apidays Helsinki & North 2025 - REST in Peace? Hunting the Dominant Design for AI-Agent Plug-Ins, Antti Törrönen (Kwork Innovations)

REST in Peace? Hunting the Dominant Design for AI-Agent Plug-Ins
Antti Törrönen, Founder (A.I. apps) at Kwork Innovations

apidays Helsinki & North 2025 - APIs for Innovation, Intelligence, and Impact
June 3 & 4, 2025

------

Check out our conferences at https://www.apidays.global/

Do you want to sponsor or talk at one of our conferences?
https://apidays.typeform.com/to/ILJeAaV8

Learn more on APIscene, the global media made by the community for the community:
https://www.apiscene.io

Explore the API ecosystem with the API Landscape:
https://apilandscape.apiscene.io/

Transcript

1. REST in Peace? Why API Thinking Needs an Agent Upgrade 1
2. Agenda • What is an A.I. agent? • Examples • What is a dominant design? • Approaches to A.I. integration • Vision for the future
3. Antti ”Andy” Törrönen • ML apps since 2015 • Generative A.I. SaaS since 2021 • Focus on Generative A.I. SaaS apps • Build with me? • LinkedIn Antti ”Andy” Törrönen • X torronen
4. Examples 5
9. Miten ennusteet?
10. Agentic agent with agency 0 Generates text/media on request 1 Fetches data from external sources 2 Has internal monologues or multiple agents converse 3 Manipulates through predefined interfaces 4 Autonomously discovers new services and completes tasks through them 5 Works as an autonomous professional 11
12. Dominant Design Theory 14
14. Suarez, Fernando & Grodal, Stine & Gotsopoulos, Aleksios. (2015). Perfect Timing? Dominant Category, Dominant Design, and the Window of Opportunity for Firm Entry. Strategic Management Journal. 36. 10.1002/smj.2225.
16. A.I. Agent-to-world: Current contenders 19
17. Approaches 1. Conventional: Large Language Model is only a subcomponent 2. JSON-centric: ’Tools use’ / ’Structured output’ 3. UI-centric: ’Operator’ / ’computer-use’ 4. Codegen: Self-coded middleware 5. Specialized context services (Model Context Protocol, llm.txt) 6. Emerging contenders: agent2agent, LangChain AgentProtocol… … 20
19. JSON-Centric Interfaces 23 MCP – Model Context Protocol Tool Calling Structured JSON Output Final answer from the model {"tool":"name", "args":{…}} JSON What you get back Model API itself (auto-runs the integrations) You (call the tool the model names) You (use JSON how you wish) Who invokes the integrations? MCP server URL List of tools + each tool’s JSON schema Response JSON schema What you must send every request Fully-managed plug-in / agent execution Orchestrate multi-step workflows Get structured data for API, UI or DB Typical use case
20. ’Operator’ or ’computer use’ Clickable user interface as the machine readable interface Allows A.I. agents to do everything humans can do 24
24. Automatically Coded Clients A.I. will code a script / mini-app on demand 28
28. What’s next? Discoverability Payments (Authorization) 32
30. REST in Peace? "Any headline that ends in a question mark can be answered by the word no.“ Betteridge's law of headlines 34
31. Build with me! LinkedIn Antti ”Andy” Törrönen X torronen antti.torronen@kwork.fi +358 44 323 7002