of pills to a voice agent is also a foreseeable image. Some practitioners think this is one of the best use cases for AI agents. And of course, someone in the comments said, “I built this months ago.”

These doubts are all valid. But separating data entry from care delivery so that doctors understand the patient before they speak—this direction itself deserves recognition. As for whether it can reach the stage of “learning from every patient,” that depends on future research. The more complete framework is on Concurrence’s official Superintelligent Care page.