Amelia Wattenberger is a designer, engineer, and data-visualization author who has spent her career on one problem rendered two ways: first making data legible to people, then making increasingly autonomous software steerable by them. As of 2026 she is a partner at Sutter Hill Ventures and product lead for Intent, Augment Code's agent-orchestration workspace. Before that she spent two stints at GitHub Next, GitHub's small R&D team, where her prototypes helped define the Copilot era — and where she wrote "Why Chatbots Are Not the Future," a 2023 essay that became a reference point in the argument over how people should actually work with large language models.
From neuroscience to the browser
Both of her parents were programmers, which is precisely what pushed her away from computers: an office job sounded boring, and she planned to become something like a prison psychologist. She studied neuroscience and psychology at Trinity College in Hartford, then moved to Texas to work as a research assistant and lab manager in a neuroimaging lab at the University of Texas at Austin — a co-authored paper on neurogenesis followed. Hanging around graduate students long enough to see the PhD track up close, she opted out, taught herself to code through Codecademy and by rebuilding her personal site, and landed a frontend job at an analytics startup.
She spent roughly a decade in that world — most visibly four years (April 2016 to March 2020) as a senior UX engineer at Parse.ly, the content-analytics platform — designing dashboards, visualizing attention data, and moving between design, dataviz, and the query layer as small startups require.
The dataviz years
Alongside the day job she built a public teaching practice: blog posts on D3, React, SVG, and the CSS cascade; "code sketches" for learning one concept at a time; and in 2019 Fullstack D3 and Data Visualization, a book-and-course whose chapters she drafted code-first. The same year she built the interactive charts for the 2019 State of JavaScript survey — the overview visualization was admired widely enough to put her on JS Party. In March 2020 she joined The Pudding as a journalist-engineer. Her single bylined piece there (November 2020) is characteristic: playable mini crosswords, built with Russell Samora and Michelle Pera-McGhee, that let readers feel the racial and gender breakdown of puzzle clues rather than just read the percentages.
GitHub Next and the Copilot era
She joined GitHub Next in 2021, shortly before GitHub Copilot's public preview, and helped with the launch and early user reactions. The team's method matched hers: small, timeboxed, public prototypes. Flat Data (May 2021, with Idan Gazit, Matt Rothenberg, and Irene Alvarado) versioned working datasets in git. "Visualizing a Codebase" (August 2021, solo) drew circle-packed repository "fingerprints" and shipped a GitHub Action to keep them current. GitHub Blocks (announced November 2022) let repos embed custom interactive views. Then came the LLM run: Code Brushes (January 2023) painted code with ML transformations inside Copilot Labs; Copilot for Docs (March 2023) answered questions from a project's own documentation, with citations and sliders to tune responses to the reader; Code Atlas (June 2023) pushed models to emit inspectable workflows of code blocks — fluid reasoning bound to rigid structure. Most of these previews were sunset in December 2023, which was always the deal: GitHub Next ships explorations, not promises.
Beyond chatbots
In May 2023 she published "Why Chatbots Are Not the Future," arguing that chat windows are a thin wrapper on a strange new capability: text inputs carry no affordances, prompts are unstructured piles of context, responses are isolated without a working buffer, and the loop forces constant switching between implementing and evaluating. Her prescription — bake context into controls, keep humans in charge, build "more tools and fewer operated machines" — circulated widely after Simon Willison and others amplified it. The nuance often missed: a companion essay, "LLMs as a tool for thought," argues the same chatbots make fine brainstorming partners — they hold pinned ideas outside working memory and match the thinker's cadence. She is not anti-chatbot; she is anti-default.
Adept, the return, and Intent
In July 2023 she left GitHub Next for Adept's design team, working on interfaces for AI that reads screens and operates software; that October, at AI Engineer Summit in San Francisco, she gave "Climbing the Ladder of Abstraction" — spreadsheets and zoomable maps as precedents for augmentation through many small automations. By January 2024 she was back at GitHub Next as principal research engineer, a stint her LinkedIn ends in February 2025. She then joined Sutter Hill Ventures — first listed as designer in residence, then partner from July 2025 — and took product lead for Intent at Sutter Hill-backed Augment Code. Intent, which opened in public beta on February 10, 2026, is her "what comes after the IDE" answer: a workspace per task (each backed by an isolated git worktree), a living spec humans approve, and coordinator, implementor, and verifier agents — including third-party agents like Claude Code and Codex — doing bounded work inside it. On the 2026 interview circuit she argues the arc ran autocomplete → chat → CLI → back to UI, and that agents make the first seventy percent of a project cheap — leaving the last thirty, polish and judgment, as the new bottleneck.
What the record does not settle
The seams are date-shaped. Her GitHub Next start, the 2021–2022 title gap, the exact designer-in-residence-to-partner transition, and any uncredited Pudding engineering all sit between a self-described "shortly before Copilot" and LinkedIn's tidy rows. She has no Wikidata or Wikipedia entry; this index leans on her own site, employer records, and interviews, and her homepage appears not to have caught up with her current job.
This index was compiled from public sources and does not imply the subject's endorsement. Citations live in the packet's source catalog.