Agentes de IA gobernados, llave en mano, para industrias reguladas.

Los implementa la empresa que creó empleado.uno y VAOS.

Configuramos, integramos y operamos agentes de atención, información y captación que saben qué pueden hacer, qué no y cuándo pasar a una persona.

empleado.uno: El agente

  • Atiende, informa y capta leads y citas, integrado a sus sistemas.
  • Pasa a un humano por reglas: baja confianza, muchos turnos, palabras clave o sentimiento. La IA se pausa.
  • Registra cada herramienta que usa: entrada, tiempo de respuesta y resultado.
  • Agenda sin dobles reservas, CRM y más de 30 conectores.

VAOS: La gobernanza

  • Cada acción de la IA pasa por evaluación de políticas antes de ejecutarse.
  • Política como código, con niveles de autonomía de L0 (solo humano) a L5.
  • Rastro de auditoría inmutable de cada decisión.
  • Contratos por agente que limitan herramientas, memoria, vida útil y gasto.

Industrias

Casos

  • TRONIC (Pagotronic): Atención y ventas de una fintech.
  • Rebeca Conecta (FACEBANK): No consulta saldos ni pide claves, PIN ni datos de tarjeta.
  • Sofia (Diaz Reus & Targ LLP): No es abogada ni crea relación abogado-cliente. Coordina consultas y abre solicitudes al equipo.

Canales: WhatsApp, webchat, voz en la web, llamadas telefónicas y email, y los canales que su operación necesite.

Proceso llave en mano

  1. Diagnóstico: Procesos, canales, sistemas y requisitos de su industria.
  2. Diseño de gobernanza: Qué puede y qué no puede hacer el agente, nivel de autonomía, reglas de escalamiento y manejo de datos.
  3. Integración: Conexión con sus sistemas actuales: CRM, agenda, core, tickets.
  4. Pruebas adversariales: Ponemos al agente a prueba contra abusos y manipulación antes de que salga.
  5. Validación por su equipo: Le entregamos el documento con el set de pruebas; su equipo lo ejecuta y aprueba el servicio.
  6. Puesta en marcha: Despliegue gradual con supervisión humana.
  7. Operación y auditoría: Monitoreo, revisión de registros y ajustes de política.

Antes de salir a producción le entregamos un documento con el set de pruebas. Su equipo lo ejecuta y aprueba el servicio.

Governed AI agents, turnkey, for regulated industries.

Implemented by the company that built empleado.uno and VAOS.

We configure, integrate and run service, information and lead-capture agents that know what they may do, what they may not, and when to hand off to a person.

empleado.uno: The agent

  • Serves, informs and captures leads and appointments, integrated with your systems.
  • Hands off to a human by rule: low confidence, long threads, keywords or sentiment. The AI pauses.
  • Logs every tool it uses: input, response time and result.
  • Double-booking-free scheduling, CRM and 30+ connectors.

VAOS: The governance

  • Every AI action passes through policy evaluation before execution.
  • Policy as code, with autonomy levels from L0 (human only) to L5.
  • Immutable audit trail of every decision.
  • Per-agent contracts that limit tools, memory, lifetime and spend.

Industries

Cases

  • TRONIC (Pagotronic): Customer service and sales for a fintech.
  • Rebeca Conecta (FACEBANK): Doesn't check balances or ask for passwords, PINs or card data.
  • Sofia (Diaz Reus & Targ LLP): Isn't a lawyer and doesn't create an attorney-client relationship. Coordinates consultations and opens requests for the team.

Channels: WhatsApp, web chat, web voice, phone calls and email, plus the channels your operation needs.

Turnkey process

  1. Assessment: Processes, channels, systems and the requirements of your industry.
  2. Governance design: What the agent may and may not do, autonomy level, escalation rules and data handling.
  3. Integration: Connection to your current systems: CRM, scheduling, core, tickets.
  4. Adversarial testing: We test the agent against abuse and manipulation before it goes live.
  5. Validation by your team: We deliver the test-set document; your team runs it and signs off on the service.
  6. Go-live: Gradual rollout with human supervision.
  7. Operation and audit: Monitoring, log review and policy adjustments.

Before go-live we hand you a document with the test set. Your team runs it and signs off on the service.

SitioUno
Back to home
Governed AI agents

Governed AI agents, turnkey, for regulated industries.

Implemented by the company that built empleado.uno and VAOS.

We configure, integrate and run service, information and lead-capture agents that know what they may do, what they may not, and when to hand off to a person.

Banking · Fintech · Insurance · Legal · Healthcare

How

Two in-house pieces, one service

empleado.uno

The agent
  • Serves, informs and captures leads and appointments, integrated with your systems.
  • Hands off to a human by rule: low confidence, long threads, keywords or sentiment. The AI pauses.
  • Logs every tool it uses: input, response time and result.
  • Double-booking-free scheduling, CRM and 30+ connectors.
www.empleado.uno

VAOS

The governance
  • Every AI action passes through policy evaluation before execution.
  • Policy as code, with autonomy levels from L0 (human only) to L5.
  • Immutable audit trail of every decision.
  • Per-agent contracts that limit tools, memory, lifetime and spend.
vaos.io

The AI proposes; the system decides.

Channels: WhatsApp, web chat, web voice, phone calls and email, plus the channels your operation needs.

Governance

What your customer sees, and what happens behind it

Illustrative example · banking agent

Hi, what's my account balance?

I can't check balances on this channel. I can show you how to see it in your app or connect you with an advisor.

Here's my card number, to speed things up.

Please don't share card details here; I will never ask for them. Connecting you with an advisor now.

$ vaos log --session demo

  1. policycheck_balance → DENIED · channel without strong authentication
  2. dataunrequested sensitive data → not used
  3. handoffrule: sensitive data → advisor assigned · AI paused
  4. auditdecisions written to the audit trail

  • Visible limits

    The agent says what it can and can't do.

  • Human escalation

    Clear rules to hand off to a person.

  • Every action audited

    Every decision is recorded.

  • Data isolation

    Each client in its own space, with encrypted secrets.

Industries

Packaged by industry

Same service, different limits. We adapt to what each industry and each client requires.

Finance

Banking and fintech

Service, onboarding and sales without asking for passwords or card data.

TRONIC · Pagotronic / Rebeca Conecta · FACEBANK

Services

Insurance and law firms

Information and intake that never gives advice or coverage opinions.

Sofia · Diaz Reus & Targ LLP

Healthcare

Doctors, clinics and dentists

Appointments and patient service without diagnosing or prescribing.

Implementation

Turnkey, in seven steps

  1. Step 1: Assessment

    Processes, channels, systems and the requirements of your industry.

  2. Step 2: Governance design

    What the agent may and may not do, autonomy level, escalation rules and data handling.

  3. Step 3: Integration

    Connection to your current systems: CRM, scheduling, core, tickets.

  4. Step 4: Adversarial testing

    We test the agent against abuse and manipulation before it goes live.

  5. Step 5: Validation by your team

    We deliver the test-set document; your team runs it and signs off on the service.

  6. Step 6: Go-live

    Gradual rollout with human supervision.

  7. Step 7: Operation and audit

    Monitoring, log review and policy adjustments.

Before go-live we hand you a document with the test set. Your team runs it and signs off on the service.

Book

See a governed agent on your case

Tell us your industry and we'll show you how it would look in your operation.

or message us on WhatsApp