Private Trial · launch offer

Private comparison of two versions

Your agent has changed. Has it actually improved?

ScoreIA compares A and B on the same cases and shows what your new version fixed, kept or broke.

Internal technical reference · synthetic · not a client case or model evaluation

Private Trial · launch offer

€690 excl. VAT · €190 deposit · €500 balance · 5 seats

1 agent · 1 mission · 2 versions · 10 scenarios · 20 executions

Founding seats: 5 remaining.

Request my free dry run See the internal synthetic HOLD reference

You provide 2 ready versions, 1 mission and 1 critical error. ScoreIA provides a dry run and 2 example scenarios before payment. You keep your keys.

Payment never buys a verdict. The public Chamber stays free.

Understand the verdict · interactive demonstration

Two fixes. One regression. Why HOLD?

Explore A, B and a targeted correction C in a 3D scene, select each case and verify local signatures. Three scripted synthetic cases, not a model evaluation or a customer case; this demonstration does not replace the ten-scenario Trial.

Explore the release scene

Deliverable

What you get

Frozen trial contract

The decision, versions, ten cases, critical failures and rules are approved before the first run.

20 paired executions

A and B face the same ten situations from the same host, with one primary changed variable.

A readable decision

A case-by-case matrix shows what B fixed, kept or broke—without hiding a critical failure inside an average.

Verifiable evidence

Private cards, redacted replays, digests, signatures and a Release Receipt bound to the tested configurations.

30–45 minute review

We walk your team through the verdict, limits, critical cases and priority corrections.

3 targeted reruns

One rerun of each of the three selected critical cases is included during the 30 calendar days following the review, without erasing the original result.

From framing to decision

How it works

  1. Free qualification. Within two business days, you receive a feasibility and scope answer.
  2. Free dry run. One case checks connectivity, tools, sealing and the absence of secrets. No deposit if the integration does not work.
  3. Contract, then deposit. You approve the changed variable, critical failures, confidentiality and deliverables, then pay €190 excl. VAT.
  4. Approved suite. Ten cases from existing ScoreIA scenario families are parameterized for your mission. You review two examples before the deposit, then approve all ten cases and their oracles before the first run. The launch offer does not include building a custom trial.
  5. A/B campaign. You launch A and B from your host and keep your keys. ScoreIA observes only what the contract permits.
  6. Control and sealing. All twenty artifacts are checked, compared and bound to their evidence. If a client oracle is not independently re-executed by ScoreIA, the Release Receipt says so explicitly.
  7. Delivery. Report and evidence within five business days after the twentieth valid seal, followed by the review within five business days and the €500 excl. VAT balance invoice, payable within 15 calendar days.
  8. Corrections. One rerun of each of the three selected cases remains available during the 30 calendar days following the review.

No surprises after payment

Your effort and our responsibility

What you do

  • Appoint one owner able to validate the release decision
  • Provide launch-ready A and B without sending us your keys
  • Approve the ten cases, critical failures and allowed data
  • Launch both versions from your environment
  • Allow time for framing and a 30–45 minute review meeting

What ScoreIA guarantees within the Trial

  • The same contract and same cases for A and B
  • Private by default, with no implicit publication
  • Immediate notice when a campaign is blocked or invalid
  • No critical regression hidden by an aggregate score
  • The original history remains intact when fixed cases are rerun

Indicative schedule: two example cases within 3 business days after the dry run and before any deposit, full suite within 5 business days after the deposit, a 10-business-day client execution window, report within 5 business days after the twentieth valid seal, then the review within 5 business days. Any scope change is disclosed before execution, never discovered on the invoice.

Perimeter

What €690 excl. VAT buys — and what it does not

Included

  • One business decision
  • Two versions
  • Report and replays
  • Three targeted reruns
  • No custom integration

Out of scope

  • Custom connector
  • CI/CD
  • Sensitive data
  • General certification

After the Trial

A standalone professional service

  1. Your starting point

    One mission, two versions

    Qualification and dry-run before any deposit. No Chamber or Labyrinth participation is required.

  2. Private Trial

    Create the baseline

    €690 excl. VAT · 5 seats

  3. Regression Watch

    Watch later versions

    €290 excl. VAT / month after a delivered Trial

Method and transparency

Payment never buys a verdict. It funds framing, the private trial and the report. 10 paired scenarios are an exploratory diagnostic, not a claim about the model. 30 paired seeds per version = Benchmark 30 option (+€590 excl. VAT), not this shelf. An Ed25519 signature is not provider identity.

Method sample (1 config, not A/B) · A/B 10×20 skeleton · Verifiable synthetic HOLD reference (3/6/1) · Method pipeline status (20/20 grok-4.6 × cursor — not a ranking)

Three MCP doors. Public Chamber: https://scoreia.ai/mcp · Auth none. Public Labyrinth: https://scoreia.ai/open-labyrinth/mcp · no authentication — it is free and stays free. Private MCP: https://scoreia.ai/private-trials/mcp · Auth Bearer spt-. Chamber free.

No online payment is open. Requesting a free dry run sends a brief: no reservation, token or deposit. Reply within 2 business days. A dry run and two example scenarios precede any deposit. After the Trial Order Form is approved, ScoreIA sends the €190 payment link individually and dates the full schedule. Five seats. €690 excl. VAT = €190 deposit + €500 balance.

Admission · no secrets

Frame the two versions

No API key, token or sensitive personal data. You keep your keys and launch the agents from your host. ScoreIA issues the private token after the dry run, technical acceptance and recorded deposit, hosts the trial and seals the results. Immediate acknowledgement. Reply within 2 business days. This form never issues a token.

1. Qualification
2. Feasibility

Trial cards are private by default. Publication is never inferred from this form. If several variables change, ScoreIA can compare the releases but does not claim that one individual change caused the difference.