Connecting external agents to 'AI for All': What to check before onboarding

Among the three consortia—SK Telecom, Kakao, and KT—selected by the Ministry of Science and ICT in August 2026 as operators of the 'AI for All' project, Kakao and KT have proposed structures that let external companies link and distribute their AI agents. Newspim reported in a feature article on November 12, 2026, that SK Telecom is also pursuing expanded integration with specialized firms.

What was decided under 'AI for All'

• According to the government announcement, 2026 will see support with 512 Nvidia B200 GPUs, and participating operators will build general-purpose chatbots as well as AI agent services specialized for public and daily life.
• The 2027 government budget proposal allocates 250 billion won to the 'AI for All' project, including a plan to assign 10 billion won each in operating costs to the three consortia.
• The 2027 government proposal also allocates 50 billion won to support the AI agent ecosystem.

Each consortium emphasized its already-established points of contact with the public: SK Telecom through phone and text usage, Kakao through KakaoTalk, and KT through its portal and telecom, media, and partner service linkages. They explain this can reduce the burden on users of having to learn a new AI or separately find and install a service.

What changes for sellers when an AI assistant does the choosing

Newspim noted that when an AI assistant handles search, comparison, and booking on a consumer's behalf, the number of candidates the consumer directly reviews may shrink, and the small set of options the AI presents can become the actual starting point for a transaction. It also noted that businesses that had been losing out in advertising spend and search visibility competition could be discovered through AI-driven comparison, while businesses not connected to a particular platform risk being excluded from consideration even if their prices are reasonable.

In a July 2026 analysis of competition in the AI market, the OECD assessed that while technological competition in AI models is intensifying, companies holding computing infrastructure, data, and customer access points may continue to hold an advantage.

What conditions should external companies check before onboarding

Newspim noted that onboarding itself does not guarantee success, and listed the following items companies should examine.

• The cost of building connection technology and passing review
• Platform fees
• The burden of handling refunds and disputes
• The scope of customer data access
• Contract termination conditions

The same article noted it also matters whether a company can offer the same service through other AI platforms without being locked into a single channel, and whether service history can be transferred with customer consent. It also noted that the pace of product improvement depends on how much performance data held by the platform external companies can access.

Implications for AX at Korean companies

If you are considering linking your own AI agent or service to a major platform, it makes sense to examine contract terms before user scale. In particular, the scope of customer data access and termination conditions should be read together with whether the same service can be offered on other AI platforms.

If you run a sales or service business, it is worth checking whether your product information and shipping/cancellation terms are organized in a form AI can compare. In a structure where an AI assistant narrows down candidates, that information becomes the material for comparison.

This week, pull out the contracts for external platforms you currently operate on or are considering, and check which clauses specify the scope of customer data access and the conditions for contract termination.

Source: [Reading the Economy Through AI] ② Government-grown AI platforms: Is the market door open?