Probe GTM brain

Probe gives AI research teams experiment tracking, shared agent context and compute observability for coding-agent research.

category

Agent-native ML experiment tracking

market profile

  • text: AI researchers need to preserve experimental context and compare runs across researchers and agents.
  • text: Experiment tracking competes with established platforms and open-source tooling.

personas

  • title: AI research scientist or research engineer
    pains:
    • Agents repeat failed experiments when prior decisions are missing
    • Comparing baselines requires reconstructing metrics and logs
  • title: ML infrastructure lead
    pains:
    • Failed runs and idle GPUs waste compute budget
    • Setting up compute and resuming interrupted jobs slows research
  • title: Head of AI research
    pains:
    • Researchers and agents lack a shared view of ongoing work
    • Research knowledge is lost in handoffs and status meetings

icp

  • text: AI and ML research labs running iterative training experiments with coding agents.
  • text: Teams using Weights & Biases that want to carry existing projects into an agent-oriented research workflow.
  • text: GPU-intensive teams with multiple researchers and expensive or interruptible runs.

message pillars

  • text: Record research automatically as coding agents work.
  • text: Keep shared decisions and prior experiments available to agents.
  • text: Identify failed or unhealthy GPU runs sooner.
  • text: Offload compute setup to the agent.

objections

  • objection: We already use Weights & Biases; would migration discard our work?
    response: Probe says it can sync existing W&B projects and offers tracking parity; test a representative project.
  • objection: Can this protect expensive research runs?
    response: The site describes failure alerts and checkpoint resumption, but says shutdown saves depend on time available.
  • objection: Does it fit our coding agent?
    response: The site names Claude Code, Codex and Cursor; validate your environment with a pilot.

competitors

  • name: Weights & Biases
    why: Established collaborative experiment tracking platform that Probe explicitly compares against.
  • name: MLflow
    why: Open-source alternative for experiment tracking and ML lifecycle management.
  • name: Comet
    why: Experiment tracking and ML platform competing for research workflows.

keywords

  • ML experiment tracking
  • coding agent research memory
  • GPU experiment monitoring
  • AI research infrastructure
  • Weights and Biases alternative
  • automated ML research
  • experiment reproducibility

voice

  • technical
  • direct
  • researcher-oriented

signals

  • text: Research team is scaling coding-agent use and repeatedly rerunning prior experiments.
  • text: A lab reports wasted GPU capacity or lost context after failed training runs.
  • text: Team uses W&B and wants a shared context layer for agents.

stack

  • Claude Code
  • Codex
  • Cursor
  • Weights & Biases
  • Harbor

pricing

model: unknown

motion

product_led

channels

  • X
  • developer documentation
  • product website

confidence

medium

Founders

  • name: Mahit Namburu
    title: Founder
    role: Founder
    bio: Co-Founder and CPO @ Probe. Previously studied EECS and conducted SpecDec research at Berkeley. Worked on inference for LoRA fine-tuned models at Databricks. Developed machine translation models at Apple that were SOTA with Apple data. Turned down ML at Together AI to start Probe.
  • name: Richard Wei
    title: Founder
    role: Founder
    bio: Co-Founder and CTO @ Probe. Previously studied EECS at Berkeley. Worked at Willow Voice (P25) as first engineering hire; led 0-to-1 development and overhaul of ML infra. Worked at Google on YouTube Shorts (feature went live with millions) and hand interactions for Android XR. Started a drone repair business in middle school. Turned down SWE at Palantir to do YC.
  • name: Ashwarye Yadav
    title: Founder
    role: Founder
    bio: Co-Founder and CEO @ Probe. Previously studied EECS at Berkeley. Repeat founder who built an EdTech startup backed by the Qatar Foundation that scaled to a large user base, and co-founded the Chemistry Catalyst Fellowship with Chemistry VC. Shipped production ML systems at Amazon, Rippling, and Sony. Built DetectBot, a viral AI image manipulation bot that reached hundreds of thousands on Twitter (X). Turned down an FDSE at Palantir to build Probe.

Back to YC companies