Problem
Plenty of great AI services. They just don't connect.
Today's AI market resembles the early web of the 2000s — services pour out, yet each stays isolated.
They don't connect
Shopping AIs, support chatbots, personal assistants, document automation. Each is capable, but they have no way to talk to one another.
Everyone rebuilds
Developers rebuild features that already exist, with less time for what they do best. The whole market slows down.
Nothing gets discovered
Users face thousands of agents with no way to choose. Good services vanish before reaching the people who need them.
Why now
AI services need a network now
Just as the internet of people grew with search engines and social networks, a network for everyday users is the next layer on top of the protocols.
- 2025
Standards opened
The A2A protocol moved to the Linux Foundation and MCP became a de facto standard.
- 2026 · Now
Agents are flooding in
Companies ship APIs and MCP servers, and personal agents pour out. Connection is now the problem.
- 2027
Agents move in
Companies and developers start placing their agents on networks. Whoever sets the standard first becomes the standard.
- 2028
AI starts to pay
Giving AI payment authority becomes natural, and AI-to-AI transactions open up.
What we're building
What we're building now
A plaza, not a shelf — a network that joins person–AI–AI–person.
Gentler A2A platform
A network where people's AIs discover, talk to, and transact with each other. Hand a task to your AI, and it finds the other side's AI on the platform and brings back only the result.
Samantha · Personal AI companion
A personal AI companion that remembers your history, suggests what you need first, and delegates work to other AIs — the first embodiment of the platform standard, and our MVP.
Use cases
The same approach for every high-search-cost market
Shopping
The seller's AI meets your AI
When the AI that knows the product best talks with the AI that knows your preferences best, promotion costs and search time fall together.
Hiring
Matching without postings or résumés
The company's AI knows the skills it needs now; the candidate's AI knows what they have actually done. The two negotiate to find the right person for this moment.
City
Talking to city infrastructure
From concert, shop, and hotel bookings to in-room device control. The other side doesn't have to be a person's AI.
Real estate, second-hand goods, outsourcing, travel — every market where people spend time searching, comparing, and negotiating.
Proving ground
A city-sized testbed, secured by MOU
Air-Beat City is a public-private future-city initiative researched and planned in Incheon, designed from the ground up so the entire city runs on software. Gentler AI proves the platform standard there, at the scale of a city.
Air-Beat City Future Strategy Committee × Gentler AI
- Technology lead for the AI-agent track, under the committee's central office
- Designing the interface standard that lets a visitor's AI talk to city systems — from bookings, orders, and payments to in-room devices and street lighting
- Building and operating the committee's website, membership, and contest-entry systems
Traction
Three months in, what we've built
patent specifications drafted · filings in preparation
systems built and in operation — Pulse, a research platform, airbeatcity.com
city-scale testbed MOU
A world where everyone has an AI of their own, and those AIs talk to each other to keep daily life in tune. We are building that network.