Agent Orchestra
About this app
One expert, many hands: the year slider 2006→2026 shows that it is not the speed that changed but the unit of work — from a function per hour to a feature per afternoon, because one person steers parallel agents.
The idea
AI agents change the role of experts in software development. Someone who used to write every line now describes goals, hands tasks to several agents and reviews what comes back. This app is a thinking model for that claim. It does not measure real productivity data; it contrasts four stylised eras: statistical learning, deep learning in teams, chat assistants and agents that do the work themselves. The core message: the speed of the human stays the same, while the unit of their work grows from a single function to a whole product feature.
What you see
An expert sits at the centre of a 3D scene and sends work packets to “hands” on an inner ring; finished packets land as results on an outer ring. In 2006 that is one pair of hands, in 2014 a team of three, in 2023 a chat assistant whose drafts flow back to the expert, and in 2026 eight agents building in parallel. A values panel shows the unit of work, the number of parallel hands, results per day and the human’s role. In the agent era, some packets fly back for review.
Try it
Drag the year slider slowly from 2006 to 2026, or start the time-lapse. Watch two things: packets get bigger (function, model, draft, feature), and from 2026 on they move at the same time. Compare 2023 with 2026: with a chat assistant the expert still passes on and integrates every draft, with agents what remains is deciding and reviewing. Notice how quickly the result ring fills in each era. The numbers are illustrative and meant to show the ratio between eras.
Frequently asked questions
- What is an AI agent in software development?
- An AI agent is a language model that works through a task in steps instead of only answering: it reads code, edits it, runs tests and reports the result. Several agents can work on separate subtasks in parallel. The human sets goals and reviews the output.
- How does a chat assistant differ from an agent?
- A chat assistant returns answers or drafts that a person has to copy, check and integrate. An agent carries the work through to a reviewable result, such as a finished code change. Human work shifts from implementing to describing, deciding and reviewing.
- Will AI agents make experts redundant?
- The model in this app suggests the opposite: the more agents work in parallel, the more you need someone who states goals clearly and judges results. Expertise moves into review and decision-making. How this plays out depends heavily on the field.
- Are the numbers in the app measured?
- No. Values such as “results per day” are illustrative and chosen to make the ratio between the four eras visible. The app is a thinking model for talks and workshops on AI transformation, not an empirical study.
Subject: AI transformation in business
More from AI transformation in business
A VisuApp by heyprof: interactive, free in your browser, no sign-up. What is a VisuApp? · All apps