Singular Bank implemented ChatGPT and Codex: bankers save up to 90 minutes a day

Singular Bank from Madrid launched an internal assistant Singularity based on ChatGPT and Codex to reduce routine work for private bankers. According to the bank, the system saves each specialist from 60 to 90 minutes a day and allows preparing for a client meeting in less than a minute.

From Preparation to Conversation

Previously, analyzing a client portfolio in private banking required manual assembly of information from multiple systems: it was necessary to export positions, verify data, understand imbalances, and compile conclusions before the meeting. This cycle repeated for almost every client and consumed a significant portion of the working day.

Singularity changes the workflow: a banker immediately sees key signals from the portfolio, can quickly assess concentration, excess exposure, or imbalance, and move to discussing solutions rather than reciting numbers. The system doesn't just display data, but also suggests next steps: reduce concentration in individual positions, lock in profits, or rebalance the portfolio into a more stable structure. After the meeting, the assistant helps compose a personalized follow-up message and determine further actions. Since the tool is integrated into core banking systems, each result is saved in structured form, which means the bank can more easily track communication history and prepare reports.

"Previously, I prepared extensively for every meeting.

Now I can analyze the portfolio in real time and focus on the conversation with the client."

Where Time Is Saved

The most noticeable effect from OpenAI and Singular Bank appears in repetitive tasks that previously took minutes or tens of minutes, but now fit into seconds. The bank claims that it is precisely on these everyday actions that the cumulative savings of up to an hour and a half per day per banker adds up. This is not a one-time demonstration, but a workflow that employees use constantly.

• Meeting preparation: approximately 20 minutes before, now less than 1 minute
• Call reports: 15–20 minutes before, now less than 30 seconds
• Investment arguments: 10–15 minutes before, now around 20 seconds
• Client messages: 5–10 minutes before, now less than 30 seconds

In 30 days, bank employees performed over 3,500 operations through Singularity—an average of about 120 per day across 19 different scenarios. This is an important detail: the tool didn't remain an experiment for presentations, but became embedded in the team's everyday workflow. Additionally, the bank notes that texts and recommendations are generated from approved data sources and through structured processes, so speed is not achieved at the expense of compliance and regulatory discipline.

Why This Matters

For private banking, speed itself is not the main metric. Clients pay not for a manager quickly opening a spreadsheet, but for the quality of advice, understanding of risk, and confidence in the next step. Therefore, in Singular Bank's case, what matters more than automation for automation's sake is the transfer of time from back-office routine to the actual conversation with the client.

When a banker comes to a meeting already with a complete picture and action options, the consultation becomes more substantive. The bank's head, Javier Marín, formulates this as a transition from the operational part to applied intelligence: the banker doesn't disappear from the process but gains a more powerful tool for judgment. This aligns well with the current trend in corporate AI: companies increasingly deploy not a universal chatbot for everything at once, but internal assistants for specific contexts, connected to working systems, with clear scenarios and traceability requirements for every answer.

What This Means

Singular Bank's story shows how ChatGPT and Codex enter not only development or support, but also conservative processes like private banking. When AI is connected to verified data and embedded in regulations, it begins to save not seconds in demos, but hours in actual operational work.