> ## Content Index
> Fetch the complete content index at: https://futureinuse.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# AI agents on WhatsApp: what the published Agibank and Riachuelo cases describe
- URL: https://futureinuse.ai/agibank-riachuelo-agentes-whatsapp/
- Published: 2026-10-14T12:22:00.000Z
- Updated: 2026-10-19T19:50:34.000Z
- Description: A Brazilian digital bank and a Brazilian fashion retailer feature in materials Salesforce and Genesys published in September. Both are cases released by the vendors themselves, with numbers that have not been independently audited.
- Author: Rafael Cordeiro
- Tags: Agents at Work

In September 2026, two software vendors published cases of Brazilian companies using AI agents on WhatsApp. On September 4, Salesforce released a customer story about Agibank. On September 30, Genesys put out a press release about Riachuelo. Everything below comes from those materials, written by the technology vendors.

## Agibank puts one agent in charge of routing the conversation

According to Salesforce, Agibank already ran specialized agents on WhatsApp, one of its main service channels, but changing the subject meant starting a new conversation. The solution described is a "superagent" that receives every message in Portuguese, reads the customer's intent and decides which of six agents should take over, covering loan status, FAQs, lead qualification or SDR, retention, satisfaction surveys and self-service banking.

In the vendor's example, a customer checks the status of a loan application, asks about a digital account and requests loan portability, all in the same conversation. Before routing, the superagent verifies the customer's identity with their document number. When a request could go more than one way, it shows a numbered list of options. Conversations involving compliance or fraud go to a human agent whenever the bank's rules require it.

Salesforce says six agents and 11 bots are being migrated into this single architecture and that 70,000 requests a month are handled autonomously by specialized agents. The material gives no resolution rate, no satisfaction score and no comparison with the prior period.

## Riachuelo moves from chatbot to an agent that opens tickets

In the Genesys release, Riachuelo uses a virtual agent to run its WhatsApp Business channel. Customers sort out issues such as late deliveries on their own and can reach a human agent when they need one. When a problem needs more attention, the agent automatically creates a support ticket and hands off the interaction with its context intact.

Genesys attributes three numbers to the project, with 84% of customers who use self-service with the agent resolving their request without help from an employee, against 30% with the previous traditional chatbots, customer satisfaction up 60% and technology costs down 26%. The release gives no period, no volume and no method for measuring satisfaction.

## What the two cases have in common

Read side by side, the two materials describe three similar choices. The first is the channel, WhatsApp, where the customer already is. The second is the link to internal systems, whether to look up accounts and products at Agibank or to open back-office tickets at Riachuelo. The third is a way out to a human, built into both designs.

For anyone evaluating similar projects, the gaps these materials leave open make a good checklist.

- **Period and volume** over which the numbers were measured.
- **Definition of "resolved"**, and whether the customer came back later for the same reason.
- **Handoff rate** to humans and the time until a person takes over.
- **Satisfaction method**, whenever satisfaction is cited.
- **Total cost**, beyond technology cost.

Vendor cases show what can be built and which pieces were used. When it is time to decide on a project, the numbers from your own operation are still the benchmark.

---

## Sources

- Salesforce, Customer Stories, "Agibank connects customers to the right AI agent with Agentforce", September 4, 2026.
- Genesys, Newsroom, "Riachuelo Advances Autonomous Customer Experience with Genesys Cloud Agentic Virtual Agent", September 30, 2026.

*Future in Use is an editorial initiative linked to OperOS AI.*