Free
AVAILABLE NOW$0single account
2 AI investigations per UTC month. 1 saved case, 250 records, 5 revisions and up to 5 MiB saved storage. Single-account access. Local measurements and export are also available in the demo above.
Subscribe FreeAI EVIDENCE INVESTIGATOR / 001
See what
the evidence
supports.
Turn supplied AI test results into a reviewable acceptance record. Recalculate the measures, identify evidence gaps, and export the basis for your decision.
FREE SUBSCRIPTION/2 AI INVESTIGATIONS PER MONTH
EVIDAGIO / EVIDENCE_SNAPSHOTSYNTHETIC
SUPPLIER CLAIM95.0%
95 / 100 scored attempts
ALL SUPPLIED ATTEMPTS79.2%
95 / 120 initiated attempts
95 SUCCESS05 FAILURE20 EXCLUDED
F_01
Different denominator. Different result.
20 supplied attempts sit outside the reported rate.
B_01
Calculation reproduced. Source unverified.
Submitted outcomes remain supplier assertions.
ILLUSTRATIVE SNAPSHOTNO LIVE SYSTEM TESTED
BOUNDED AGENT / VISIBLE TOOLS / STRUCTURED FINDINGSExplore the protocol
01 / FREE LOCAL WORKSPACE
Change the rule.
Inspect the result.
Start with fictional evidence or import a small file. Measurements run in your browser. This workspace does not call an AI model or test a live target.
EVIDAGIO / MEASUREMENT_CONSOLESYNTHETIC DATA
One local case. Up to 250 records.
Import CSV or a JSON array. Required: id, outcome (success, failure, excluded), requiresApproval (true or false). Optional: cohort, costs, human minutes, approval status, execution status, version and source reference. .
Imports replace the current browser session only. Records are not uploaded or saved on this site. Outcomes and origin labels are unverified.
OBSERVED IN SUPPLIED ROWS
%
95 successes ÷ 120 initiated attempts × 100
SUCCESS
FAILURE
EXCLUDED
95% Wilson interval Assumes independent attempts
A rate threshold is one criterion. It does not establish source accuracy, complete coverage, or deployment safety.
Record inspector
120 synthetic task records / declared outcomes
| Record | Cohort | Outcome | Run cost | Human min. | Approval required | Approved | Executed | Source |
|---|---|---|---|---|---|---|---|---|
| Enable JavaScript to inspect and export the sample records. | ||||||||
120 records
LOCAL CALCULATION / NO MODEL CALLS / NO CASE STORAGEFREE LIMIT: 250 ROWS · 256 KiB · ONE SESSION CASE
02 / THE AGENT
The tools are bounded.
The questions are precise.
The hosted investigator checks five aggregate evidence sections, then can inspect individual records. Its coverage panel shows the sections and record count actually read. Open a Free account to run it; the local workspace above calculates measurements without an AI call.
TOOL_01
inspect_evidence(section)
Inspect deterministic metrics, cohorts, declared approval status, tested versions and provenance findings.
TOOL_02
inspect_records(outcome, offset)
Read pages of supplied records. Follow failures, exclusions and approval exceptions back to their source references.
01 / INPUTEvidence + criteria
→02 / INVESTIGATERead-only tool calls
→03 / OUTPUTFindings + references
→04 / REVIEWHuman decision
YOUR REVIEW RECORD
Take the evidence into the decision.
Download a printable HTML dossier from the hosted console: acceptance criteria, recalculated results, source references, inspection coverage and saved review history when available. Keep the JSON archive for the complete evidence. Both exports are included in Free and every paid tier.
Explore the review consoleCode calculates the measurements. AI interprets the evidence and can be wrong. Neither tool contacts a live target. A content hash identifies a snapshot; it does not authenticate its contents.
03 / ENTERPRISE & GOVERNMENT
Bring the decision.
We will follow the evidence.
Start with a synthetic walkthrough. Agree scope and data handling before supplying sensitive material.
ENT_01 / ENTERPRISEMove beyond
Evidence for
Move beyond
the headline metric.
Examine acceptance conditions for support and account-change agents.
Request a pilot ↗GOV_01 / GOVERNMENTEvidence for
the intended use.
Explore an internal acquisition research scenario with explicit scope boundaries.
Request a pilot ↗Enterprise and Government pilots begin with an agreed scope. A walkthrough does not activate team access or imply government authorization or security certification. Do not send CUI, classified or export-controlled material.
04 / SELF-SERVE SUBSCRIPTIONS
Start small.
Make the evidence count.
Subscribe Free for 2 AI investigations per month and 1 saved case. Choose Pro, Growth or Business for more capacity. All four include a printable review dossier in the hosted console. These subscriptions are for individual accounts; Enterprise and Government team offerings begin with a scoped pilot.
Pro
EARLY ACCESS$49/ month
30 AI investigations per UTC month. 25 saved cases. 2,500 records and 50 revisions per case. Up to 50 MiB saved storage. Single-account access.
Subscribe ProGrowth
EARLY ACCESS$149/ month
150 AI investigations per UTC month. 100 saved cases. 5,000 records and 50 revisions per case. Up to 250 MiB saved storage. Single-account access.
Subscribe GrowthBusiness
EARLY ACCESS$399/ month
500 AI investigations per UTC month. 500 saved cases. 5,000 records and 50 revisions per case. Up to 1 GiB saved storage. Single-account access.
Subscribe BusinessEarly-access prices are in USD. Pro, Growth and Business are monthly subscriptions that renew automatically until canceled. These are capacity tiers for individual reviewers; they do not include expert review or team access. Paid access begins after server-confirmed payment and an active subscription. Stripe shows any applicable tax and the final total before you confirm checkout.
Hosted AI allowances reset at 00:00 UTC on the first day of each calendar month, separately from your billing cycle. Unused allowance does not roll over. Once a run starts, failed and canceled attempts count. No usage overages. All four self-serve subscriptions allow up to 200 review entries per case.
Shared hosted capacity is 25 AI investigations across all Free accounts per UTC day. Pro, Growth and Business share capacity of 250 AI investigations per UTC day. Personal allowance does not guarantee an immediate run. Local calculations and saved evidence remain available when daily AI capacity is exhausted.
Saved storage counts current case copies, retained revisions and review notes. The first case, record, revision or storage limit reached applies. A saved snapshot must be under 1.9 MB, including its computed report and accounted metadata; long records may reach this limit before the row cap. Saved storage also shares a 7 GiB service budget and is not reserved capacity. Export and delete saved cases to recover space.
Self-serve cases are private to the signed-in owning account. Shared workspaces and SSO are not included in Free, Pro, Growth or Business. Manage billing and cancel future renewal from your account through Stripe. Existing customers can use the billing portal even when new purchases are paused. For billing or access questions, contact hello@syntonyresearch.ai.
05 / TEAM OFFERINGS
For your team.
Start with a pilot.
Enterprise and Government are team offerings that begin with a scoped pilot. We scope seats, reviewer roles, identity, deployment and data requirements with you before agreeing delivery and activation.
Enterprise
PILOT SCOPINGCustomagreed scope
Agree team seats, reviewer roles, identity and deployment as part of the pilot scope. Implementation and acceptance precede activation.
Request a pilotGovernment
PILOT SCOPINGCustomagreed scope
Mission-specific team access, permitted data, identity and deployment conditions agreed in the pilot scope before activation.
Request a pilotTeam capabilities are scoped requirements for an agreed engagement and require implementation and verification before activation. Shared workspaces, reviewer roles and SSO are not currently available in the standard app. A pilot request creates no subscription or payment obligation; team activation follows an agreed scope. No government authorization or security certification is implied.
AWAITING YOUR QUESTION