Rob Haisfield

Rob Haisfield is a WebSim cofounder whose work spans generative browser experiences, tools for thought, and behavioral product strategy. In his WebSim presentation published in April 2024, he explained a browser where invented URLs and natural-language requests generate interactive pages on the fly. His earlier research explored how people could structure, query, and synthesize information through programmable interfaces.[4][1][9]

Haisfield credits Karan Malhotra’s WorldSim demonstration with inspiring WebSim and says Malhotra introduced him to his cofounder at a hackathon. He framed the idea as “what if we simulated a world, but like within a browser instead of a CLI, right?” In the presentation published April 27, 2024, he demonstrated slides generated inside WebSim using reveal.js, with other generated pages embedded as iframes. The URL bar accepted ordinary language as well as URL-shaped prompts.[1]

Haisfield encouraged a remix competition after Dylan Field shared a generated Figma interface. He credited a user named Neil with turning the initially nonfunctional interface into a drawing tool: Neil prompted it to use HTML canvas elements and script tags, add drawing controls, and fix disappearing rectangles. Haisfield demonstrated drawing and saving an image, then explained how shareable browser links let people iterate on one another’s creations. He also described the team’s addition of script-tag execution as enabling an early videosynth tool.[1]

For Haisfield, the browser gave language models more ways to express themselves. He experimented with dynamic typography, assigning fonts to words to express their meanings. He also recounted visiting a simulated Wikipedia in a universe where octopuses rather than humans were sapient, exploring what octopus computer interaction might look like. Adding table-view instructions organized the generated material; a “secrets equal revealed” parameter changed the styling and introduced hidden text. These were playful explorations of imagined worlds.[1]

Before WebSim, Haisfield worked on behavioral product strategy and tools for thought. That earlier work supplies context for his proposals about interfaces that help newcomers learn while giving experienced users more expressive building blocks.[6][8][9]

Haisfield studied behavioral economics in college. In his career account, he reports spending three years as an independent consultant specializing in behavioral product strategy and gamification, while also working as a behavioral product strategist for Spark Wave, including work involving GuidedTrack. Across more than a dozen startups, he repeatedly worked on onboarding and retention challenges.[6]

He chose consulting over pursuing a PhD because he wanted applied work: reading research, testing its usefulness in products, and learning from what worked. His view of professional credibility was succinct: “Your output becomes your credentials.” He qualified that decision by acknowledging that a one-year master’s program might have made consulting easier and taught him more.[7]

Drawing on GuidedTrack, Haisfield argued for domain-specific languages as end-user software. High-level functions let people build forms, studies, and stories; loops, conditional logic, variables, and reusable programs extend those capabilities. Indentation expresses relationships, while a structured editor helps newcomers generate code before they know the language. He also valued text’s portability: programs could be edited or reviewed in other tools, although GuidedTrack’s web app remained necessary to run and deploy them.[8]

His concrete example was a multi-select question followed by a question for each selected answer. He reported that building this in Tripetto took three minutes and at least 20 clicks outside typing, even after he knew the procedure; his GuidedTrack version used fewer than 15 lines of code. The comparison illustrated his preference for composable primitives that let users create behavior themselves.[8]

Haisfield reports receiving a Graph Foundation research grant with Joel Chan and Brendan Langen during August 2021–March 2022. The team surveyed prior work and interviewed 40 users and builders of tools for thought. Their collaborative Scaling Synthesis notebook, initially published in May 2022, investigates data structures and interfaces intended to support synthesis and innovation in a decentralized discourse graph. Its associative hypertext format presents evolving ideas rather than a finished system.[6][5]

He began working for Tana in August 2022. In his biography, he says he moved to an advisory role in mid December because of time-zone coordination challenges.[6]

By April 2023, Haisfield was presenting AI-related interface work: the AI UX meetup report, published April 26, lists his presentation titled “AI Zettelkasten: With prompts on your notes taken.”[2]

In his discourse-graph DSL proposal, last updated June 7, 2023, Haisfield proposed operations such as query, sort, group, and view-as-map, alongside custom functions. The interface could use spreadsheets, canvases, or block editors, with graphical controls serving as scaffolding. He suggested placing formulas or queries alongside displayed numbers so readers could verify how values were produced. Expressivity and efficiency remained hypotheses, and the graph’s syntax was explicitly unresolved.[9]