Demo · OutFigure

Step 1

15 answers across 4 sessions

Everything the learner did arrives as events: which question, whether it was right, and what help they had. OutFigure keeps every answer, and decides which ones count as evidence of what they can do alone.

Answers from learner_7f3a

Synthetic · times in UTC

Every answer, with the help used and whether it counts as evidence
When	Topic	Result	Help	Counts?
3 Sep, 18:02	DNS records	Right	None	Counts
3 Sep, 18:05	DNS records	Right	None	Counts
3 Sep, 18:09	DNS records	Wrong	None	Counts
5 Sep, 19:10	DNS records	Right	None	Counts
5 Sep, 19:14	DNS records	Right	None	Counts
10 Sep, 20:01	Subnetting	Right	None	Counts
10 Sep, 20:06	Subnetting	Right	Tutor hints	Set aside Hints or outside help were used
10 Sep, 20:12	Subnetting	Right	Tutor hints	Set aside Hints or outside help were used
14 Sep, 18:31	Subnetting	Right	Tutor hints	Set aside Hints or outside help were used
14 Sep, 18:36	Subnetting	Right	Outside help	Set aside Hints or outside help were used
14 Sep, 18:42	Subnetting	Wrong	Tutor hints	Set aside Hints or outside help were used
14 Sep, 18:48	TLS certificates	Wrong	None	Counts
14 Sep, 18:52	TLS certificates	Right	None	Counts
14 Sep, 18:57	Access policies	Right	Not reported	Set aside No record of whether help was used
14 Sep, 19:01	Access policies	Right	Not reported	Set aside No record of whether help was used

Step 2

What they've shown on their own

For each topic OutFigure reports only what answers given alone support. With fewer than five of them, or all in one session, it says the evidence is thin instead of guessing a score.

Subnetting
Not enough evidence

1 of 1 right alone

With help
4 of 5 right

Sessions
1

Latest solo answer
10 Sep
DNS records
Enough to summarize

4 of 5 right alone

Sessions
2

Latest solo answer
5 Sep
TLS certificates
Not enough evidence

1 of 2 right alone

Sessions
1

Latest solo answer
14 Sep
Access policies
Not enough evidence

0 answers alone

Help not reported
2 answers

Sessions
0

Latest solo answer
None

Step 3

What to check next

OutFigure ranks where one answer given alone would tell you the most, and suggests questions the learner hasn't seen. Your product decides whether to ask them; OutFigure never contacts the learner.

Check plan

Issued 15 Sep, 09:00 · rule next-check-v0

1
Access policies

No answers on their own yet. Every answer on this topic came with help or without saying whether help was used.
- iam_q03 new
- iam_q04 new
- iam_q05 new
2
Subnetting

Help since the last solo answer. Hints or outside help came after the last solo answer, so it is unclear what the learner can do alone now.
- sub_q07 new
- sub_q08 new
- sub_q01 seen before
3
TLS certificates

Too little to go on. Fewer than 5 solo answers, or all of them in one session.
- tls_q03 new
- tls_q04 new
- tls_q05 new
4
DNS records

Time to check it stuck. The topic looked solid, and enough time has passed to see whether it lasted.
- dns_q06 new
- dns_q07 new
- dns_q01 seen before

Step 4

Did it stick?

Over the next 7 days, later answers are matched to the plan. Only an answer given alone counts. Help doesn't, and neither does silence: no answer is reported as missing, not as a fail.

Follow-up

Window of 7 days · read 23 Sep

Access policies
Not observed

No answer given alone within 7 days. 1 answer arrived without saying whether help was used, so it doesn't count.
Subnetting
Observed

Wrong on a new question, answered alone 33.5 hours after the plan.
TLS certificates
Not observed

No answer given alone within 7 days.
DNS records
Observed

Right on a new question, answered alone 47 hours after the plan.

Step 5

In your stack

The same loop runs through the API: send events from your backend, issue a plan when you want to know what to ask, and read its follow-up later.

API reference

Issue a check plan

POST /v1/learners/learner_7f3a/check-plans
Authorization: Bearer lsk_test_…
Idempotency-Key: plan-7f3a-0915

{ "max_checks": 5 }

Reading the plan a week later, abridged

{
  "rule_version": "next-check-v0",
  "mode": "shadow",
  "validation_status": "unvalidated",
  "checks": [
    {
      "position": 2,
      "concept_id": "networking.subnets",
      "reason": "assisted_since_independent_evidence",
      "follow_up": {
        "status": "observed",
        "observed": {
          "item_id": "sub_q07",
          "correct": false,
          "delay_hours": 33.5,
          "item_novelty": "new_to_learner"
        }
      }
    }
  ]
}

One learner, start to finish

15 answers across 4 sessions

What they've shown on their own

What to check next

Did it stick?

In your stack

Want to see this on your own product?