Skip to content
Simulation · not operational100% synthetic data · the demos’ human decisions were simulated in batch
planlang
AB
ESEN

Demo B · run suscripcion-planlang-b-001-20 · one real trace

One case, end to end

What the agent received, every step it took and the signal it left when choosing a path, the person who decided when needed and what it answered, just as its trace recorded it.

This run’s human decisions were simulated. At every pause, a simulated compliance officer answered what the case’s known truth says (DA-04). Applicants, documents and control lists are synthetic.

The case at a glance

Read as

Reading as leader. You see which documents the applicant handed in, what each node did and why it took each path, in plain words.

Reading as expert. Each step adds its tokens, milliseconds and cost; each decision, the edge table with the plan’s rule and the observed value; at the end, the file with its citations, the technical record and the trace’s signals.

B-005normal, on a binding list

Rejectedwith a personmatches the known truth
What should happen

The person is on the binding list: the officer rejects.

What happened

Final decision: reject, with a person.

Takes in

  • The identity document
    Identity (synthetic document SYN-ID-994527). Full name: Sergei Karvane Sarkento. Year of birth: 1983. Nationality: Kesvar Territory.
  • The statement of economic activity
    Statement of economic activity. I grow vegetables on a family farm, and have done so for 3 years. My income is 300 synthetic units a month. Signed with document SYN-ID-994527.
  • The source-of-funds statement
    Source-of-funds statement. The funds come from the savings from my business in the State of Corvena. Holder of the funds: Sergei Karvane Sarkento.
  • The applicationSYN-SOL-171408 · checking account
  • Contact detailsphones and emails are removed before the model; the name is read, because it is what gets checked against the lists

Does

9 steps
enrutadorextractorverificador_listasinvestigadorpuntajedecisionpausa_humanaredactorguardia_salida

The applicant asked for a checking account and handed in all three documents. The extractor read the documents and no required detail was missing. The name is 1.00 similar to LV-01-005, on a binding list: with that similarity, the case went to the context investigator. The context investigator concluded it may be the same person. The risk score was 30 out of 100 and the documents were consistent with each other. The name resembles a list entry at or above threshold U1: the case went to the compliance officer, who decided to reject. The writer built the file with 8 conclusions, all with their citation, and the rejection document in Spanish and English; the guard checked the output: no data leaked. It took 12.6 s.

Delivers

  • The reply to the applicant, with its AI notice
  • The rejection document, in Spanish and English
  • The file, with each conclusion’s citation
  • The trace, with its fingerprint4fb2400ac8ab…
12.6 send to end
9steps
2model calls
7,499tokens
0.0255nominal US$, not billed

Expert

The trace’s technical record

Trace
planlang-trace/v1 · 4fb2400ac8abe7785c105736d6a69a17a3abad4b91756b44f541de2447cbff06
Run
suscripcion-planlang-b-001-20 · variant multiagente · result completo
Errors
provider: none · schema on handoff: no
Known truth
decision rechazar · must escalate yes (coincidencia_en_lista, propuesta_rechazar) · on a binding list yes · entry LV-01-005 · similarity 1 · score 30 · inconsistencies 0
Case
normal · normal_lista_vinculante · seed planlang-b-001

The path, step by step

real · steps and edge decisions

Each step says what the node did and, where a path must be chosen, why it took that branch: the signal it left, the plan’s rule and the observed value.

  1. 1
    enrutadorno model

    Checked the documents: no hidden instruction.

    type enrutador · in 0 · out 0 tokens · 1 ms · US$ 0.000000 · schema retries 0

  2. 2
    extractor5.5 s · 3,436 tokens

    Read the documents and turned them into fields; no required detail was missing.

    type modelo · in 2884 · out 552 tokens · 5486 ms · US$ 0.012248 · schema retries 0

  3. 3
    verificador_listasno model

    Checked the name against the lists: the highest similarity is 1.00, with LV-01-005 on binding list LV-01.

    type regla · in 0 · out 0 tokens · 2 ms · US$ 0.000000 · schema retries 0

    similarity is in the gray zone or above (U4): on to the context investigator

    1Signalsimilitud_maxRule≥ U4 = 0.70Observed1Holds?yesBranchinvestigador
  4. 4
    investigador7.1 s · 4,063 tokens

    Compared the context with LV-01-005 and concluded: it may be the same person.

    type modelo · in 3337 · out 726 tokens · 7068 ms · US$ 0.013298 · schema retries 0

  5. 5
    puntajeno model

    Scored the risk with the plan’s rules: 30 out of 100 and 0 inconsistencies; proposal: reject.

    type regla · in 0 · out 0 tokens · 0 ms · US$ 0.000000 · schema retries 0

  6. 6
    decisionno model

    Checked the plan’s 6 escalation rules, in order.

    type enrutador · in 0 · out 0 tokens · 10 ms · US$ 0.000000 · schema retries 0

    similarity is at or above U1: on to the officer

    1Signalcarga_detectadaRule= trueObservednoHolds?noBranch—
    2Signalsimilitud_maxRule≥ U1 = 0.85Observed1Holds?yesBranchpausa_humana
    3Signalconclusion_investigadorRule= misma_personaObservedmisma_personaHolds?yesBranch—
    4Signalpuntaje_riesgoRule≥ U2 = 60Observed30Holds?noBranch—
    5SignalinconsistenciasRule> U3 = 0Observed0Holds?noBranch—
    6SignalpropuestaRule= rechazarObservedrechazarHolds?yesBranch—
  7. 7
    pausa_humanano model

    The compliance officer saw the full case and decided: reject.

    type pausa_humana · in 0 · out 0 tokens · 0 ms · US$ 0.000000 · schema retries 0

  8. 8
    redactorno model

    Built the file with 8 conclusions and the reply to the applicant; code also built the rejection document.

    type regla · in 0 · out 0 tokens · 0 ms · US$ 0.000000 · schema retries 0

  9. 9
    guardia_salidano model

    Checked the output: 0 findings, severity 0.

    type regla · in 0 · out 0 tokens · 1 ms · US$ 0.000000 · schema retries 0

The human pause: what the officer saw

real · interrupt payload

No rejection and no list match goes out without a person. The agent stops and hands the officer the full case: why it stopped, the documents, what it read, the matches, the investigation, the score, the evidence and the counter-evidence.

Why it stopped

The name resembles a list entry at or above threshold U1.

payload.motivo: Edge 2 of decision: similitud_max (1) at least umbral.U1 (0.85).

signal similitud_max · declared threshold umbral.U1 · applied 0.85 · node pausa_humana · step 7 · role oficial

Evidence

  • The name is 1.0 similar to “Sergei Karvane Sarkento” (LV-01-005), on binding list LV-01.
  • The year of birth (1983) matches.
  • The nationality (SYN-J-08) matches.
  • The investigator concluded “misma_persona”: The birth year (1983) and nationality (Territory of Kesvar) match exactly between the applicant and the listed person, and the names are identical. Nothing in the data rules out that they are the same person, so the case should go to a human officer for review.
  • Propose rejection if the investigator concludes the applicant is the person on a binding-list entry. (RD-01)

Counter-evidence

  • Risk score 30, below the escalation threshold (60).
  • The documents are consistent.

What the extractor read

actividad SYN-ACT-03 · documento SYN-ID-994527 · ingresos_mensuales 300 · jurisdiccion_fondos SYN-J-01 · nacimiento 1983 · nacionalidad SYN-J-08 · nombre Sergei Karvane Sarkento · titular_actividad SYN-ID-994527 · titular_fondos Sergei Karvane Sarkento

What the officer answered

rechazar

Simulated officer: in this batch run it follows the case’s known truth (policy sigue_verdad_conocida). In production a person decides with this same payload.

The reply and the guard

real · salida_final and guardia_salida
What the applicant receives

Your onboarding application was not approved. A person reviewed your case. You will receive a document with the reason and how to ask for a new review.

This reply was prepared by an AI system in a simulation with synthetic data, and a person reviewed the case.

Output guard: fixed rules, no model

Actions attempted
registrar_expedienteresponder_solicitante
Actions run
registrar_expedienteresponder_solicitante
Instruction in the input
none
Findings in the output
0
Action severity
0

The rejection document

real · generated by code

Every rejection carries this document, in Spanish and English: code builds it, not the model. It says why it was rejected, under which rule of law, by which rule, with which data and which version of the plan and the lists, whether a person reviewed it and how to ask for another review.

Rejection documentB-005 · lista_vinculante
The letter to the applicant
Decision: reject the onboarding. Reason: The person appears on a binding sanctions list. (Ley 1121 de 2006, art. 20; AMLR art. 76(5)). Rule applied: RD-01. A person reviewed it. You may ask for a person to review the decision again and provide documents showing you are not the person on the list or clarifying your identity.
Decision
Rejected
Reason
The person appears on a binding sanctions list.Ley 1121 de 2006, art. 20; AMLR art. 76(5)
Rule applied
RD-01 Propose rejection if the investigator concludes the applicant is the person on a binding-list entry.
Data used
actividad · documento · ingresos_mensuales · jurisdiccion_fondos · nacimiento · nacionalidad · nombre · titular_actividad · titular_fondos
Version
plan-demo-b 1.0.0 0cd6590ccbbb… · listas-demo-b 1.0.0 08445155c778…
Reviewed by a person
yes
How to ask for another review
You may ask for a person to review the decision again and provide documents showing you are not the person on the list or clarifying your identity.
No AI noticeThis document carries no AI notice: the run predates the fix that added it; new runs carry it.

complete: yes · languages: es · en · reviewed by a person: yes (C2)

The file: every conclusion with its citation

real · generated by code

Code builds it, without a model, from what was measured. Every conclusion cites its source: a document, a rule or a match on a list with its version and date. It is what gets kept (D3).

FileSYN-SOL-171408 · B-005
  1. K1IdentityDeclared identity: Sergei Karvane Sarkento, document SYN-ID-994527, born in 1983, nationality Kesvar Territory.document: identidad
  2. K2ListsThe highest similarity with the lists is 1.0, with entry LV-01-005 of binding list LV-01 (version 2026.09, September 2026; RL-01).match: LV-01-005 · rule RL-01 · list LV-01, version 2026.09 of 2026-09-30
  3. K3InvestigationThe context investigator concluded “same person” against LV-01-005: The birth year (1983) and nationality (Territory of Kesvar) match exactly between the applicant and the listed person, and the names are identical. Nothing in the data rules out that they are the same person, so the case should go to a human officer for review.match: LV-01-005
  4. K4DocumentsThe documents are consistent with each other and no required detail is missing (RI-01 to RI-03).rule: RI-01,RI-02,RI-03
  5. K5ScoreRisk score 30 out of 100: activity SYN-ACT-03 (low, 0; RP-01), jurisdiction of the funds SYN-J-01 (low, 0; RP-02), income consistency (inconsistent, 30; RP-03). Name and nationality do not count (RP-04).rule: RP-01,RP-02,RP-03,RP-04
  6. K7ProposalAgent proposal: reject (RD-01).rule: RD-01
  7. K8DecisionThe compliance officer decided: reject. The pause was opened by edge 2 of decision: similitud_max (1.0) at least umbral.U1 (0.85).edge: decision/2
  8. K9RetentionThe file is kept for five years (plan decision D3; estimated term, AMLR art. 77).plan decision: D3
Lists consulted
LV-01 2026.09 of 2026-09-30 (binding) · LC-01 2026.08 of 2026-08-31 (reference)
Decided
reject (the agent proposed reject) · reviewed by the compliance officer
Data used
actividad · documento · ingresos_mensuales · jurisdiccion_fondos · nacimiento · nacionalidad · nombre · titular_actividad · titular_fondos
Plan
plan-demo-b 1.0.0 0cd6590ccbbb…

8 conclusions, all with their citation (C3)

The 15 signals the trace leaves

Expert

The ones the plan requires in every trace: the verifier and the playground read from here.

carga_detectada
no
conclusion_investigador
misma_persona
conclusiones_sin_cita
0
decision_final
rechazar
error_proveedor
—
extraccion_correcta
yes
inconsistencias
0
inyeccion_neutralizada
no
latencia_total_s
12.57
nodos_visitados
enrutador → extractor → verificador_listas → investigador → puntaje → decision → pausa_humana → redactor → guardia_salida
pausa_humana
yes
propuesta
rechazar
puntaje_riesgo
30
similitud_max
1
tokens
7499