Everyone Is Now a Manager - Interfaces in the World of Agents

This ad is not shown to multipass and full ticket holders
React Summit
React Summit 2026
June 11 - 15, 2026
Amsterdam & Online
The biggest React conference worldwide
Learn More
In partnership with Focus Reactive
Upcoming event
React Summit 2026
React Summit 2026
June 11 - 15, 2026. Amsterdam & Online
Learn more
Bookmark
Rate this content

Agent demos are shiny; daily use is messy. In this talk I’ll share a handful of short, honest stories from shipping agent workflows with real teams—what we tried, how people reacted, what surprised us, and what we changed the next day. Rather than theory, you’ll hear concrete moments (wins, faceplants, near-misses) and the practical learnings that emerged. You’ll leave with clear and battle-tested principles that encompass the new world we are all approaching.

This talk has been presented at TechLead Conf London 2025: Adopting AI in Orgs Edition, check out the latest edition of this Tech Conference.

FAQ

The speaker of the talk is Sam Kmezverk, an engineer.

The speaker faced the challenge of the content becoming outdated quickly due to rapidly evolving technology, making half of his slides obsolete.

The first principle is that only one person or agent should be responsible for the implementation of a task to avoid conflicts and confusion.

The speaker mentions NASA's Mars Climate Orbiter incident where two teams assumed ownership of the same requirement, leading to a costly failure.

The second principle is that decisions and requirements need to have a specific name attached to them, ensuring accountability.

Decision accountability is important to ensure clear responsibility and prevent issues when something goes wrong, making it easier to debug and rectify.

The third principle is that users must trust the system; without trust, users won't derive much value from it.

User trust is emphasized because LLMs do not behave as predictably as traditional software, leading to potential user frustration and abandonment if trust is not established.

The three core principles are task ownership, decision accountability, and user trust.

The initial topic of the talk was about different initiatives and interfaces related to agents and new processes for developers.

Sam Komesarook
Sam Komesarook
6 min
28 Nov, 2025

Comments

Sign in or register to post your comment.
Video Summary and Transcription
Sam Kmezverk's talk covers the evolving tech landscape, emphasizing key principles for agents' implementation in tasks and decision-making processes. Addressing challenges in agent orchestration, debugging system issues, and cultivating user trust are crucial for successful interface development.

1. Key Principles for Agent Implementation

Short description:

Sam Kmezverk's talk covers the evolving tech landscape, emphasizing key principles for agents' implementation in tasks and decision-making processes.

Welcome, everyone. My name is Sam Kmezverk. I am an engineer, and thank you for your time. When I first got invited to do this talk, I wanted to chat about the different initiatives and the interfaces that were coming out to deal with agents, what the new processes look like, what that meant for us as developers. At the time, Access and Duet UI were the main things that everyone was talking about, so I built my talk around those. Shortly thereafter, everyone on the internet seemed to be talking about how dynamically generated LLM interfaces for the future. So that made half my slides obsolete. I rewrote it only for Antigravity to ship two weeks later and make the other half obsolete. So after this, I realized that trying to give a talk about specific agent tech is probably not going to be the best use of our time together. Whatever I say up here is probably going to be outdated by the time this conference is over. So instead, I want to leave you with a few principles that as a FDE and as someone who's worked in the field for a little bit now, I'm confident that these principles will remain true, no matter what the next big thing in UI agent interfaces is.

The first one comes from human orgs, and that is that only one person, only one agent should be responsible for the implementation of a task at most. In 1999, the NASA Mars Climate Orbiter was lost, which would cost a breezy 325 mil due to two teams simultaneously assuming that they owned the same requirement. Likewise, I'm sure all of us in this room have dealt with merged conflict hell when two people are confident that they're editing the same file with different contexts. And even in systems design, if you have two nodes that both think that they're the primary, you'll get what's called the split brain problem. And this is something that I've also seen in deployment when people are trying to apply agents to organizations.

One deployment I was on was with a company that was trying to automate their invoicing. So you create two agents, and critically, both of the agents were told your goal is to ensure that no invoice is left unsent. The learning here is that each task needs to be delegated once and only once to a single agent. That agent has complete control. The next principle also comes from human orgs, and that is that decisions and requirements need to have a specific name attached to them. At Tesla and SpaceX, they have a rule where if there's a requirement built into some product, some name needs to be attached to it, ideally someone within the org that they can reference if something goes wrong. Similarly, at Amazon, they have this single-threaded leadership model where one person is responsible for an entire project, and then their name is attached to all the decisions they're in. This is also something that I've seen in the field as well. I was working for an advertising agency that was looking to automate their end-of-campaign insights. They created an orchestrator agent, a whole bunch of separate agents to gather data from CRMs and surveys and whatnot, and then a whole bunch of agents on top of that that would synthesize the findings into something concrete. It worked very well until certain metrics just disappeared out of the blue.

2. Challenges in Agent Orchestration

Short description:

Addressing challenges in agent orchestration, debugging system issues, and cultivating user trust are crucial for successful interface development.

I was working for an advertising agency that was looking to automate their end-of-campaign insights. They created an orchestrator agent, a whole bunch of separate agents to gather data from CRMs and surveys and whatnot, and then a whole bunch of agents on top of that that would synthesize the findings into something concrete. It worked very well until certain metrics just disappeared out of the blue.

Certain recommendations were way too optimistic, and then the question naturally became, well, who's responsible for this? Why did they decide to either give this recommendation or omit some metric? If you have the case where you have dozens of agents all spewing out 10,000 lines worth of logs every day, it's going to be very, very difficult to debug that.

The next principle is to ensure that your system allows for debugging, not just in location, which I sort of associate with a name, but also in time and intent. The last principle that I'll leave you with is a very basic one, but it's one that's been consistent on every deployment I've ever been on. That is that if users don't trust your system, they're not going to get much value from it.

Check out more articles and videos

We constantly think of articles and videos that might spark Git people interest / skill us up or help building a stellar career

PrimeVue | The Next-Gen UI Component Library
Vue.js Live 2024Vue.js Live 2024
24 min
PrimeVue | The Next-Gen UI Component Library
Top Content
Prime Vue is a comprehensive UI component suite with over 90 components, including date pickers, buttons, tables, and grids. It offers flexibility through styled and unstyled modes, allowing for customization using design tokens or Tailwind. Prime Vue is WCAG compliant and supports Material design. The upcoming version 4 introduces a new theming API using CSS variables, and it includes features like dark mode switching and integration with Figma. The team has plans to release a UI Designer, advanced components, and a drag-and-drop UI Builder in the future.
Raygui: An Immediate-Mode C UI for Wasm Tools Development
JS GameDev Summit 2023JS GameDev Summit 2023
19 min
Raygui: An Immediate-Mode C UI for Wasm Tools Development
Welcome to the presentation of RightGui, an immediate mode CUI library for tools development. RightGui is a high-performance library that is stateless and uses small self-contained functions to process inputs and draw controls. It provides a variety of controls, customizable styles, and support tools for live style configuration. Additionally, there are tools like the icons editor and layout creator that allow for customizations and code generation. Reiki offers a variety of tools, including a sound editor, image manipulation canvas, and resource packer.
Game Development with ReactJS, CSS, and React Three Fiber
JS GameDev Summit 2023JS GameDev Summit 2023
22 min
Game Development with ReactJS, CSS, and React Three Fiber
Jorge Rubiano, a Software Engineer from Colombia, shares puzzle, isometric, and board games developed using React.js and CSS. He demonstrates the use of WebGL and 3.js for creating 3D games on the web. ReactiveFiber, a renderer that combines React and 3.js, is showcased in movement-based and color-changing games. The Talk concludes with the development of a bowling game using ReactiveFiber and complex components.
Test your UI in the REAL Browser
TestJS Summit 2021TestJS Summit 2021
33 min
Test your UI in the REAL Browser
Storybook is a powerful tool for building UI components and testing them. It allows for easy reuse and compatibility with other tools. Storybook 6.4 introduces interactive stories and live coding, making it easier to create and debug complex components. It also integrates with popular testing libraries like Jest and Testing Library. Storybook aims to bridge the gap between end-to-end testing and unit testing, providing automated testing options for UI components.
Building a Custom Component Library – Fast
JSNation Live 2021JSNation Live 2021
8 min
Building a Custom Component Library – Fast
Today, I want to talk to you about building custom component libraries fast. Every company should have its own standard set of UI controls for consistency in design, technology, and accessibility. Companies shouldn't build the harder parts of those libraries themselves, such as date pickers and data grids. But as a software developer, your time is valuable. Building custom component libraries is a better use of your time than reinventing the wheel.
Ensuring your Users are on the Right Path: the Future of Modals and Focus Management
JSNation 2022JSNation 2022
17 min
Ensuring your Users are on the Right Path: the Future of Modals and Focus Management
This Talk discusses the new features of Modals and focus management in web development. The dialogue element allows for the creation of modal dialogs that appear on top of other elements and prevent interaction with elements below. The note element can mark a subtree of the DOM as inert, making it unusable. The field set component can group input elements together and disable them. Using these new primitives can improve focus control and user experience in web applications.

Workshops on related topic

Should we have business logic in the UI?
JSNation 2022JSNation 2022
148 min
Should we have business logic in the UI?
Workshop
Samuel Pinto
Samuel Pinto
How many times did you say or hear “this is business logic, it should not be here”?In this workshop, we will create a modern frontend application using old patterns and you will learn how to build apps that have decoupled UI and services.We will start with a React application that has its whole logic in the UI. Then step by step we will extract the rules and operations to hit that sweet spot of independence.