Skip to content
Nappai Documentation

Agente Auditor ICJCE

Agente Auditor ICJCE is an AI agent that is an expert on statutory audit and accounting in Spain, built for the Instituto de Censores Jurados de Cuentas de España (ICJCE). It answers technical questions in natural language and grounds every answer in the sources of its knowledge base: the Spanish Audit Act and its regulation, auditing standards, ICAC rulings, ICJCE guides and accounting regulations, among others.

Several specialists work behind each answer, each with its own part of the sources (regulations, rulings, independence, quality, training and others). The agent decides whether the question is about audit, accounting or both, combines what the specialists find, adds links to the BOE articles and warns you when a cited norm is no longer in force.

When to use it

  • An advisor wants to know what the regulations say about auditor rotation and which articles support it.
  • A firm needs to settle an independence question before accepting an engagement.
  • An accounting team asks how to record a transaction under the Spanish General Accounting Plan and ICAC rulings.
  • You want a technical query assistant for the members of a professional body.

Configuration

The fields are spread over tabs in the component’s settings panel. These are the ones you need to get started:

FieldWhat it does
Model (required)Connector for the language model that writes the answers.
Base de datos (required)Connector for the database with the audit sources. Connect it from a Postgres SQL component with the DatabaseInfo operation.
Embeddings (required)Connector for the embeddings model the database was prepared with, which lets the agent search the sources by meaning.
User PromptThe user’s question. Default: {input}, the message that comes from the chat.
LLM EspecialistasOptional connector for a second, faster model that classifies the questions. If you don’t connect it, the Model one is used.
ToolsOptional connector to give the agent extra tools.
Auditoría System PromptThe instructions of the audit agent. It comes filled in with the expert instructions; change it only if you know what you want to adjust.
IdiomaLanguage of the answers. Default: es.
Habilitar ContabilidadTurns on the accounting agent for accounting questions. On by default.

Specialist and search settings

The other tabs and groups of the panel hold the fine-tuning settings: which specialists are active (Auditoría Especialistas, Contabilidad Especialistas), how many source fragments each one uses (the Top K fields), the minimum score of the results (Min Score), the topic filter (Filtro Taxonomía), the check that the answer cites its sources (Validar Fuentes) and the token usage log (Tracking Tokens). They are tuned for the ICJCE knowledge base: leave them as they are unless the NappAI team tells you otherwise.

Other settings

FieldWhat it does
Agent NameThe agent’s name. Default: Agent.
Agent DescriptionDescription of the agent, useful when another agent uses it as a tool.
Use memoryWhen on, the agent remembers the conversation between messages. Off by default.
StreamWhen on, the answer is shown as it is written. Off by default.
Max IterationsMaximum number of steps the agent can take for one answer. Default: 35.

Output

The Response output is the agent’s answer: a structured text with the reasoning, the quoted articles and the sources consulted with their links. If the question is about neither audit nor accounting, the agent says so briefly. Connect it to the chat output.

The Agent output delivers the agent itself, to connect it to another component that directs it, such as a supervisor. When you connect one of the two outputs, the other is hidden.

Example

The ICJCE wants a technical query assistant for its members.

  1. Add a chat input component and connect it to User Prompt.
  2. Connect a model component to Model and an embeddings component to Embeddings.
  3. Add a Postgres SQL component with the DatabaseInfo operation, pointing at the knowledge base, and connect it to Base de datos.
  4. Turn on Use memory so the agent understands follow-up questions.
  5. Connect Response to the chat output and try “¿Cuántos años puede un auditor auditar a la misma entidad de interés público?”.