01 / THE PROBLEM

Review the evidence.
Not a black box.

Exports repeat records, hide conflicts and invite expensive guesses. OBSIDIAN provides small, inspectable data workflows with explicit limits and human decisions.

Synthetic proof only. No customers, revenue or testimonials claimed.

02 / WHAT EXISTS TODAY

A read-only duplicate audit.

One CSV → exact repeats, case/spacing-only matches, key conflicts → spreadsheet row references, readable summary, source fingerprint.

No merge, deletion, fuzzy identity, accounting conclusion, live-system change or guaranteed import.

03 / THE EVIDENCE

Six records.
Three expected groups.

Exact repeat: rows 2–3. Near match: rows 4–5. Key conflict: rows 6–7.

Fresh execution compared against predeclared exact findings; original bytes and hash unchanged. Clean and malformed-input cases included. Synthetic fixture, not universal accuracy.

Inspect the verification receipt · Inspect actual output

04 / A BOUNDED COMMERCIAL PILOT

CA$29 duplicate audit.

One CSV, up to 2,000 rows. Read-only groups CSV, readable audit and source SHA-256. Agency retains client relationship and migration decisions.

Separate offers: CA$15 cleaning, up to 50 rows; CA$39 quality audit, up to 100 rows / 30 columns / 1 MiB. No bundle, recurring commitment or wholesale terms implied.

Confirm scope, permitted data, transfer safeguards and delivery date BEFORE payment. Intake, payment and payout require verification; no guaranteed delivery clock added here.

05 / TRUST & LIMITS

Small promises.
Checkable results.

Independent expected-value checks. Source preservation. Explicit unknowns. No confidential customer intake until verified and approved safeguards.

Current evidence does not establish general AI intelligence, autonomous customer support, enterprise compliance or production integrations. Costs, margins and willingness to pay remain unvalidated.

06 / THE NEXT CONVERSATION

One real problem.
One agreed scope.

Customers: describe an export-review problem without sharing data. Partners: evaluate the synthetic sample before discussing a subcontract. Advisors: challenge the commercial hypothesis.

Vision — conceptual, not launched: more useful AI-assisted tools with inspectable evidence, consent and human control. No fundraising offer or investment return claimed.