← Cases
South KoreaFinance and insurance2020 – ongoing

Shinhan Bank

AI voice agent Solly in the call centre

Shinhan Bank's voice AI took over 95% of outbound notice calls and resolved about a quarter of all calls without an agent

WorkingCompany's own claim

Problem

Shinhan Bank's call centre handled 40,000 to 80,000 calls a day on average. Much of the outbound work, such as maturity notices and follow-up calls after product sales, was routine, and customers waited to reach agents at peak times.

Approach

From May 2020 the bank used Solly, a voice agent built on Naver's CLOVA engine. A first phase moved outbound notification calls to the AI; a second phase, built over about a year by a consortium of Shinhan, Naver CLOVA and the contact-centre firm Bridgetec, extended it to inbound calls with 352 scenarios. During the 2021 year-end tax season Solly handled more than 17,000 of about 28,000 calls in roughly ten days.

Results

95%share of outbound notification calls handled by the AICompany's own claim[1][2][3]
~25%share of calls resolved without transfer to an agent after the inbound phaseCompany's own claim[2]
27s → 6scustomer waiting time reported after the second phaseCompany's own claim[2]

Takeaway

The bank began with scripted outbound notices where AI led the conversation, then added inbound questions, where about half of calls involved AI but only about 25% were resolved without an agent against a 40% target. Sources reported different wait-time gains (27 to 6 seconds and 2–3 minutes to 30–40 seconds), without explaining measurement methods.

Categories

Business function
Customer service
Innovation type
Process innovation
Evidence
Company's own claim

Sources

  1. 은행 상담에 인공지능이 접목되다
    Byline Network23 Mar 2021News article
  2. 신한은행, 콜센터 25% AI가 처리한다
    Byline Network8 Sep 2021News article

Source summaries

[1]Byline Network23 Mar 2021

Describes Solly, in service since May 2020, which handles 95% of the bank's outbound calls such as maturity notices and follow-up calls and can talk with about 100 customers at once. During the 2021 year-end tax season it handled more than 17,000 of some 28,000 calls in about ten days. A second phase was to teach it account opening, loans and foreign exchange.

[2]Byline Network8 Sep 2021

Reports the completion of the second phase, in which a consortium of Shinhan, Naver CLOVA and Bridgetec applied 352 scenarios to inbound calls. Of 40,000–80,000 daily calls, the AI handles about half and about 25% end without an agent; waiting time fell from 27 to 6 seconds. The bank aimed to raise the resolution rate to 40%.

[3]Kyunghyang Shinmun7 Feb 2021

The Kyunghyang Shinmun reports on AI use at Korean banks. It says Shinhan introduced the voice bot with Naver in May 2020, that waiting time to reach an agent fell from 2–3 minutes to 30–40 seconds and that the bot handles 95% of outbound work. It also mentions KB Kookmin's use of AI in branch staff postings and NH NongHyup's AI investment-profile analysis.

Full details are in the original sources.

Added 25 Oct 2026 · Updated 25 Oct 2026
Open case page ↗