← All organizations

Healthcare AI and clinical workflow automation

Anterior

Anterior builds clinical AI for health plans to automate administrative work, including prior authorization, claims adjudication, and risk adjustment. Its Actions are modular tools for gathering clinical records, checking information, structuring documents, applying medical policies, and generating summaries. Health plans can integrate these tools into their own agents and workflows or use managed deployments with Anterior’s clinicians and engineers. The platform is FHIR-native and API-first, with audit logs and human review built into workflows.

Founded in 2022 by Abdel Mahmoud and Zahid Mahmood, Anterior is led by co-founder and CEO Abdel Mahmoud, a physician and former Google product leader. Its engineering combines OCR and vision to process clinical documents while preserving references to original files and pages. A rules compiler converts medical guidelines into version-controlled decision trees; the reasoning engine records the evidence behind each decision. Its prior-authorization AI recommends approval or escalation to a clinician, rather than independently denying care.

In 2026, the company reported supporting organizations covering 50 million lives, with production deployments including Geisinger Health Plan. Anterior also closed a $40 million funding round that year, bringing total capital raised to $64 million, with participation from NEA, Sequoia Capital, FPV, and Kinnevik.

www.anterior.com

Start here

  1. Mission-Critical Evals at Scale: Learnings from 100,000 Medical Decisions

    Start here to learn how confidence scores can route uncertain medical decisions to stronger models or clinicians as review volume grows.

    Christopher LovejoyAI Engineer Summit 2025

  2. Don't be data poor

    Learn how a structured synthetic-data workflow can support evaluation when protected medical records cannot be retained, while reducing expensive ground-truthing.

    Anuj IravaneAI Engineer World's Fair 2026

  3. Why Your Enterprise Tech Stack Isn't Ready for AI Agents - And What to Build Instead

    Use this discussion to understand why a working healthcare agent prototype can still face deployment barriers in enterprise integrations and customer-controlled infrastructure.

    Christopher Lovejoy · Saul HowardAI Engineer World's Fair 2026

  4. Make your LLM app a Domain Expert: How to Build an LLM-Native Expert System

    Christopher Lovejoy, presenting for Anterior, explains how to build domain-native LLM applications around clinician expertise rather than model sophistication alone.

    Christopher LovejoyAI Engineer World's Fair 2025

Messages from the stage

Turn clinical judgment into evaluation workflows

Christopher Lovejoy describes categorizing extraction, reasoning, and rules errors, prioritizing harmful failures, and monitoring regressions. His evaluation talk adds expert ground truths, reference-free judging, and confidence scoring to address the limits of fixed-rate manual review.

Build diversity into synthetic records

Anuj Iravane explains why one-shot generation produces insufficiently diverse medical records. His alternative combines policy-guided symbolic decision trees, scenario sampling, and coarse-to-fine construction with clinical expertise.

Design infrastructure for human and model oversight

The joint session with Saul Howard discusses application, control, and data planes alongside audit trails, access controls, and clinician escalation. It argues for consistent context and oversight across human and model agents.

Affiliations reflect each recorded session, not necessarily current employment. The enterprise AI-agent session is a joint discussion with Saul Howard of Anterior and Christopher Lovejoy of Anthropic.

Company sources · checked 2026-08-27