NSK AI
Products
·
Data Annotation

Labelled data you can defend in a review.

Human annotation for systems that have to be right. Domain specialists work to a written spec, disputed items are adjudicated rather than averaged, and every batch ships with its agreement rate and its edge cases named.

Pass 00 — raw

Unlabelled

A batch arrives as items with no structure: transactions, calls, notes, photographs. Nothing here is wrong yet, and nothing here can train anything.

Pass 01–03 — independent

Three passes

Every item is labelled three times, by specialists who cannot see each other’s work. A coarse shape appears. Most of the batch settles immediately.

Contested

Disagreement

The items that split the passes are not noise. They sit on the boundary between classes, which is precisely where a model in production gets things wrong.

Ruled

Adjudicated

A named specialist rules on each contested item and the spec is amended underneath it. The boundary stops moving, and it stays where it is for the rest of the batch.

κ 0.00 · block 32 px · contested —

Specialists, screened on your task

Annotators qualify on your data with your guidelines, and we publish who labelled what.

Adjudicated, not averaged

Multi-pass review with a named adjudicator on disagreements, so the label has a reason behind it.

Reported, batch by batch

Agreement rate, throughput and the edge cases that forced a spec change, in writing.

What comes with it

Text, speech, documents and images
Guidelines written with your domain experts
Low-resource and West African languages
Labelling inside your environment where required
Adjudication trace

Three passes, one ruling, and the reason in writing.

Where annotators disagree, most vendors take the majority and move on. We open the item, rule on it, and record why. Every ruling is attributable, and the ones that expose a gap amend the spec for the rest of the batch.

item 0413 · batch NB-24
12,480 items · 3 passes
TRF/REV NIP 08:14 — debit posted, credit not received, value date T+2
ANNOTATOR A-07
Failed
Credit never landed, so the transfer did not complete.
ANNOTATOR A-11
Reversed
The reversal flag is present on the record.
ANNOTATOR A-02
Pending
Value date has not passed; the ledger can still settle.
1 · 1 · 1 — no majority, sent to adjudication
Ruling — adjudicator O. Bassey, payments
Unresolved

A reversal flag records an instruction, not an outcome. Until the ledger settles at value date the customer is still out of funds, and a model trained on “reversed” will close the ticket early.

Spec amended §4.2 — settlement state, not the flag, decides the label. 214 earlier items re-labelled under the amendment.
Where labels collide

One agreement number hides the class that will break you.

We report agreement per class, and the pairs that annotators confuse most. Those pairs are where a model fails in production, and they are the pairs your reviewers should read first.

0.91
Cohen’s κ, batch mean
0.62
κ on the weakest class
hover a cell
Failed
Reversed
Pending
Duplicate
Settled
Disputed
Failed
14
9
2
0
3
Reversed
14
21
1
4
2
Pending
9
21
0
11
1
Duplicate
2
1
0
0
17
Settled
0
4
11
0
1
Disputed
3
2
1
17
1
Disagreements per label pair, batch NB-24. Darker means the pair collides more often.

What ships with every batch

Labels are the smallest part of the delivery. The rest is the evidence that lets your model risk team sign them off.

In the handover
Batch
NB-24
Items
12,480
Adjudicated
431
Closed
14 Aug 2026
01
Agreement report

Cohen’s κ per class and per annotator pair, the confusion pairs ranked by volume, and drift against the previous batch. The weakest class is named rather than averaged away.

agreement.csv
agreement.pdf
02
Adjudication log

Every overturned label with the original three passes kept intact, the named adjudicator, and the reasoning. Re-labelled items stay flagged in place.

rulings.jsonl
03
Edge-case register

The items that forced a spec amendment, dated, with worked examples. Held back from training and handed over as an evaluation set.

edge-cases/
spec-history.md
04
Chain of custody

Who labelled what and when, the access granted to each annotator, and the environment the data never left. Signed off before the batch closes.

custody.pdf
κ 0.91 · 431 rulings · 3 spec amendments
signed — O. Bassey, batch lead

The rest of the system

Own your own AI future.

Label the data your model will be judged on, with the evidence attached.