Sources#
Summary#
Thariq Shihipar's answer to the obvious objection against HTML plans: HTML isn't as easily editable as markdown, so what do you do when you want to change a complex table or a set of rules buried in an HTML artifact? You ask Claude to build a temporary, disposable user interface — a micro-app — just for editing that one specific part of the plan. Instead of editing the artifact by hand or grinding through chat, you generate the ideal tool for the edit and use it once.
"It's like this is not even personal software. This is like sub… it's like micro-software on top of micro-software."
The worked example#
Thariq's HTML implementation plan contained a set of decision rules for turning different CSV data types into visualizations. He felt the rules were "a bit arbitrary" and wanted to refine them — a tedious thing to do in prose or in raw HTML. The prompt:
I want to create an edible HTML artifact to help me define the decision rules...
I don't like the ones we have right now. Make this a custom UI that helps me with
structure but gives me flexibility. Design the ideal interface for this problem.
("edible" = editable — another harmless typo the model handled.) Claude returned a dedicated webpage: input fields, dropdowns, add/remove buttons, and a button to copy the final configuration back to the clipboard as markdown. Thariq edited the rules in a structured, visual way, got the output right, and brought the perfected data back into his main workflow.
Two design notes that make it work#
- "Design the ideal interface for this problem." The human specifies the problem, not the widget set, and lets the model pick the interface — constraint plus latitude, the same prompting move as in HTML as the New Markdown.
- Copy-back-to-markdown. The micro-app is a transient editing surface over durable text. This patches HTML's worst weakness (diffing/versioning) — the source of truth round-trips through markdown, while the rich UI exists only for the duration of the edit.
Why this is now rational: the abundance mindset#
Spinning up a single-use application to make one editing decision would be absurd under scarcity economics. It is rational under the Compute Allocator's abundance mindset: generation is cheap enough that throwaway tools are a sensible investment in your own process. "We can now afford to generate these throwaway tools to make our own process more efficient and enjoyable." This is Printing Press Software Democratization at the scale of a single task — software so cheap to produce that it becomes a disposable convenience, not a project.
Connections#
- Thariq Shihipar — originator of the micro-app workflow
- Claude Code — the product used to generate the throwaway UIs on demand
- HTML as the New Markdown — the plans these micro-apps edit; micro-apps answer its editability objection
- Compute Allocator — the role/economics that make disposable tooling rational
- Living Design System — the durable counterpart: a micro-app you keep and reuse as context
- Printing Press Software Democratization — software cheap enough to be disposable
- Harness Shrinkage as Models Improve — micro-apps are human-facing scaffolding (built per-task for the human), distinct from the model-facing harness that shrinks
- Software 3.0 — micro-apps are Software-3.0-native: Karpathy's MenuGen "that app shouldn't exist" is the same logic — apps cheap enough to barely exist
- Building Is Cheap, Arguing Is Expensive — throwaway built artifacts as the unit of decision; cheap building makes the disposable UI rational
Open questions#
- Where's the line between a disposable micro-app and tool sprawl? If every edit spawns a bespoke UI, does the workflow fragment?
- Does the copy-back-to-markdown round-trip generalize beyond config-shaped data (rules, tables) to richer artifacts?
- Could these micro-apps be templated/reused rather than regenerated — and at what point does that defeat the "disposable" framing and turn into durable tooling?
Derived#
- Does the Human-Facing Harness (HTML Artifacts) Hit Its Own Bloat Ceiling? — micro-apps are the decomposition escape valve that softens the bloat ceiling, but tool sprawl is the axis the ceiling relocates onto
Sources#
Cited by 11
- Building Is Cheap, Arguing Is Expensive
"In technical debate, code wins": generate three PRs vs whiteboard; prototype over design doc; reduce design docs
- Claude Code
Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React; CLI/desktop/web/mobile/IDE sur…
- Compute Allocator
The human's evolving role: deciding what's worth spending compute on; ~1% of generated tokens ship, 99% is scaffolding…
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- HTML as the New Markdown
Thariq Shihipar's thesis: as models improve, thousand-line markdown plans overwhelm the *human*; HTML artifacts (visual…
- Does the Human-Facing Harness (HTML Artifacts) Hit Its Own Bloat Ceiling?
Yes — HTML raises and reshapes the human-attention ceiling but can't remove it; bloat relocates from document-length to…
- Living Design System
`design_system.html` extracted from repos as a portable, human- and machine-readable source of truth; component playgro…
- Narrow Wedge into a Legacy Market
Disrupt without being feature-complete: be the best for a narrow customer profile (tech cos outgrowing QuickBooks); Goo…
- Printing Press Software Democratization
Boris Cherny's analogy: 1400s literacy expansion → AI software-writing expansion; domain knowledge displaces coding ski…
- Software 3.0
Karpathy's taxonomy: 1.0 code, 2.0 weights, 3.0 prompting; LLM as programmable interpreter; MenuGen "shouldn't exist";…
- Thariq Shihipar
Engineer on the Claude Code team at Anthropic; "HTML is the new markdown" and "compute allocator" framings; three HTML-…
Related articles
- Compute Allocator
The human's evolving role: deciding what's worth spending compute on; ~1% of generated tokens ship, 99% is scaffolding…
- Boris Cherny
Creator of Claude Code at Anthropic; phone-driven workflow with hundreds of agents; primary advocate of `/loop` primiti…
- HTML as the New Markdown
Thariq Shihipar's thesis: as models improve, thousand-line markdown plans overwhelm the *human*; HTML artifacts (visual…
- Thariq Shihipar
Engineer on the Claude Code team at Anthropic; "HTML is the new markdown" and "compute allocator" framings; three HTML-…
- Claude Code
Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React; CLI/desktop/web/mobile/IDE sur…
