01 · Onboarding
Docs → spec in 3 minutes
Drop a PDF, form screenshot, or FAQ doc. Five Claude stages extract every field, rule, and routing path with a verbatim source quote. No hallucinated rules.
Solutions
Automated Expense Policy
Enforcement in Lark
Kopi learns your expense policies from every approver decision and automatically checks every new submission for compliance — before it ever reaches your approval queue.
📖
Learned policy rules
Every approver confirmation is saved as a rule. Kopi builds a living policy rulebook from real decisions — no manual configuration required.
🚫
Spending cap enforcement
Set per-category spending caps. Kopi flags any submission that exceeds the cap with a red alert, before the approver sees the request.
🏷️
Category validation
Expenses submitted under the wrong category are detected and reclassified. Finance teams get clean data without manual correction.
📋
Vendor approval lists
Define approved and prohibited vendor lists. Submissions from prohibited vendors are flagged automatically for human review or auto-rejection.
⏰
Time-based policy checks
Enforce submission deadlines, detect late reimbursements, and flag expense claims outside of allowed submission windows.
📈
Compliance reporting
Track policy violation rates, approval decisions, and rule match rates in the Kopi dashboard. Spot trends before they become audit findings.
How policy enforcement learns
Kopi builds your policy library automatically from approver decisions — no manual setup needed.
01
Approver confirms a decision in Lark
When an approver taps green, yellow, or red on a Kopi check card in Lark, their decision is recorded — including the expense category, amount, and vendor.
02
Kopi saves it as a rule
The confirmed category is added to your rule library. Same expense type next month — Kopi recognizes it instantly without human review.
03
Policy violations are flagged automatically
New submissions are checked against your rule library before reaching the approver. Violations are flagged red. Policy-compliant expenses are pre-approved green.
04
Rules compound over time
The more your team uses Kopi, the more accurate the policy enforcement becomes. Most teams see 60–80% auto-approval rates within 4 weeks.
Related solutions
Free for Singapore SMBs in private beta
No credit card. No time limit. Set up in 10 minutes.
Built on Mirror
One agent reads every policy doc — across every domain.
Policy enforcement is what Mirror does by definition — every rule in a TemplateSpec is a policy clause Mirror extracted with a verbatim quote. There is no separate enforcement engine; the runtime IS the policy enforcement.
02 · Runtime
Decisions in 4ms
runTemplate(spec, request) → Decision is pure code, no LLM at decision time. The spec is data, the runtime is deterministic. Three-color verdict with full routing trace.
03 · Channels
Web · Email · Slack · Lark
A submission can come from anywhere. Reviewers see the same triage card across all surfaces. Web + email is the default; Slack and Lark are accelerators for already-connected tenants.
Citation contract — non-negotiable
Every field, rule, and routing approver Mirror emits carries a verbatim quote from your source document. If she can’t cite it, she drops the item. No fabricated policies — your AI is anchored to your own words.
Read the docs →Architecture, full pipeline detail, page map.