❌

Normal view

There are new articles available, click to refresh the page.
Before yesterdayMain stream

쿠팑, 3370만 계정 κ°œμΈμ •λ³΄ μœ μΆœΒ·Β·Β·μΈμ¦ν‚€ 관리 λΆ€μ‹€ 의혹 제기

1 December 2025 at 03:45

이컀머슀 업체 쿠팑이 29일 λ°œν‘œν•œ μž…μž₯문에 λ”°λ₯΄λ©΄, νšŒμ‚¬λŠ” 11μ›” 18일 μ•½ 4,500개 κ³„μ •μ˜ κ°œμΈμ •λ³΄ 무단 λ…ΈμΆœμ„ ν™•μΈν•˜κ³  κ²½μ°°μ²­, ν•œκ΅­μΈν„°λ„·μ§„ν₯원, κ°œμΈμ •λ³΄λ³΄ν˜Έμœ„μ›νšŒμ— μ‹ κ³ ν–ˆλ‹€. 이후 쑰사 κ³Όμ •μ—μ„œ ν”Όν•΄ 규λͺ¨κ°€ μ•½ 3,370만 개 κ³„μ •μœΌλ‘œ ν™•λŒ€λœ κ²ƒμœΌλ‘œ 확인됐닀.

유좜된 μ •λ³΄λŠ” 이름, 이메일 μ£Όμ†Œ, 배솑지 μ£Όμ†Œλ‘(μž…λ ₯ 이름, μ „ν™”λ²ˆν˜Έ, μ£Όμ†Œ), 일뢀 주문정보닀. 쿠팑 츑은 결제 정보, μ‹ μš©μΉ΄λ“œ 번호, 둜그인 μ •λ³΄λŠ” ν¬ν•¨λ˜μ§€ μ•Šμ•˜λ‹€κ³  λ°ν˜”λ‹€. 2025λ…„ 6μ›” 24일뢀터 ν•΄μ™Έ μ„œλ²„λ₯Ό 톡해 무단 접근이 이루어진 κ²ƒμœΌλ‘œ μΆ”μ •ν•˜κ³  μžˆλ‹€. λ˜ν•œ μΉ¨ν•΄ 원인에 λŒ€ν•΄ ν˜„μž¬ κ΄€λ ¨ λ‹Ήκ΅­κ³Ό ν˜‘λ ₯ν•˜μ—¬ 쑰사λ₯Ό μ§„ν–‰ 쀑라고 μ–ΈκΈ‰ν–ˆλ‹€.

κ³Όν•™κΈ°μˆ μ •λ³΄ν†΅μ‹ λΆ€, μ„œμšΈκ²½μ°°μ²­ λ“± 관계 기관은 11μ›” 19일 침해사고 신고와 11μ›” 20일 κ°œμΈμ •λ³΄ 유좜 μ‹ κ³ λ₯Ό μ ‘μˆ˜ν•œ λ’€ ν˜„μž₯ 쑰사λ₯Ό μ‹€μ‹œν–ˆλ‹€. 쑰사 κ²°κ³Ό, κ³΅κ²©μžκ°€ 쿠팑 μ„œλ²„μ˜ 인증 취약점을 μ•…μš©ν•΄ 정상적인 둜그인 절차λ₯Ό κ±°μΉ˜μ§€ μ•Šκ³  고객 정보λ₯Ό μœ μΆœν•œ κ²ƒμœΌλ‘œ 확인됐닀.

μ •λΆ€λŠ” 11μ›” 30일뢀터 민관합동쑰사단을 κ°€λ™ν•˜κ³  있으며, κ°œμΈμ •λ³΄λ³΄ν˜Έμœ„μ›νšŒλŠ” 쿠팑이 κ°œμΈμ •λ³΄ 보호 κ΄€λ ¨ μ•ˆμ „μ‘°μΉ˜ 의무(μ ‘κ·Όν†΅μ œ, μ ‘κ·ΌκΆŒν•œ 관리, μ•”ν˜Έν™” λ“±)λ₯Ό μœ„λ°˜ν–ˆλŠ”μ§€ 쑰사 쀑이닀. β€˜ν•œκ΅­μ˜ μ•„λ§ˆμ‘΄β€™μœΌλ‘œ 뢈릴 만큼 λ§Žμ€ μ‚¬μš©μžκ°€ μ΄μš©ν•˜λŠ” μ„œλΉ„μŠ€μΈ 만큼 11μ›” 29μΌμ—λŠ” λ°”λ‘œ 2μ°¨ ν”Όν•΄ λ°©μ§€λ₯Ό μœ„ν•œ λŒ€κ΅­λ―Ό λ³΄μ•ˆ 곡지λ₯Ό μ§„ν–‰ν•˜κΈ°λ„ ν–ˆλ‹€. λ˜ν•œ 11μ›” 30일뢀터 3κ°œμ›”κ°„ 인터넷상 κ°œμΈμ •λ³΄ 유좜 및 λΆˆλ²• μœ ν†΅ 점검 κ°•ν™” κΈ°κ°„μœΌλ‘œ μš΄μ˜ν•œλ‹€.

ν•œνŽΈ ꡭ회 κ³Όν•™κΈ°μˆ μ •λ³΄λ°©μ†‘ν†΅μ‹ μœ„μ›νšŒ 졜민희 μœ„μ›μž₯은 30일 λ³΄λ„μžλ£Œλ₯Ό 톡해 이번 μ‚¬κ³ μ˜ ꡬ체적인 원인을 λΆ„μ„ν•œ κ²°κ³Όλ₯Ό κ³΅κ°œν–ˆλ‹€. μΏ νŒ‘μœΌλ‘œλΆ€ν„° 받은 μžλ£Œμ— λ”°λ₯΄λ©΄, μΏ νŒ‘μ€ 토큰 μ„œλͺ…ν‚€ 유효 인증기간에 λŒ€ν•΄ β€œ5~10λ…„μœΌλ‘œ μ„€μ •ν•˜λŠ” 사둀가 λ§Žλ‹€β€λΌλ©° β€œλ‘œν…Œμ΄μ…˜ 기간이 κΈΈκ³ , ν‚€ μ’…λ₯˜μ— 따라 맀우 λ‹€μ–‘ν•˜λ‹€β€κ³  λ‹΅λ³€ν•œ κ²ƒμœΌλ‘œ μ „ν•΄μ‘Œλ‹€.

졜 μœ„μ›μž₯ 츑은 이번 사고λ₯Ό μΆœμž… μ‹œμŠ€ν…œμ— λΉ„μœ ν•΄ μ„€λͺ…ν–ˆλ‹€. λ‘œκ·ΈμΈμ— ν•„μš”ν•œ β€˜ν† ν°β€™μ΄ 일회용 μΆœμž…μ¦μ΄λΌλ©΄, β€˜μ„œλͺ…ν‚€β€™λŠ” μΆœμž…μ¦μ„ λ°œκΈ‰ν•˜λŠ” 인증 도μž₯κ³Ό κ°™λ‹€. μΆœμž…μ¦μ΄ μžˆμ–΄λ„ 인증 도μž₯이 μ—†μœΌλ©΄ μΆœμž…ν•  수 μ—†μ§€λ§Œ, μ„œλͺ…ν‚€λ₯Ό μž₯κΈ°κ°„ λ°©μΉ˜ν•  경우 μ§€μ†μ μœΌλ‘œ μ•…μš©λ  수 μžˆλ‹€λŠ” 것이닀.

졜민희 μ˜μ›μ‹€μ΄ ν™•μΈν•œ κ²°κ³Ό, 쿠팑의 둜그인 μ‹œμŠ€ν…œμ—μ„œλŠ” 토큰이 생성 ν›„ μ¦‰μ‹œ νκΈ°λ˜λ„λ‘ μ„€κ³„λ˜μ–΄ μžˆμœΌλ‚˜, 토큰 생성에 ν•„μš”ν•œ μ„œλͺ… 정보가 λ‹΄λ‹Ή 직원 퇴사 μ‹œ μ‚­μ œλ˜κ±°λ‚˜ κ°±μ‹ λ˜μ§€ μ•Šκ³  λ°©μΉ˜λ˜μ–΄ λ‚΄λΆ€ 직원에 μ˜ν•΄ μ•…μš©λœ κ²ƒμœΌλ‘œ νŒŒμ•…λλ‹€.

졜민희 μœ„μ›μž₯은 λ³΄λ„μžλ£Œλ₯Ό 톡해 β€œμ„œλͺ…ν‚€ 갱신은 κ°€μž₯ 기본적인 λ‚΄λΆ€ λ³΄μ•ˆ μ ˆμ°¨μž„μ—λ„ μΏ νŒ‘μ€ 이λ₯Ό μ§€ν‚€μ§€ μ•Šμ•˜λ‹€β€λΌλ©° β€œμž₯κΈ° 유효 인증킀λ₯Ό λ°©μΉ˜ν•œ 것은 λ‹¨μˆœν•œ λ‚΄λΆ€ μ§μ›μ˜ μΌνƒˆμ΄ μ•„λ‹ˆλΌ, 인증체계λ₯Ό λ°©μΉ˜ν•œ 쿠팑의 쑰직적·ꡬ쑰적 문제의 결과”라고 λ°ν˜”λ‹€.

이번 μΉ¨ν•΄ μ‚¬κ³ μ˜ ν”Όν•΄ λŒ€μƒμžμ—κ²ŒλŠ” 이메일 λ˜λŠ” 문자 λ©”μ‹œμ§€λ₯Ό 톡해 κ΄€λ ¨ 사싀이 μ•ˆλ‚΄λλ‹€. μΆ”κ°€ λ¬Έμ˜κ°€ ν•„μš”ν•œ 경우 고객 μ„Όν„° λ˜λŠ” incident_help@coupang.com으둜 μ—°λ½ν•˜λ©΄ λœλ‹€. κ΄€λ ¨ μ •λ³΄λŠ” λ³„λ„μ˜ μ•ˆλ‚΄ νŽ˜μ΄μ§€λ₯Ό ν†΅ν•΄μ„œλ„ 확인할 수 μžˆλ‹€.

μΉ¨ν•΄ 사싀 μ•ˆλ‚΄λ¬Έκ³Ό λ³„λ„λ‘œ λ°•λŒ€μ€€ 쿠팑 λŒ€ν‘œμ΄μ‚¬λŠ” 30일 λ³„λ„μ˜ μž…μž₯문을 톡해 β€œκ΅­λ―Ό μ—¬λŸ¬λΆ„κ»˜ 큰 뢈편과 걱정을 λΌμ³λ“œλ € μ§„μ‹¬μœΌλ‘œ μ‚¬κ³Όλ“œλ¦°λ‹€β€λΌλ©°β€œμΏ νŒ‘μ€ κ³Όν•™κΈ°μˆ μ •λ³΄ν†΅μ‹ λΆ€, κ°œμΈμ •λ³΄λ³΄ν˜Έμœ„μ›νšŒ, ν•œκ΅­μΈν„°λ„·μ§„ν₯원, κ²½μ°°μ²­ λ“± 민관합동쑰사단과 κΈ΄λ°€νžˆ ν˜‘λ ₯ν•˜μ—¬ 좔가적인 ν”Όν•΄ μ˜ˆλ°©μ„ μœ„ν•΄ μ΅œμ„ μ„ λ‹€ν•˜κ² λ‹€β€λΌκ³  λ°ν˜”λ‹€.
jihyun.lee@foundryco.com

에이전틱 컀머슀의 뢀상 : AIκ°€ β€˜κ΅¬λ§€μžβ€™κ°€ 될 λ•Œ

25 November 2025 at 21:03

λͺ‡ λ…„ μ „λ§Œ 해도 직접 온라인 쇼핑을 ν•  μ—¬μœ κ°€ μžˆμ—ˆλ‹€. ν•˜μ§€λ§Œ μ΄μ œλŠ” κ·Έμ‘°μ°¨ κ΅¬μ‹μ²˜λŸΌ λŠκ»΄μ§„λ‹€. 일과 좜μž₯, μˆ˜λ§Žμ€ λ””μ§€ν„Έ μ•Œλ¦Ό μ‚¬μ΄μ—μ„œ 가격을 λΉ„κ΅ν•˜κ³  결제 정보λ₯Ό μž…λ ₯ν•  μ‹œκ°„μ‘°μ°¨ μ—†λ‹€.

κ³§ 그럴 ν•„μš”λ„ μ—†μ–΄μ§ˆ 전망이닀. AIκ°€ μƒν’ˆ μΆ”μ²œλΏ μ•„λ‹ˆλΌ β€˜μ§μ ‘ κ΅¬λ§€β€™κΉŒμ§€ 배우고 있기 λ•Œλ¬Έμ΄λ‹€.

이것이 λ°”λ‘œ β€˜μ—μ΄μ „ν‹± 컀머슀(agentic commerce)β€™μ˜ κ°œλ…μ΄λ‹€. μ•žμœΌλ‘œ 자주 λ“£κ²Œ 될 μš©μ–΄λ‘œ, 자율적인 AI μ—μ΄μ „νŠΈκ°€ μ‚¬μš©μžλ₯Ό λŒ€μ‹ ν•΄ ꡬ맀 ν–‰μœ„λ₯Ό μˆ˜ν–‰ν•˜λŠ” 세계λ₯Ό λœ»ν•œλ‹€. 예λ₯Ό λ“€μ–΄ β€œλ‚΄κ°€ μ’‹μ•„ν•˜λŠ” 원두가 λ–¨μ–΄μ§€λ©΄ λ‹€μ‹œ μ£Όλ¬Έν•΄μ€˜β€λΌκ³  λ””μ§€ν„Έ λΉ„μ„œμ—κ²Œ λ§ν•˜λ©΄, AIκ°€ 검색·비ꡐ뢀터 κ²°μ œμ™€ 배솑 μΆ”μ κΉŒμ§€ μ „ 과정을 μ•Œμ•„μ„œ μ²˜λ¦¬ν•œλ‹€.

CIOμ—κ²Œ 이 λ³€ν™”λŠ” μ‚¬μ†Œν•œ μ£Όμ œκ°€ μ•„λ‹ˆλ‹€. 고객을 ν™•λ³΄ν•˜κ³  μœ μ§€ν•˜λŠ” κΈ°μ—…μ˜ 방식 자체λ₯Ό λ‹€μ‹œ μ“°κ²Œ 될 μ°¨μ„ΈλŒ€ λ””μ§€ν„Έ μ „ν™˜μ˜ 물결이 될 수 있기 λ•Œλ¬Έμ΄λ‹€.

ν΄λ¦­μ—μ„œ μœ„μž„μœΌλ‘œ

κΈ°μ‘΄ μ „μžμƒκ±°λž˜λŠ” μΈκ°„μ˜ 주의λ ₯에 μ˜μ‘΄ν•œλ‹€. μ‚¬μš©μžκ°€ λ‘˜λŸ¬λ³΄κ³  λΉ„κ΅ν•˜κ³  결제 λ²„νŠΌμ„ ν΄λ¦­ν•œλ‹€. κΈ°μ—…μ˜ 역할은 λ§€λ ₯적인 μ›Ήμ‚¬μ΄νŠΈμ™€ λ§€λ„λŸ¬μš΄ 결제 과정을 μ„€κ³„ν•˜λŠ” κ²ƒμ΄μ—ˆλ‹€.

에이전틱 μ»€λ¨ΈμŠ€λŠ” 이런 곡식이 μ™„μ „νžˆ 바뀐닀. 이제 μ†ŒλΉ„μžλŠ” 행동이 μ•„λ‹ˆλΌ β€˜μ˜λ„β€™λ₯Ό μœ„μž„ν•œλ‹€. β€œμ›ν•˜λŠ” 것을 λ§ν•˜λ©΄ AIκ°€ μ‹€ν–‰ν•˜λŠ”β€ λ°©μ‹μœΌλ‘œ λ°”λ€ŒλŠ” 것이닀. 즉, λ‹€μŒ 고객은 μ‚¬λžŒμ΄ 아닐 μˆ˜λ„ μžˆλ‹€. λˆ„κ΅°κ°€λ₯Ό λŒ€μ‹ ν•΄ μ‡Όν•‘ν•˜λŠ” μ•Œκ³ λ¦¬μ¦˜μΌ μˆ˜λ„ μžˆλ‹€.

CIO μž…μž₯μ—μ„œ ν₯λ―Έλ‘œμš°λ©΄μ„œλ„ λΆˆμ•ˆν•œ λ³€ν™”λ‹€. μ‚¬μš©μž κ²½ν—˜μ—μ„œ 기계 κ°„ κ²½ν—˜μœΌλ‘œ 쀑심이 μ΄λ™ν•˜κΈ° λ•Œλ¬Έμ΄λ‹€. 이제 μŠΉλΆ€λŠ” ν™”λ €ν•œ μ΄λ―Έμ§€λ‚˜ λ§ˆμΌ€νŒ… 문ꡬ가 μ•„λ‹ˆλΌ, AI μ—μ΄μ „νŠΈκ°€ 이해할 수 μžˆμ„ 만큼 μ •ν™•ν•˜κ³  κ΅¬μ‘°ν™”λœ 데이터 ν’ˆμ§ˆμ— 달렀 μžˆλ‹€.

λ˜ν•œ 에이전틱 μ»€λ¨ΈμŠ€λŠ” 검색 μ΅œμ ν™”μ˜ κ°œλ…λ„ 근본적으둜 λ°”κΎΌλ‹€. κ²€μƒ‰μ˜ λŒ€μƒμ΄ 더 이상 μ‚¬λžŒμ΄ μ•„λ‹ˆλΌ AI이기 λ•Œλ¬Έμ΄λ‹€. AIλŠ” νŽ˜μ΄μ§€λ₯Ό νƒμƒ‰ν•˜μ§€ μ•Šκ³  데이터λ₯Ό λΆ„μ„ν•œλ‹€. 이 μƒˆλ‘œμš΄ μ„Έκ³„μ—μ„œ 데이터 ν’ˆμ§ˆμ΄ κ³§ β€˜λ§€μž₯ μ§„μ—΄λŒ€β€™κ°€ λœλ‹€.

λ³€ν™”μ˜ 초기 μ‹ ν˜Έ

λ³€ν™”λŠ” 이미 μ‹œμž‘λλ‹€. μ•„λ§ˆμ‘΄μ€ μžμ‚¬μ— μ—†λŠ” μƒν’ˆμ„ λ‹€λ₯Έ μ†Œλ§€μ μ—μ„œ λŒ€μ‹  κ΅¬λ§€ν•΄μ£ΌλŠ” β€˜λ‚˜λ₯Ό μœ„ν•œ ꡬ맀(Buy for Me)’ κΈ°λŠ₯을 μ‹œν—˜ 쀑이닀. μ˜€ν”ˆAI의 μ±—GPTλŠ” λŒ€ν™”μ°½ μ•ˆμ—μ„œ μ¦‰μ‹œ κ²°μ œν•  수 μžˆλŠ” κΈ°λŠ₯을 μΆ”κ°€ν–ˆλ‹€. ꡬ글은 λͺ©ν‘œ 가격에 λ„λ‹¬ν•˜λ©΄ μžλ™μœΌλ‘œ ꡬ맀λ₯Ό μˆ˜ν–‰ν•˜λŠ” AI μ‡Όν•‘ λͺ¨λ“œλ₯Ό μ‹€ν—˜ν•˜κ³  μžˆλ‹€.

결제 λ„€νŠΈμ›Œν¬λ„ 발 λΉ λ₯΄κ²Œ 움직이고 μžˆλ‹€. λΉ„μžμ™€ λ§ˆμŠ€ν„°μΉ΄λ“œλŠ” 승인된 μ˜ˆμ‚° λ‚΄μ—μ„œ 인증된 AI μ—μ΄μ „νŠΈκ°€ κ²°μ œν•  수 μžˆλ„λ‘ ν•˜λŠ” APIλ₯Ό 개발 쀑이닀. νΌν”Œλ ‰μ‹œν‹°λŠ” νŽ˜μ΄νŒ”κ³Ό ν˜‘λ ₯ν•΄ μ‚¬μš©μžκ°€ ν•˜λ‚˜μ˜ λŒ€ν™” μ“°λ ˆλ“œ μ•ˆμ—μ„œ μƒν’ˆμ„ νƒμƒ‰ν•˜κ³  ꡬ맀할 수 μžˆλ„λ‘ ν•˜κ³  μžˆλ‹€.

이 λͺ¨λ“  것은 미래의 이야기가 μ•„λ‹ˆλ‹€. 이미 μ‹œλ²” 운영 쀑이닀. 기술적 κΈ°λ°˜μ€ λ³΅μž‘ν•˜μ§€λ§Œ, κ²°κ³ΌλŠ” λ‹¨μˆœν•˜λ‹€. 마찰이 사라지고 쇼핑이 보이지 μ•Šκ²Œ λœλ‹€.

IT 리더가 μ£Όλͺ©ν•΄μ•Ό ν•˜λŠ” 이유

기술이 λ§ˆμ°°μ„ μ œκ±°ν•˜λ©΄ 거래 μ†λ„λ§Œ λΉ¨λΌμ§€λŠ” 것이 μ•„λ‹ˆλΌ, κ°€μΉ˜ μ‚¬μŠ¬μ˜ ꢌλ ₯ ꡬ쑰 μžμ²΄κ°€ 바뀐닀. AI μ—μ΄μ „νŠΈκ°€ ꡬ맀의 μ£Όμš” 관문이 되면 μ–΄λ–€ μƒν’ˆμ΄ λ…ΈμΆœλ˜λŠ”μ§€λ₯Ό κ²°μ •ν•˜λŠ” 주체도 AIκ°€ λœλ‹€. μΈκ°„μ˜ 클릭 κ²½μŸμ—μ„œ μ•Œκ³ λ¦¬μ¦˜ μ‹ λ’° 경쟁으둜 λ¬΄κ²Œμ€‘μ‹¬μ΄ μ΄λ™ν•˜λŠ” 것이닀.

CIO와 λ””μ§€ν„Έ 리더가 μ£Όλͺ©ν•΄μ•Ό ν•  μ„Έ κ°€μ§€ λ³€ν™”λŠ” λ‹€μŒκ³Ό κ°™λ‹€.

1. APIκ°€ μƒˆλ‘œμš΄ λ§€μž₯이닀. AI μ—μ΄μ „νŠΈλŠ” μ›ΉνŽ˜μ΄μ§€λ₯Ό νƒμƒ‰ν•˜μ§€ μ•ŠλŠ”λ‹€. κ΅¬μ‘°ν™”λœ 데이터λ₯Ό μ§ˆμ˜ν•œλ‹€. μ •λˆλœ μƒν’ˆ λͺ©λ‘, μΌκ΄€λœ 메타데이터, μ‹€μ‹œκ°„ 재고 정보가 AI κ΅¬λ§€μžμ—κ²Œ λ…ΈμΆœλ  수 μžˆλŠ”μ§€λ₯Ό μ’Œμš°ν•œλ‹€.

2. λ§ˆμΌ€νŒ…μ€ μ„€λ“μ—μ„œ μ •λ°€ν•¨μœΌλ‘œ 바뀐닀. AI μ—μ΄μ „νŠΈλŠ” 가격, ν’ˆμ§ˆ, 지속가λŠ₯μ„±, 이행 신뒰도 같은 객관적 사싀을 μ€‘μ‹œν•œλ‹€. λΈŒλžœλ“œλŠ” 이런 차별화λ₯Ό 데이터 μžμ²΄μ— λ…Ήμ—¬μ•Ό ν•œλ‹€.

3. λ³΄μ•ˆκ³Ό 신원 관리가 μ „λž΅μ΄ λœλ‹€. β€˜κ³ κ°μ„ μ•„λŠ” 것(KYC)’이 β€˜μ—μ΄μ „νŠΈλ₯Ό μ•„λŠ” 것(KYA)β€™μœΌλ‘œ 바뀐닀. 각 κ±°λž˜λŠ” AIκ°€ μ§„μ§œ μ‚¬μš©μžλ₯Ό λŒ€ν‘œν•˜λŠ”μ§€ 증λͺ…ν•΄μ•Ό ν•˜λ©°, μ§€μΆœ ν•œλ„λ₯Ό ν†΅μ œν•˜λŠ” μž₯μΉ˜κ°€ ν•„μš”ν•˜λ‹€.

이 λ³€ν™”λ₯Ό μ‹€ν—˜μ΄ μ•„λ‹Œ 인프라 과제둜 λ°›μ•„λ“€μ΄λŠ” 기업이 κ°€μž₯ λ¨Όμ € μŠΉλ¦¬ν•  것이닀.

μ•žμœΌλ‘œμ˜ 과제

μ• λ„λ¦¬μŠ€νŠΈλ“€μ€ 이미 이런 흐름을 μΆ”μ ν•˜κ³  μžˆλ‹€. λ§₯ν‚¨μ§€λŠ” 에이전틱 μ»€λ¨ΈμŠ€κ°€ 2030λ…„κΉŒμ§€ μ „ 세계 μ†Œλ§€ λ§€μΆœμ—μ„œ μ—°κ°„ 3μ‘°~5μ‘° λ‹¬λŸ¬ 규λͺ¨μ˜ 영ν–₯을 λ―ΈμΉ  수 μžˆλ‹€κ³  μΆ”μ‚°ν•œλ‹€. 어도비에 λ”°λ₯΄λ©΄, 2025λ…„ 7μ›” μƒμ„±ν˜• AI 기반 λΈŒλΌμš°μ €μ™€ μ±„νŒ… μ„œλΉ„μŠ€μ—μ„œ λ―Έκ΅­ μ†Œλ§€ μ‚¬μ΄νŠΈλ‘œ μœ μž…λœ νŠΈλž˜ν”½μ€ μ „λ…„ 동기 λŒ€λΉ„ 4,700% μ¦κ°€ν–ˆλ‹€.

숫자만 보면 κ±°λŒ€ν•˜μ§€λ§Œ, μ§„μ§œ μ΄μ•ΌκΈ°λŠ” κΈ°μ—… λ‚΄λΆ€μ—μ„œ μ–΄λ–€ λ³€ν™”κ°€ μΌμ–΄λ‚˜λŠ”μ§€μ— μžˆλ‹€.

μ†Œλ§€ 기업은 ν‘œμ€€ν™”λœ APIλ₯Ό 톡해 μƒν’ˆ 데이터λ₯Ό λ…ΈμΆœν•˜κΈ° μœ„ν•΄ 기술 μŠ€νƒμ„ μž¬νŽΈν•˜κ³  μžˆλ‹€. 결제 λ„€νŠΈμ›Œν¬λŠ” AI μ—μ΄μ „νŠΈκ°€ μ‹œμž‘ν•˜λŠ” κ²°μ œμ— μ‚¬μš©ν•  수 μžˆλŠ” μ§€μΆœ ν•œλ„ 기반 자격 증λͺ…을 μ‹œλ²” μ μš©ν•˜κ³  μžˆλ‹€. μ•Œλ¦¬λ°”λ°”μ™€ 미라클(Mirakl) 같은 κΈ€λ‘œλ²Œ λ§ˆμΌ“ν”Œλ ˆμ΄μŠ€λŠ” AI κ΅¬λ§€μžκ°€ μ—¬λŸ¬ 판맀자 간에 λŠκΉ€ 없이 μƒν’ˆμ„ νƒμƒ‰ν•˜κ³  ꡬ맀할 수 μžˆλ„λ‘ 곡유 에이전틱 컀머슀 계측에 νˆ¬μžν•˜κ³  μžˆλ‹€.

μƒνƒœκ³„κ°€ μ‹€μ‹œκ°„μœΌλ‘œ κ΅¬μΆ•λ˜κ³  μžˆλ‹€.

κΈ°νšŒμ™€ μœ„ν—˜

λͺ¨λ“  기술 도약이 κ·Έλ ‡λ“―, κΈ°νšŒμ™€ μœ„ν—˜μ€ λ™μ‹œμ— μ°Ύμ•„μ˜¨λ‹€. κΈ°νšŒλŠ” λ‹€μŒκ³Ό κ°™λ‹€.

  • μžμ—°μ–΄ λŒ€ν™”ν˜•μœΌλ‘œ μ΄λ€„μ§€λŠ” λ§€λ„λŸ¬μš΄ 온라인 쇼핑은 μž₯λ°”κ΅¬λ‹ˆ μ΄νƒˆμ„ 쀄이고 고객 좩성도λ₯Ό 높일 수 μžˆλ‹€.
  • 반볡적인 B2B 쑰달 μ—…λ¬΄λŠ” μ™„μ „ μžλ™ν™”λ˜μ–΄ μ‚¬λžŒμ΄ 더 높은 λΆ€κ°€κ°€μΉ˜ 업무에 집쀑할 수 μžˆλ‹€.
  • 데이터λ₯Ό AIκ°€ νƒμƒ‰ν•˜κΈ° 쒋은 ν˜•νƒœλ‘œ κ΅¬μ‘°ν™”ν•œ μ†Œκ·œλͺ¨ νŒλ§€μžλŠ” μƒˆλ‘œμš΄ 고객측에 도달할 수 μžˆλ‹€.

μœ„ν—˜μ€ λ‹€μŒκ³Ό κ°™λ‹€.

  • λ³΄μ•ˆ ν‘œμ€€μ΄ μ„±μˆ™ν•˜κΈ° μ „κΉŒμ§€λŠ” 사기와 신원 λ„μš©μ΄ 증가할 수 μžˆλ‹€.
  • μƒν˜Έμž‘μš©μ„ AI μ—μ΄μ „νŠΈκ°€ λŒ€μ‹  μ²˜λ¦¬ν•˜λ©΄μ„œ λΈŒλžœλ“œλŠ” 고객과의 직접적인 관계λ₯Ό μžƒμ„ 수 μžˆλ‹€.
  • κ·œμ œκ°€ 아직 따라가지 λͺ»ν•΄ AIκ°€ 잘λͺ»λœ μƒν’ˆμ„ κ΅¬λ§€ν–ˆμ„ λ•Œ μ±…μž„ μ†Œμž¬κ°€ λˆ„κ΅¬μ—κ²Œ μžˆλŠ”μ§€ λΆˆν™•μ‹€ν•˜λ‹€.

해법은 κ· ν˜•μ΄λ‹€. μžλ™ν™”λ₯Ό μ „μ œλ‘œ μ‹œμŠ€ν…œμ„ κ΅¬μΆ•ν•˜λ˜, 관리와 κ°μ‹œλŠ” μœ μ§€ν•΄μ•Ό ν•œλ‹€. μ—μ΄μ „νŠΈκ°€ ν–‰λ™ν•˜λ„λ‘ ν—ˆμš©ν•˜λ˜, 투λͺ…ν•œ 감사 좔적 체계λ₯Ό 섀계해야 ν•œλ‹€. 무엇보닀도 ν†΅μ œκΆŒμ€ 항상 κ³ κ°μ—κ²Œ μžˆμ–΄μ•Ό ν•œλ‹€.

기업이 μ€€λΉ„ν•΄μ•Ό ν•  것

ν•„μžκ°€ λͺΈλ‹΄μ€ μ‘°μ§μ—μ„œλŠ” AI μ—μ΄μ „νŠΈλ₯Ό λͺ¨λ°”μΌμ΄λ‚˜ μŒμ„± μΈν„°νŽ˜μ΄μŠ€ μ΄ˆμ°½κΈ°μ™€ λΉ„μŠ·ν•œ μƒˆλ‘œμš΄ λ””μ§€ν„Έ μ±„λ„λ‘œ 보기 μ‹œμž‘ν–ˆλ‹€. 이λ₯Ό μœ„ν•΄ μ„Έ κ°€μ§€ μΈ‘λ©΄μ—μ„œ μ€€λΉ„ν•˜κ³  μžˆλ‹€.

  • 데이터. μ‚¬λžŒκ³Ό 기계가 λͺ¨λ‘ 이해할 수 μžˆλ„λ‘ μƒν’ˆΒ·μ„œλΉ„μŠ€ 데이터λ₯Ό ν‘œμ€€ν™”ν•˜κ³  μžˆλ‹€.
  • λ³΄μ•ˆ. AI μ—μ΄μ „νŠΈκ°€ 행동할 λ•Œ 적절히 인증되고 μ±…μž„ μ†Œμž¬λ₯Ό λΆ„λͺ…νžˆ ν•  수 μžˆλ„λ‘ μœ„μž„ 승인 방식을 κ²€ν† ν•˜κ³  μžˆλ‹€.
  • λ§ˆμŒκ°€μ§. μΈν„°νŽ˜μ΄μŠ€μ™€ 클릭이 μ•„λ‹ˆλΌ μ˜λ„μ™€ 결과의 κ΄€μ μ—μ„œ μ‚¬κ³ ν•˜λ„λ‘ νŒ€μ„ κ΅μœ‘ν•˜κ³  μžˆλ‹€.

에이전틱 μ»€λ¨ΈμŠ€κ°€ μ„±μˆ™ν•˜λŠ” 데 μ‹œκ°„μ΄ 더 걸리더라도, 이런 μ—­λŸ‰μ€ μ§€κΈˆ λ‹Ήμž₯ λ””μ§€ν„Έ 운영 μ „λ°˜μ„ κ°•ν™”ν•˜λŠ” 데도 도움이 λœλ‹€.

μžλ™ν™”μ˜ 인간적 μΈ‘λ©΄

μΌκ°μ—μ„œλŠ” AI μ—μ΄μ „νŠΈμ—κ²Œ ꡬ맀λ₯Ό λ§‘κΈ°λ©΄ μƒκ±°λž˜μ—μ„œ 인간적인 감성이 μ‚¬λΌμ§ˆ 것이라고 μš°λ €ν•œλ‹€. ν•„μžμ˜ 생각은 λ‹€λ₯΄λ‹€.

기술이 반볡적인 업무λ₯Ό μ²˜λ¦¬ν•˜λ©΄ μ‚¬λžŒμ€ μƒˆλ‘œμš΄ μƒν’ˆμ„ νƒμƒ‰ν•˜κ³  관계λ₯Ό κ΅¬μΆ•ν•˜λ©° κ²½ν—˜ 자체λ₯Ό μ¦κΈ°λŠ” λ“± 더 높은 μˆ˜μ€€μ˜ μ˜μ‚¬κ²°μ •μ„ ν•  μ‹œκ°„μ„ ν™•λ³΄ν•˜κ²Œ λœλ‹€. 이 μƒˆλ‘œμš΄ ν™˜κ²½μ—μ„œ μŠΉμžκ°€ λ˜λŠ” 기업은 μžλ™ν™”μ— 신뒰와 투λͺ…성을 κ²°ν•©ν•  것이닀. 이런 기업은 고객이 반볡적인 선택을 AI에 μœ„μž„ν•˜λ”λΌλ„, AIκ°€ 고객의 이읡을 μ΅œμš°μ„ μœΌλ‘œ μ‚Όκ³  μžˆλ‹€λŠ” 확신을 쀄 수 μžˆμ–΄μ•Ό ν•œλ‹€.

λ””μ§€ν„Έ μ „ν™˜μ˜ λͺ©μ μ€ 항상 κ°™λ‹€. μ‚¬λžŒμ„ λŒ€μ²΄ν•˜λŠ” 것이 μ•„λ‹ˆλΌ μ‚¬λžŒμ΄ 이룰 수 μžˆλŠ” μ„±κ³Όλ₯Ό μ¦ν­ν•˜λŠ” 것이닀.

μƒˆλ‘œμš΄ μœ ν˜•μ˜ 고객

λͺ¨λ“  큰 λ””μ§€ν„Έ μ „ν™˜μ€ μƒˆλ‘œμš΄ μœ ν˜•μ˜ 고객을 λ°λ €μ˜¨λ‹€.

  • 웹은 μ „ 세계 μ–΄λ””μ—μ„œλ‚˜ μ ‘μ†ν•˜λŠ” κΈ€λ‘œλ²Œ 고객을 λ§Œλ“€μ—ˆλ‹€.
  • λͺ¨λ°”일은 항상 μ—°κ²°λœ 고객을 λ“±μž₯μ‹œμΌ°λ‹€.
  • AIλŠ” μ—μ΄μ „νŠΈλ‘œ λŒ€ν‘œλ˜λŠ” 자율적인 고객을 데렀올 것이닀.

CIOμ—κ²Œ 이것은 λ„μ „μ΄μž μ΄ˆλŒ€μ΄λ‹€. AI μ—μ΄μ „νŠΈκ°€ λ””μ§€ν„Έ 문을 λ‘λ“œλ¦¬κΈ°λ₯Ό 마λƒ₯ 기닀릴 μˆ˜λ„ 있고, 이듀을 λ§žμ΄ν•  μ€€λΉ„κ°€ 된 μ‹œμŠ€ν…œμ„ μ§€κΈˆλΆ€ν„° 섀계할 μˆ˜λ„ μžˆλ‹€.

μΈν”„λΌλŠ” 이미 κ°–μΆ°μ§€κΈ° μ‹œμž‘ν–ˆκ³  ν‘œμ€€μ€ λ§Œλ“€μ–΄μ§€κ³  있으며, κΈ°νšŒλŠ” ν™œμ§ μ—΄λ € μžˆλ‹€. λ‹€μŒμ— AI λΉ„μ„œκ°€ β€œκ°€μž₯ 쒋은 쑰건을 μ°Ύμ•„ 이미 μ£Όλ¬ΈκΉŒμ§€ μ™„λ£Œν–ˆλ‹€β€λΌκ³  λ§ν•˜λŠ” μˆœκ°„, 에이전틱 컀머슀 μ‹œλŒ€κ°€ κ³΅μ‹μ μœΌλ‘œ μ—΄λ Έλ‹€λŠ” 사싀을 μ•Œκ²Œ 될 것이닀.
dl-ciokorea@foundryco.com

The rise of agentic commerce: When AI becomes the shopper

25 November 2025 at 12:05

A few years ago, I could still keep up with my own online shopping. Now, even that feels old-fashioned. Between work, travel and a thousand digital distractions, I do not have time to compare prices or fill out checkout forms.

Soon, I may not have to. Artificial intelligence is learning to do it for me; not just to recommend, but to buy.

That is the promise of agentic commerce, a term we will all be hearing more often. It describes a world where autonomous AI agents act as shoppers on our behalf. You tell your digital assistant, β€œReorder my favorite coffee beans when I am running low,” and it quietly handles everything from searching and comparing to paying and tracking delivery.

For CIOs, this is not a side show. It is the next wave of digital transformation; one that could rewrite how businesses attract, serve and keep customers.

From clicking to delegating

Traditional e-commerce depends on human attention. People browse, compare, decide and click. The company’s job has been to design beautiful sites and frictionless checkout flows.

Agentic commerce flips that script. Customers will increasingly delegate intent, not actions. They will say what they want and an AI agent will execute.

That means your next β€œcustomer” might not be a person at all. It could be an algorithm shopping on behalf of one.

As a CIO, I find that idea both thrilling and unsettling. It moves us from user experience to machine-to-machine experience. The winning brands will not be the ones with the best imagery or marketing copy. They will be the ones whose data is clean, structured and trustworthy enough for an AI agent to understand.

Also, agentic commerce fundamentally reshapes search optimization because the audience for search is no longer just human. It’s now AI agents and they don’t β€œsearch” the way people do. They parse data, not pages.

In this world, data quality is the new storefront.

Early signals of the shift

We are already seeing the change.

Amazon is testing a β€œBuy for Me” feature that lets its AI purchase items from other retailers if Amazon does not carry them. OpenAI’s ChatGPT added instant checkout inside chat conversations. Google is piloting an AI shopping mode that tracks prices and buys automatically when a target price is reached.

Payment networks are following fast. Visa and Mastercard are building APIs that let verified AI agents spend within approved budgets. AI startups like Perplexity are partnering with PayPal so users can discover and buy products in a single chat thread.

None of this is futuristic. It is already piloting. And while the technical underpinnings are complex, the outcome is simple: friction disappears. Shopping becomes invisible.

Why this matters to leaders

When technology removes friction, it doesn’t just speed up transactions. It changes who holds power in the value chain.

If AI agents become the dominant gateway to purchasing, they will decide which products get visibility. That is a fundamental shift from competing for human clicks to competing for algorithmic trust.

For CIOs and digital leaders, I see three implications:

  1. Your APIs are your new storefront. Agents do not browse web pages. They query structured data. Clean catalogs, consistent metadata and real-time inventory feeds will determine if you are even visible to AI shoppers.
  2. Marketing will evolve from persuasion to precision. Agents care about facts such as price, quality, sustainability and fulfillment reliability. Brands will need to embed their differentiation into the data itself.
  3. Security and identity become strategic. β€œKnow your customer” becomes β€œKnow your agent.” Every transaction will need proof that an AI truly represents its user, with controls to limit what it can spend.

The companies that treat these changes as infrastructure work, not as experiments, will win early.

The road ahead

Analysts are already tracking the momentum. McKinsey estimates that agentic commerce could influence up to $3 to $5 trillion annually in global retail sales by 2030. Traffic to US retail sites from GenAI browsers and chat services increased 4,700% year-over-year in July 2025, according to Adobe.

Those are big numbers, but the real story is what happens inside the enterprise.

Retailers are reorganizing their technology stacks to expose product data through standardized APIs. Payment networks are piloting agent-initiated credentials with built-in spending caps. And global marketplaces such as Alibaba and Mirakl are investing in shared β€œagentic commerce” layers so AI shoppers can discover and buy across multiple sellers seamlessly.

The ecosystem is being built in real time.

Opportunities and risks

As with every technological leap, opportunity and risk arrive together.

The opportunity:

  • Seamless, conversational online shopping could reduce abandoned carts and increase loyalty.
  • Routine B2B procurement can become fully automated, freeing people for higher-value work.
  • Small merchants can reach new audiences if their data is structured for AI discovery.

The risk:

  • Fraud and identity spoofing could rise before security standards mature.
  • Brands may lose direct customer relationships as agents handle the interaction.
  • Regulations have not caught up, leaving questions about liability if an AI buys the wrong product.

The way forward is balance. Build for automation but preserve oversight. Allow agents to act, but design transparent audit trails. And above all, keep the customer in control.

What I am telling my own teams

Within my organization, we have started treating AI agents as a new type of digital channel, similar to the early days of mobile or voice. That means preparing in three ways:

  1. Data readiness. We are standardizing product and service data so it is understandable by both humans and machines.
  2. Security readiness. We are exploring delegated authorization so that when an AI agent acts, it is properly authenticated and accountable.
  3. Mindset readiness. We are teaching teams to think in terms of intents and outcomes rather than interfaces and clicks.

Even if agentic commerce takes longer to mature, these capabilities strengthen every part of digital operations today.

The human side of automation

Some worry that letting AI agents shop for us removes the human touch from commerce. I see it differently.

When technology handles repetitive tasks, people gain time for higher-order decisions such as exploring new products, building relationships or enjoying the experience.

The winners in this new landscape will combine trust and transparency with automation. They will let customers delegate routine choices while staying confident that the AI is acting in their best interest.

That has always been the purpose of digital transformation: not replacing people, but amplifying what they can achieve.

A new kind of customer

Every major digital shift brings a new kind of customer.

  • The web brought the global customer.
  • Mobile brought the always-connected customer.
  • AI will bring the autonomous customer, represented by an agent.

For CIOs, this is both a challenge and an invitation. We can wait until these agents start knocking on our digital doors or we can design systems ready to welcome them.

The infrastructure is forming, the standards are emerging and the opportunity is wide open.

And the next time your AI assistant says, β€œI found the best deal and already placed the order,” you will know that the era of agentic commerce has officially begun.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

❌
❌