← All organizations

AI for life sciences

Log10

Log10 builds Everest, an AI platform for life sciences that helps pharmaceutical, medical-device, and contract research organizations prepare clinical, regulatory, and commercial documents. Teams provide prompts and source data to generate intermediate reports and final drafts with quality checks and validation. Its scope includes clinical study reports, regulatory applications, and the statistical work behind them: Everest generates tables, listings, and figures alongside R and SAS code to reproduce the analysis.

Co-founders Arjun Bansal, CEO, and Niklas Nielsen, CTO, bring experience from AI infrastructure: Bansal co-founded Nervana Systems, while Nielsen previously led product at MosaicML. Log10’s research includes ClinReg, a benchmark covering literature screening, regulatory drafting, and clinical-trial analysis. Its evaluation approach combines multiple model judges with deterministic checks for missing or incorrect extracted values, and validation during task execution—addressing both document fidelity and the reliability of generated analyses.

Log10 raised $7.2 million in seed financing in 2024, led by TQ Ventures and Quiet Capital, with Essence Venture Capital participating. Its commercial partnerships include Merigold, whose Bloom AI combines imaging-trial expertise with Everest to draft imaging charters and other study documents. In their 2025 partnership announcement, the companies reported up to 70% faster study start-up at Merigold.

everest.log10.io

Start here

  1. Best Practices for Evaluating Large Language Model Applications with llmeval: Niklas Nielsen

    Start here to learn how Hydra-based configuration organizes local evaluation work and how reports compare results across Claude, GPT-4, and GPT-3.5.

    Niklas NielsenAI Engineer Summit 2023

  2. What It Actually Takes to Deploy GenAI Applications to Enterprises

    Learn how conversation-intelligence systems normalize customer interactions and apply configurable LLM pipelines to surface insights across conversation datasets.

    Arjun Bansal · Trey DoigAI Engineer World's Fair 2024

Messages from the stage

Explicit criteria for evaluation

Nielsen’s demonstration pairs strict output comparisons with customizable metrics and model-based grading, including scored assessments of Mermaid diagrams.

Trust and throughput in enterprise systems

Bansal and Doig discuss using Log10 AutoFeedback to address weaknesses in LLM-as-a-judge evaluation and track hallucinations. Their joint session also considers self-hosted models for production-scale processing.

Affiliations reflect each recorded session, not necessarily current employment. The enterprise deployment session is a joint discussion with Arjun Bansal of Log10 and Trey Doig of Echo AI.

Company sources · checked 2026-08-28