Touchmark GTM brain

Touchmark operates a market for buying and reselling future AI model inference capacity.

category

Reserved AI inference marketplace

market profile

  • text: AI application operators can buy inference on demand or reserve capacity directly from providers, trading flexibility against price and service guarantees.
  • text: Touchmark matches buyers and providers using fixed delivery windows, service levels and a resale mechanism.

personas

  • title: AI infrastructure or platform engineering lead
    pains:
    • Must secure throughput and latency for predictable production demand
    • Needs a fallback when model traffic forecasts change
  • title: Head of finance or cloud procurement
    pains:
    • Variable token costs make margins and budgets hard to forecast
    • Long capacity commitments can strand spend if usage changes
  • title: Inference provider commercial lead
    pains:
    • Needs forward demand visibility to plan GPU utilization
    • Must attract buyers without negotiating every deal from scratch

icp

  • text: AI product companies with material recurring inference volume and sufficiently forecastable baseline demand to reserve tokens.
  • text: Model inference providers able to commit capacity and service levels for named models and delivery windows.

message pillars

  • text: Reserve inference against a defined contract and service level.
  • text: Use competition among providers to find a price.
  • text: Retain flexibility if demand changes.

objections

  • objection: Our forecast may be wrong and capacity may expire unused.
    response: Size reservations to a conservative baseline and assess resale liquidity and expiration terms before committing.
  • objection: We need reliable production performance.
    response: Compare the named provider's availability and latency commitments with your workload requirements.
  • objection: How does the quoted price compare to our current spend?
    response: Calculate effective input, output and cache costs using the listing's token weights and actual traffic mix.

competitors

  • name: Fireworks AI
    why: Direct inference provider and alternative source of capacity.
  • name: Together AI
    why: Offers model inference and reserved capacity directly.
  • name: OpenRouter
    why: Aggregates inference providers for buyers, though its routing model differs from a forward market.

keywords

  • reserved inference capacity
  • LLM inference pricing
  • buy AI tokens in advance
  • inference capacity marketplace
  • AI inference SLA
  • LLM throughput reservation
  • reduce inference costs

voice

  • Financial market language
  • Concise and technical
  • Commercially focused

signals

  • text: AI inference spend is rising and a predictable baseline can be separated from peak demand.
  • text: A provider seeks forward commitments to improve capacity planning.

stack

  • OpenAI-compatible API

pricing

model: unknown

motion

hybrid

channels

  • Y Combinator launch
  • LinkedIn
  • press

confidence

medium

Founders

  • name: Ilia Bolgov
    title: Founder/CEO
    role: CEO
    bio: Co-founder of Touchmark. Product @ Revolut, Maths @ Imperial College London.
  • name: Roman Yanushevskyi
    title: Founder/CTO
    role: CTO
    bio: Co-Founder at Touchmark. Previously 2x Quantitative Research intern at Citadel Securities, AI Engineering intern at Lovable. Math and CS graduate from the National University of Singapore. IOI gold medalist.

Back to YC companies