# Credit case: complete files, no chasing | Shimli > The agent collects and validates documentation over chat, reads each document with OCR and hands the analyst a file ready for a decision. Credit ## Credit analysis with AI: from the incomplete file to a decision with everything in view. The agent collects and validates documentation over chat, assembles the file and hands it to the analyst ready to decide. Consumer lenders and savings cooperatives The starting point The credit decision takes minutes. What took days was gathering the paperwork to make it. What reached the analyst - ID document WhatsApp blurry photo - Proof of income Email arrived 3 days later - Utility bill — never arrived - Bank statement WhatsApp expired 4 months ago Four documents, three channels and no way to know what was missing without opening them one by one. The file wasn’t held up by judgment: it moved at the speed of paperwork. How it runs ### From the application to the file in the analyst’s hands. - 01 Application intake It arrives over WhatsApp or web and opens the case. - 02 Documents, one by one It asks one by one and stops if one is illegible. - 03 Reading and validation OCR extracts, checks validity and cross-checks. - 04 File assembly Documents, data and bureau in a single record. - 05 Handoff to the analyst The case arrives complete; failures are flagged. What it uses from Shimli ### The pieces behind AI credit analysis. #### AI agents They request, review and ask again when a document isn't legible, without anyone having to chase. - How it was used here - It requests one document at a time and won’t move on until it arrives legible. - If the photo is blurry or cropped, it explains what’s missing and asks again. - It answers requirement questions without taking an analyst’s time. Agents */* **Credit · Applications** Active CA Tools *5* - buscarCliente Database Looks up the customer by ID in use - leerDocumento Read Extracts the data and checks validity in use - validarRequisitos Logic Checks against the product requirements in use - calcularCuota Compute Computes the instalment from amount and term in use - armarExpediente Action Hands the complete file to the analyst in use Limits Can Request and validate documents Prepare the file Cannot Approve the loan Hands to a person If a document does not match #### Flow Builder Each document is a step with its own validation; the flow won't advance if one fails. - How it was used here - Four steps, one per document, each with its own validation. - The bureau isn’t queried until all four pass. - If a field doesn’t match the database, the case is flagged before it reaches the analyst. Nodes **73** Messaging *14* AI agents *9* Logic *12* Integrations *21* Portfolio loaded 420 accounts Days past due 1–30 · 31–90 · +91 Collections agent negotiates within policy Promise recorded writes to the core Collections agent *Can do* **Offer up to 6 days** **Send payment link** *Cannot* **Waive principal** *Escalates to a human* **If the balance is disputed** #### Databases The file lives in its own structure, with the fields the credit committee requires. - How it was used here - The fields are the ones the committee already used, not generic ones. - Each document is linked to the applicant and to its expiry date. - The analyst opens one record, not four scattered conversations. Databases */* **Credit applications** Edit structure CA | Name | Phone | Amount | Stage | Updated | | --- | --- | --- | --- | --- | | Lucía Torres | ···· 4471 | $2,400 | File complete | today, 10:12 | | Marco Ríos | ···· 9038 | $850 | Proof missing | today, 09:40 | | Elena Navas | ···· 2215 | $5,000 | In committee | yesterday | | Bruno Salas | ···· 7760 | $1,200 | File complete | yesterday | | Ana Medina | ···· 3184 | $3,100 | ID missing | Monday | 5 of 128 records #### Integrations & API Bureau queries and core writes, without replacing the systems already running. - How it was used here - The bureau query fires from the flow, without leaving the case. - The approved file is written to the core with the fields already validated. - The core remained the source of truth: nothing was migrated. Integrations */* **Connections** + New connection CA | System | Type | Status | Last activity | | --- | --- | --- | --- | | CRM | API REST | Connected | 2 min ago | | Core | API REST | Connected | 5 min ago | | Customer database | MCP | Connected | 1 min ago | | WhatsApp Business | Official · Meta | Connected | now | | Credit bureau | API REST | In review | | **POST** /v1/msg-wh *200* *12:04:31* **GET** /v1/contact *200* *12:04:29* **POST** /v1/crm *200* *12:04:26* The outcome ### What changes in credit operations. #### The file arrives complete The analyst opens the case and has everything: legible documents, verified data and the bureau query already run. #### Nothing moves forward half-done The flow stops at the missing document instead of dragging the gap all the way to the end of the process. #### The analyst decides, doesn't chase The team's time goes into assessing risk, which is the part that can't be delegated. This case describes how the agent operates on the process. Result figures are defined with you during discovery, against your own indicators. The other three processes ### The agent runs the same way on every one of them. [Collections From cold receivables to confirmed payment. See the case](https://www.shimliapp.com/en/casos/cobranza) [Support From the message queue to the reply that resolves. See the case](https://www.shimliapp.com/en/casos/atencion) [Operations From the WhatsApp order to the invoice issued. See the case](https://www.shimliapp.com/en/casos/operaciones) ### What people ask before getting started. Straight answers on what Shimli is, what it costs, what it integrates with and who is behind it. What is Shimli? – Shimli is a platform to build and run AI agents inside a company's operation. An AI agent is a program that understands the context of a conversation, queries the institution’s systems, decides within the policies defined for it and carries the task through to the end. It is built by Sheilim LLC, headquartered in Miami. What is the difference between a chatbot and an AI agent? + A chatbot replies from a script: it recognizes an intent and returns text written in advance. An AI agent understands context, analyzes information, executes tasks in the institution’s systems and escalates to a person when needed. A chatbot replies; an agent resolves. How much does Shimli cost? + One plan at $139 a month with every feature on the platform: there are no editions or tiers. It includes 3 human seats and 1 admin, 1 Flow Builder, WhatsApp, Facebook, Instagram and web connection, and unlimited WhatsApp templates. Added separately: an extra instance at $59 a month, an extra Flow Builder at $19, an extra human seat at $9, and AI usage at $49 per 2,450 queries. You can create the account and get into the platform yourself. ### Let's build this same case on your operation. We review your **credit** process, identify where the agent fits and define with you the indicators we're going to move. No commitment, nothing to install. Design my credit case