Capability · Error Detection

Surface errors early.
Route them right.

Compass derives errors deterministically from structured operational signals — starting with attendance and task data. It categorizes them, grades severity into clear bands, and routes each correction to the owner responsible for it. Compass surfaces and routes. Humans correct.

Deterministic from real signals. Decision-support, not blame. No content surveillance. Human-in-the-loop always.

What it is

An observe→correct loop built on structured signals

Most operational errors go untracked until they compound. Compass watches structured data sources — attendance records, task completion states, operational logs — and runs deterministic checks against defined thresholds. When something is out of range, it does not guess; it categorizes the finding, assigns a severity band, and routes it to the right person to close the loop.

🔎
Deterministic detection
Errors are derived from structured signals — attendance gaps, missed completions, threshold breaches. No probabilistic scoring, no AI guessing from unstructured content. If the signal is present and crosses the configured threshold, it is flagged. Math, not inference.
🏷
Severity grading
Every detected error is graded into one of four bands — Critical, High, Medium, or Low — based on operational impact rules you define. A Critical grade means the correction is time-sensitive and needs the right owner immediately. Low means it can be batched.
📤
Routed correction
Compass routes each flagged error to the person with the authority and context to correct it — manager, HR, or the employee themselves — based on the error category and your org's configured routing rules. No correction is left unassigned. No owner is blindsided.
Signal in practice

Errors by category — 8-week view

Compass tracks flagged errors across categories over time. Below is what a typical view looks like: categories accumulate, correction rate trends up as the loop closes, and average time-to-correct narrows as routing improves.

Compass · Error Detection — Category breakdown · 8-week trend
Total flagged · last 8 wks
147
across all categories
Correction rate
84%
↑ 12pp vs prior period
Avg. time-to-correct
18h
↓ from 31h last period

Illustrative data — shows the type of signal Compass surfaces. Real values come from your org's operational data.

Recent flags — live queue sample
CRIT Attendance gap >3 consecutive days — no leave record Routed → HR
HIGH Task completion rate below threshold for 2nd consecutive week Routed → Manager
MED KPI data missing for current period — upload overdue Routed → Admin
LOW Check-in timestamp outside configured work window Routed → Employee
Deterministic Structured signals only Human-in-the-loop
How it works

Five steps from signal to closed loop

01 — DETECT
Structured signal check

Compass reads structured operational data — attendance records, task states, KPI uploads — and runs deterministic threshold checks. A flag fires when a defined condition is met. Not probabilistic. Not guesswork.

02 — CATEGORIZE
Error type assignment

Each flag is categorized by type: attendance, task completion, data integrity, schedule, or compliance. Category determines which routing path applies — so the right owner receives it without manual triage.

03 — GRADE
Severity banding

The error is graded Critical, High, Medium, or Low based on impact rules your org configures. Severity determines urgency in the correction queue and notification priority — Critical flags surface at the top.

04 — ROUTE
Owner assignment

Compass routes each correction to the configured owner — manager, HR, admin, or the employee — based on error category and RBAC rules. No manual assignment needed. The right person sees the right item.

05 — CORRECT
Human closes the loop

The assigned owner reviews, takes action, and marks the correction complete. The loop closes. Correction rate and time-to-correct are tracked so you can see whether the system is actually working — and where bottlenecks form.

Honesty

What Error Detection is not

The word "error detection" can trigger concern. Here is what Compass does not do — and why the distinction matters.

Not content surveillance
Compass does not read email, messages, documents, or any unstructured content. Every error is derived from structured operational signals — records and states, not content.
Not automated punishment
A flagged error is a routing event, not a disciplinary action. Compass puts it in front of the right person to review and decide. No automated consequence fires. The human always decides what happens next.
Not AI guessing from behavior
There is no behavioral model, no sentiment inference, no productivity score derived from how someone works. Every flag traces back to a specific structured signal that crossed a defined threshold — transparent and auditable.
Not a black box
Every error flag shows the signal it came from, the threshold it crossed, and the rule that triggered it. Employees and managers can see why something was flagged — no mystery, no opaque score.
Decision-support
Every flagged error is a prompt for a human decision. Compass provides the signal, the category, and the severity. What happens next is always the owner's call.
Fully auditable
Every detection, routing action, and correction is written to the append-only audit log. You can trace any flag from detection to resolution — who saw it, when, and what was done.
Who it helps

Closed loops across every role

Error detection works differently depending on where you sit in the org. Each role sees what is relevant to them — nothing more.

Employee
Know before it escalates
Employees see errors routed to them directly — a data gap they need to fill, a check-in they missed, a task state to update. Fixing it promptly closes the loop before it becomes a manager conversation. No surprises at review time.
Manager
See what needs your attention
Managers receive team-level errors that require their decision — a pattern of missed completions, an attendance flag that needs context. The queue is scoped to their team. They are not flooded with noise from the whole org.
HR & Admin
Org-wide correction visibility
HR and admins see the full correction queue — correction rate trends, time-to-correct by category, which error types are recurring. Enough to spot systemic issues and adjust processes, without needing to intervene in every individual flag.
Close the loop before it compounds

Detect. Route. Correct.

Error Detection & Correction is live in Compass. Connect your operational data sources and Compass starts flagging and routing from day one — with full auditability and human-in-the-loop correction.

Compass