Last updated: 2026-08-19
Touchmark is a quality-adjusted AI billing platform: it scores every output a connected product generates and adjusts what the end customer pays based on that score. Its published AI quality indices are cited by Fortune 500 procurement teams, and poor outputs are discounted or refunded automatically.
About Touchmark
Touchmark is a quality-based AI pricing platform built by Y Combinator S26 founders Ilia Bolgov and Roman Yanushevskyi, founded in 2026 on a simple thesis: most AI products bill by token count even though a correct answer and a wrong one cost the same to generate. The founders combine quantitative finance experience at Citadel Securities with product management at fintech Revolut, a pairing that shows up in Touchmark's mix of eval infrastructure and financial derivative products.
The platform runs automated evaluations on every output a connected AI product generates, then adjusts what the end customer is billed based on the resulting quality score, discounting or refunding weak outputs and letting strong ones command a slightly higher price. Beyond the billing layer, Touchmark publishes daily industry AI quality indices covering pharma, code, and legal use cases, and that same quality data backs tradable financial instruments: forwards, options, and exchange-listed futures that let institutional buyers hedge on AI output quality over time.
Pricing
Custom pricing, contact-based. No published tiers as of June 2026. Pricing is usage-based, negotiated per AI vendor based on eval volume and billing delta volume.
No free tier.
Key Features
- Quality-Adjusted Billing: Every output is scored across 5 eval dimensions, including code quality, instruction following, and hallucination rate, and the price the end customer pays adjusts to match, with weak outputs discounted or refunded automatically.
- 10-Line TypeScript SDK: The @touchmark/sdk npm package drops into an existing AI product in under 10 lines of code: open a session, tag each output with a base price, and apply the quality-adjusted delta Touchmark returns.
- Public Session Pages: Every priced output gets its own URL at sessions.touchmark.ai showing the score breakdown, the price calculation, and a plain-language explanation that end customers can inspect directly.
- Industry AI Quality Indices: Touchmark publishes daily-updated Pharma, Code, and Legal AI quality indices that Fortune 500 procurement teams cite when comparing AI vendor quality.
- Financial Instruments on AI Quality: Forwards, options, and exchange-listed futures settle against Touchmark's eval timeseries, letting institutional buyers and sovereign wealth funds hedge or speculate on AI output quality over time.
- Custom and Third-Party Eval Support: Any deterministic signal or third-party evaluation can plug into Touchmark's [0, 1] scoring scale, letting domain-specific quality criteria influence pricing alongside the default suite.
Pros
- Outcome-based pricing directly cuts AI cost waste: customers pay full price only for outputs that clear an agreed quality threshold, with automatic refunds on poor ones.
- No architectural changes needed for the SDK integration, one of the fastest eval-to-billing setups available in 2026.
- Published industry AI quality indices give enterprise procurement teams a standardized benchmark other eval platforms don't offer.
Cons
- Small, early-stage Y Combinator-backed team with no published pricing, no public compliance certifications, and thin documentation compared to established observability platforms.
- The eval model requires developers to trust Touchmark's default scoring dimensions as fair quality proxies; custom eval integration adds effort for domain-specific use cases.
- Financial derivative products built on AI quality, like futures and options, are novel and largely untested at scale, which will draw regulatory and settlement-risk scrutiny from enterprise legal teams.
Frequently Asked Questions
How much do you pay for Touchmark?
Whatever the two sides negotiate: Touchmark publishes no tiers and prices each AI vendor on eval volume and billing-delta volume. Nothing can be bought without talking to the team, and no free tier sits underneath it. Terms are probably still soft while the company works with early design partners.
Does Touchmark have a free plan?
Touchmark has no free plan or trial. It is a B2B tool for AI companies that want quality-based billing for their own customers, not something an individual signs up for solo. The one thing anyone can access without a contract is the public session viewer at sessions.touchmark.ai, where you can inspect how a specific priced output was scored.
What are Touchmark's closest competitors?
Touchmark has no direct like-for-like competitor on hokai since quality-adjusted billing is a new category, but two adjacent eval tools are worth comparing. Archal, also a 2026 YC company, catches agent failures before they ever reach production; pick it for pre-release testing rather than a live billing layer. Crukx is built for production LLM observability and monitoring, so pick it if you want visibility into running models rather than a system that changes what customers pay.
What separates Touchmark from Archal?
Archal and Touchmark are both 2026 Y Combinator eval startups, but they act at different moments in the pipeline. Archal runs agents against sandboxed clones of GitHub, Slack, and Stripe before code ships, catching failures pre-production, and gives builders a free tier of 100 evals. Touchmark instead evaluates live production outputs and uses the resulting score to adjust what the end customer is billed, with no free tier and custom pricing only. Choose Archal to catch agent failures before release; choose Touchmark if you want output quality to directly drive customer billing.
How long does it take to get going with Touchmark?
Onboarding starts with a beta conversation rather than a signup form, so the wait is measured in emails, not minutes. Once you are approved the SDK work is short: install @touchmark/sdk, create a client with your API key, open a session per user interaction, tag the output with the price you would normally charge, then apply the returned quality delta in your billing system. The quickstart and event reference at docs.touchmark.ai cover the whole path.
Top Alternatives
- Archal: Archal tests agents before the code ships and gives you a free tier to do it. Touchmark works on live output and turns the score into what the customer actually pays.
- Crukx: Observability is Crukx's job: watch the model, read the traces. Touchmark goes one step further and lets the quality score move the invoice.
HokAI guides covering Touchmark
- ReasonBlocks vs Touchmark: Which One Actually Fixes Your AI Bill?: ReasonBlocks cuts wasted AI agent tokens by up to 52%. Touchmark, after its August 2026 pivot, sells forward contracts locking in token prices instead.
- Archal vs ReasonBlocks: You're Not Choosing Between Them, You're Choosing When: Archal tests AI agents in a sandbox before they ship. ReasonBlocks fixes them mid-run after they're live. Real pricing, real gaps, and which to buy first.
- The Three Jobs Hiding Inside "AI Agent Observability" in 2026: AI agent observability tools split into three jobs in 2026: tracing, mid-run intervention, and outcome-based billing. Here is how to tell which one you need.