Sciloop GTM brain

Sciloop supplies expert-authored STEM reasoning datasets to AI labs training and evaluating frontier models.

category

Expert AI training and evaluation data

market profile

  • text: As models improve, labs need fresh problems difficult enough to reveal remaining reasoning failures.
  • text: Specialist-authored training and evaluation data competes with broader expert-data vendors and benchmark providers.

personas

  • title: Head of model evaluation
    pains:
    • Existing benchmark questions become saturated or contaminated
    • Difficult new problems require credible independent verification
  • title: Post-training research lead
    pains:
    • Needs fresh, high-quality reasoning examples in specific domains
    • Managing specialist contributors and changing specifications consumes research time
  • title: AI data procurement lead
    pains:
    • Needs predictable delivery and evidence of data quality
    • Must match domain and output format to changing model requirements

icp

  • text: Frontier model developers and well-funded AI labs building mathematical and scientific reasoning models; account size is not stated.
  • text: AI research organizations with recurring requirements for expert-created, cross-checked datasets and evaluations.

message pillars

  • text: Recruit top specialists to create hard problems.
  • text: Target problems beyond frontier model capability.
  • text: Manage sourcing and quality assurance for the lab.
  • text: Deliver continuously as requirements change.

objections

  • objection: How do we trust the answers?
    response: Evaluate cross-checking and expert review on a sample with your own researchers before committing.
  • objection: Will the data be sufficiently challenging and fresh?
    response: Test a held-out sample against your model and agree on refresh, novelty and contamination criteria.
  • objection: Can you match our specific training format?
    response: The intake asks for domain, difficulty, format and volume; validate licensing and delivery terms directly.

competitors

  • name: Labelbox
    why: Offers expert-generated training data and evaluation services to frontier AI labs.
  • name: Surge AI
    why: Provides expert RLHF and post-training datasets.
  • name: Mercor
    why: Supplies domain experts for model evaluations and training data.

influencers

  • name: Osman Siddique
    url: https://www.mitaiconference.org/speakers/osman-siddique
    why: Sciloop co-founder and CTO with physics and mathematics data expertise.

keywords

  • expert STEM training data
  • frontier model evaluation datasets
  • math reasoning benchmark
  • physics AI training data
  • expert annotated reasoning data
  • hard math evaluation problems

voice

  • Ambitious and research-oriented
  • Technical yet accessible
  • Credibility through expert credentials

signals

  • text: A frontier lab is building a new scientific reasoning benchmark or post-training run.
  • text: A research team needs continuously refreshed, expert-reviewed problems tailored to its model's weaknesses.

pricing

model: unknown

motion

sales_led

channels

  • LinkedIn
  • X
  • website
  • expert network

confidence

medium

Founders

  • name: Bilal Asmatullah
    title: Founder/CEO
    role: CEO
    bio: Co-founder of Sciloop (YC F25), On Leave From MIT International Physics Olympiad ’22 medalist.
  • name: Osman Siddique
    title: Founder
    role: Founder
    bio: CTO and co-founder of Sciloop (YC F25). CS student at MIT (on leave). Medalist and two-time highest national scorer at the 2022 International Physics Olympiad.

Back to YC companies