Use cases · Financial servicesIn development
Accounts, payments and requests, backed by sources and read-only
Ray will answer the common questions of a wallet, a fintech or a financial institution from the information each business publishes, and report a payment’s status without touching the money. Sensitive topics, such as an unrecognized transaction, will always go to a person.
Early access with close support from the team. Ray is in development.
Illustrative example of a fictional wallet: five frequent questions. Ray will resolve three, one with read-only account data; the unrecognized charge and the request for investment advice will go to a person.
Illustrative example with sample data. Lumora is a fictional brand.
Frequent questionsIn development
Accounts, payments and procedures, explained step by step
Ray will look for the answer in the information the company publishes and, with the customer’s identity signed, in their account data, read-only. It will explain conditions; it will not give financial recommendations.
Transfers and payments
Crediting hours, published limits and the status of a transaction, with the time of the data.
Receipts and statements
Where to download them in the app, with the exact name of each menu.
Common procedures
Updating contact details, activating a card or requesting a certificate: the published steps to do it from the app.
Conditions, not advice
Fees, time limits and requirements exactly as published. Ray will not recommend products, investments or loans.
Illustrative example: a customer asks how long a transfer to another bank takes and Ray replies with a list of two time windows; then they say the afternoon transfer has not arrived, Ray reports with read-only data that it was sent and will be credited today, and the customer confirms it was solved.
Illustrative example with sample data. Lumora is a fictional brand.
TraceabilityIn development
What the team sees in the console
The customer will get the written answer, with no internal references. In the console, the team will see the cited article and its version, the transaction data looked up, the customer’s identity and the controls the answer passed.
Article and version
Each answer will be tied to the published article that backs it and to the version in force at that moment.
Data looked up
Which transaction was read, read-only, with the time of the data and the state of the customer’s signed identity.
Check before sending
Hours, time limits and fees checked against the published information. If something does not match, the answer does not go out.
Illustrative example of the record in the console, visible only to the team: the article Transfers to other banks in its version 5, the transfer looked up read-only, the check of the time window, the time limit and the path, and the signed identity and non-sensitive topic controls.
Illustrative example with sample data. Lumora is a fictional brand.
Sensitive topicsIn development
In financial services, the sensitive list is the core
An unrecognized transaction, a rejected verification or a request for advice does not get an automated reply. On those topics the customer will get a fixed holding message, and the team the conversation with the reason, the context and a draft that will never be sent on its own.
Unrecognized transactions or possible fraud
Charges the customer does not recognize or suspicious logins. They will go over in full, with a supervisor alert.
Added by the companyDisputed payments
Declines, chargebacks and complaints about a charge. The draft will not promise amounts or time limits.
DefaultRejected identity verification
A person communicates the reason and the next steps.
Added by the companyRequests for recommendations
Investments, loans or taxes: Ray will not advise, and the question will go to a person on the team.
Added by the companyComplaints and regulators
Mentions of a regulator, a report or legal action.
DefaultPersonal data and account closure
Access, correction or deletion of data and closure requests.
Default
Illustrative example. The customer writes that there is a charge they do not recognize; they get only the fixed holding message. The team gets the conversation with the reason, a supervisor alert and a draft from Ray with its sources, which a person will review before sending.
Illustrative example with sample data. Lumora is a fictional brand.
The team will handle these cases in Inbox and copilot; the list will be edited in Ray agent.
How each sensitive topic will be treated
Fixed holding message
The customer will not get an AI-written text about the topic: only the fixed holding message.
No promised amounts or time limits
The draft for the team will not promise refunds, amounts or dates.
The draft never goes out on its own
A person will review, edit and send the reply, or reject it with a reason.
Supervisor alert
Possible fraud or a risk to a person will also reach a supervisor.
Read-onlyIn development
Ray reports; the company moves the money
Ray’s tools will be read-only. It will be able to report the status of a transfer or a payment, but it will not move money, reverse transactions or change account details: that will stay with people, in the company’s systems.
Without an identity signed by the company’s server, the customer will be treated as anonymous and Ray will not look up their account data.
Ray will be able to look up
- The status of a transfer or a payment, with the time of the data
- The published hours, limits and fees
- The published steps to complete a procedure in the app
Ray will never
- Move money, reverse payments or cancel transfers
- Change an account’s limits, details or holders
- Recommend investments, loans or products
- Approve a verification or unlock an account
See Ray with the institution’s questions
A short conversation with the team that builds Ray to review the company’s case: frequent questions, existing knowledge, sensitive topics and a migration plan.