A practice, not a SaaS.

Yevheny Panin — banker and trader. Two professions that touch the same problem: the measurement of money.

• · Banker, three years as the international-payments head at a European commercial bank — responsible for the operating cost-and-yield economics of a cross-border payments book that processed billions in annual flow. The relevant skill is reading invoice-line-item data, distinguishing genuine optimisation from cosmetic re-pricing, and writing a settlement contract that survives an audit. This is the spine of how Tessera is structured.
• · Trader, five years on the floor — execution and pricing across FX, indices, and metals. Where market makers and clients have asymmetric information on the cost of liquidity, skin-in-game structures and audit-immutable measurement are how the interests get re-aligned. Tessera applies that same structural discipline to AI inference pricing — a flat subscription as the only line item, every saved dollar measured directly from proxy logs.

Yevheny holds the principal advisory role at Tessera. Account onboarding, recommendation reviews, and high-volume Enterprise / MSA discussions go through him directly. There is no account-management layer between the practice and the Client.

A SaaS sells per-seat access to a tool. A practice sells a measured outcome.

Tessera's economic structure — a flat monthly subscription priced by token volume on the paid tiers, every dollar of measured savings kept by the customer, a permanent Free Sandbox at sixty million tokens per month with zero cost, kill-switch always available — is only honest when the work is measured per-request rather than per-seat. A SaaS observability dashboard that surfaces inference spend without changing it is selling visibility, not savings. Both are useful; we are explicitly the second thing. We sit in the request path and do the optimisation.

We borrow the structural cues of a professional-services firm — audit-immutable measurement, every saved dollar traced and visible in the dashboard, vendor neutrality codified in the Terms — because those cues are what make the flat-subscription model legible to a CFO: one predictable monthly line item, and the savings it produces kept entirely by the customer. The paid-tier signup is frictionless (Stripe Checkout + ToS click-through, no engagement letter); the trust comes from the measurement loop, not from a signed document.

• · We don't accept affiliate revenue, referral fees, or any other compensation from AI providers, gateway vendors, or observability platforms we may recommend. Client fees are the only income.
• · We don't write code in your repositories. We never receive write-access to your inference endpoints. The recommendation lives in the Monthly Joint Reading; the implementation lives in your team.
• · We don't sell the data we ingest, train on it, or use it in any context outside the engagement that produced it. The Joint Baseline is a closed-book document between Tessera and the client.
• · We don't claim certifications we have not earned. See the Security posture page for an honest snapshot.

The practice is incorporated and headquartered in Tallinn, Estonia. The team is small and remote across European time zones, with the engagement footprint structured around the standard EU/UK business day plus deliberate two-hour overlap with US East. Founding team is two operators. Cap on cohort size — four to six new Pilots per quarter — keeps the practice from drifting into a generic agency.

Two concrete shapes carry 95% of Tessera paid-tier signups in Year 1:

• · Primary — Margin-Squeezed AI SaaS CTO. Series Seed-B AI-native B2B SaaS (vertical copilots, agent platforms, customer support AI, knowledge tools, RAG search, content gen, dev tools). Ten to one hundred people total. LLM spend twenty to two hundred thousand US dollars per month, growing twenty to forty per cent month over month. Trigger: the finance director or board has asked why AI COGS is growing faster than ARR. Decision speed: seventy-two hours from demo to a paid tier. No procurement, founder-to-CTO call.
• · Secondary — AI Platform Lead at a Funded Scale-up. Series B-D established SaaS adding AI features (CRM, dev tools, security, fintech, healthtech, sales/marketing). One hundred to five hundred people, dedicated AI platform team of three to ten. LLM spend fifty to five hundred thousand US dollars per month, central platform team owns the budget. Decision speed: two to four weeks, light security review, no contract redlines on the paid tiers.

Above twenty billion gross tokens per month, Enterprise Clients may negotiate a custom MSA addendum — invoice billing (NET-30/45/60), custom SLO, custom rate, and a dedicated commercial conversation. This is a small minority of Clients and runs on a different motion entirely.

Workloads tagged regulated (HIPAA, PCI-DSS, SOC 2 in-scope) never auto-route — the compliance gate blocks routing at the code level. Tessera still measures and reports on regulated workloads, but the four optimization moves remain dormant on them. If one hundred per cent of your stack is regulated, the Optimize Layer's economic surface area is zero — see Not for you.

Engagement enquiries: contact@tesseraai.io.

Or start with the apply form at the bottom of the landing — eight short questions, response within five business days.