Hi HN,

I’m a solo founder working on SentienceAPI, a perception & execution layer that helps LLM agents act reliably on real websites.

LLMs are good at planning steps, but they fail a lot when actually interacting with the web. Vision-only agents are expensive and unstable, and DOM-based automation breaks easily on modern pages with overlays, dynamic layouts, and lots of noise.

My approach is semantic geometry-based visual grounding.

Instead of giving the model raw HTML (huge context) or a screenshot (imprecise) and asking it to guess, the API first reduces a webpage into a small, grounded action space made only of elements that are actually visible and interactable. Each element includes geometry plus lightweight visual cues, so the model can decide what to do without guessing.

I built a reference app called MotionDocs on top of this. The demo below shows the system navigating Amazon Best Sellers, opening a product, and clicking “Add to cart” using grounded coordinates (no scripted clicks).

Demo video (Add to Cart): [https://youtu.be/1DlIeHvhOg4](https://youtu.be/1DlIeHvhOg4)

How the agent sees the page (map mode wireframe): [https://sentience-screenshots.sfo3.cdn.digitaloceanspaces.co...](https://sentience-screenshots.sfo3.cdn.digitaloceanspaces.co...)

This wireframe shows the reduced action space surfaced to the LLM. Each box corresponds to a visible, interactable element.

Code excerpt (simplified):

``` from sentienceapi_sdk import SentienceApiClient from motiondocs import generate_video

video = generate_video( url="https://www.amazon.com/gp/bestsellers/", instructions="Open a product and add it to cart", sentience_client=SentienceApiClient(api_key="your-api-key-here") )

video.save("demo.mp4") ```

How it works (high level):

The execution layer treats the browser as a black box and exposes three modes:

* Map: identify interactable elements with geometry and visual cues * Visual: align geometry with screenshots for grounding * Read: extract clean, LLM-ready text

The key insight is visual cues, especially a simple is_primary signal. Humans don’t read every pixel — we scan for visual hierarchy. Encoding that directly lets the agent prioritize the right actions without processing raw pixels or noisy DOM.

Why this matters:

* smaller action space → fewer hallucinations * deterministic geometry → reproducible execution * cheaper than vision-only approaches

TL;DR: I’m building a semantic geometry grounding layer that turns web pages into a compact, visually grounded action space for LLM agents. It gives the model a cheat sheet instead of asking it to solve a vision puzzle.

This is early work, not launched yet. I’d love feedback or skepticism, especially from people building agents, RPA, QA automation, or dev tools.

— Tony W

Microsoft releases native Windows feature for performance boost to Servers

Google "6-7" easter egg

"Peter introduced savo " Are they two separate individuals?

California minimum wage set to rise in 2026

Artificial sweetener found in Diet Coke and gum could trigger heart brain damage

Google lobs lawsuit at search result scraping firm SerpApi

Part 4: The Emergence of Matter

Texas Attor­ney Gen­er­al Sues Five Major TV Com­pa­nies for Spy­ing on Texans

Show HN: Semantic geometry visual grounding for AI web agents (Amazon demo)

Sequoia partner spreads debunked Brown shooting theory, testing new leadership

Approaching 50 Years of String Theory

Show HN: aic – CLI to fetch changelogs from AI coding assistants

The ______ Files

Clinton Unloads on Trump's Plot to 'Scapegoat' Him for Epstein

Is North Korea's 'princess' walking a path toward succession?

Ask HN: Anyone here bringing "AI" to small-mid farmers?

The 'untranslatable' emotions you never knew you had (2017)

Ask HN: Is GitHub becoming more and more unstable?

Trove of Epstein Files Includes New Photos and Court Records

Generating DSL Code with LLMs

To grow, we must forget but now AI remembers everything

Why homeownership in California isn't nearly the financial slam dunk it once was

US backs Bayer–Monsanto with glyphosate decision

How Lord of the Flies was rescued from the reject pile (2024)

AI Stylist for Perfect Fit and Style

In Re Tesla, Inc. Derivative Litigation [Tornetta vs. Musk Appeal]

The Cabbage Patch Kids caused a near-riot in the 1980s

The Abolition of Work – Bob Black (1991)

Cheaper, cleaner energy drives Germany's balcony-solar boom

Ruby 4.0.0 Preview3 Released

Building ScrapeForge in public. Day 1 update