Yichen Product Lab · Portfolio 2026

Products built around
proof, not promises.

Selected products

Yichen Product Lab

Three focused products. One shared discipline: evidence first, automation bounded, people in control.

01 / 03UI prototype

Product intelligence

Yichen Lens

Plan the research. Inspect the evidence.

A controllable product-intelligence workspace for one- or two-product research, with an editable plan, consistent comparison schema, and evidence one action away.

Designed for
Product managers
Signature workflow
Research plan → report → evidence
  • Source-aware
  • Editable report
  • Explicit unknowns
View product thinking
YLYichen LensProduct intelligence for confident decisions
Interactive product research
Your product intelligence companion
Research products with a plan you control.

Turn a business question into a structured, evidence-linked report.

Research planStripe vs. AdyenReady to review
Business decisionChoose the payment platform that best supports European expansion.
  1. 1Confirm productsPrepared
  2. 2Review the planPrepared
  3. 3Research public sourcesNext
  4. 4Inspect the evidenceNext
02 / 03Live product

Release assurance

Yichen Guard

Test the release. Gate it with evidence.

A release-assurance workspace for reproducible KSeF scenarios, semantic diffs, deterministic assertions, and an explicit human PASS / WARN / BLOCK gate.

Designed for
ERP and integration teams
Signature workflow
Scenario run → assertion matrix → gate
  • Reproducible
  • Deterministic first
  • Human-gated
Open live product
YGYichen GuardAI-powered e-invoicing release assurance
Demo mode
Version-aware regression assurance
Release confidence,
backed by evidence.
Current readiness Action required4 failures · 2 require reviewOpen human decision gate →
Deterministic resultFrozen portfolio runrun_portfolio_mock_v1
Total12
Pass5
Warn1
Fail4
Review2
FAILGross amount decimal shiftExpected 9876.54Observed 987.65
03 / 03Live product

Exception operations

Yichen Resolve

Turn the unresolved queue into controlled decisions.

A finance-operations workspace that explains deterministic matching, gathers source evidence, and keeps bounded AI guidance separate from the recorded human resolution.

Designed for
Finance operations teams
Signature workflow
Exception queue → evidence → resolution
  • Evidence-led
  • Bounded guidance
  • Append-only audit
Open live product
YRYichen ResolveAI-powered financial exception resolution
Reconciliation run · ready for review
August PSP payout close

Deterministic processing completed with 5 match groups and 3 cases.

Run manifestshowcase-2026-08-20.1resolve-rules-bootstrap-v0
PSPNorthstar settlement4 input rows
BankMeridian EUR account5 input rows
ERPAtlas cash clearing5 input rows
Deterministic resultCounts reconcile to every input row.14 input = 13 accepted + 1 rejected

Product thesis

AI earns trust
in the details.

Useful intelligence is not a magic layer. It is a carefully designed relationship between evidence, boundaries, and human judgment.

  1. 01
    Show the evidence

    Make the source, reasoning, and uncertainty inspectable.

  2. 02
    Design the boundary

    Be explicit about what AI can recommend—and what it cannot decide.

  3. 03
    Build for the operator

    Fit intelligence into the real workflow, not a separate novelty experience.

Behind the work

Product thinking shaped by
platform delivery.

Meet Yichen