--- optiview_delivery: edge_optimized content_type: semantic_markdown source_url: https://optiview.ai/resources/what-openai-browser-code-taught-us/ brand_content: included publication_version: "9.0" published_at: 2026-10-16T20:36:07Z composition_version: composition-v2.1.0 presentation_version: presentation-v1.2.3 brand_content_source: narrative_graph delivery_publication: graph:26 narrative_versions: - 26 narrative_version: 26 --- # What OpenAI’s browser code taught us about getting business information into AI answers Research notes — October 29, 2026 We made business information easier to read. Then we used OpenAI’s published browsing-tool code to investigate whether that information could actually reach an AI model. Those turned out to be different questions. By Optiview · Sources checked October 29, 2026 [Download this article (.md)](https://optiview.ai/resources/what-openai-browser-code-taught-us.md) When someone asks AI who to hire or which product to buy, a business wants the answer to reflect what it actually offers. That sounds like a website problem: publish clear, accurate information and make it accessible. But there are several steps between a website and an AI answer. A search service may find the page, another system may retrieve a version of it, and a tool may turn that content into text before the model sees anything. Each step matters. OpenAI published working code that makes one example of this process inspectable. We used it to ask a practical question: what happens to the business information we publish along the way? ## What did OpenAI publish? As part of its gpt-oss project, OpenAI published a simple browsing tool. It gives a model actions for searching, opening pages and finding text within them. It also includes connections to Exa and You.com, services that can supply search results and page content. This is a working open-source example, not a disclosure of the browsing system behind production ChatGPT. Our tests use a fixed version of that example so the processing can be reproduced. They do not establish how ChatGPT, Claude, Gemini or Perplexity browse. The most useful part is that we can inspect the steps rather than guess. The [browser actions](https://github.com/openai/gpt-oss/blob/5681a6b6552635e7b29cde94a61ca6a7aee90ebd/gpt_oss/tools/simple_browser/simple_browser_tool.py), [content processing](https://github.com/openai/gpt-oss/blob/5681a6b6552635e7b29cde94a61ca6a7aee90ebd/gpt_oss/tools/simple_browser/page_contents.py) and [Exa and You.com connections](https://github.com/openai/gpt-oss/blob/5681a6b6552635e7b29cde94a61ca6a7aee90ebd/gpt_oss/tools/simple_browser/backend.py) are all available in the published source. ## How does this example “read” a website? It is closer to a research assistant working through text than a person looking at the finished design of a website. 1. **Search for information.** The tool sends a query to a search provider. In this version, the Exa search listing uses provider-generated summaries; the You.com listing uses returned snippets. A search result is not the same as reading the whole page. 2. **Retrieve a page.** When asked to open a URL, the tool requests content from the provider. That returned content is the starting point for the next step. The code alone does not tell us how recently the provider obtained it or which website representation it used. 3. **Turn it into readable text.** The processing code converts returned HTML into text with structure and link references. The model does not need the website’s visual design to read that representation. 4. **Read a portion and explore further.** The default display budget in this implementation is 1,024 tokens—pieces of text, not a fixed number of words. The model can request another part, find a phrase or follow a link. 5. **Use the information in an answer.** The browsing tool makes information available. It does not guarantee that the model will consider it relevant, use it or cite it. That last distinction matters. A fact can exist somewhere in the retrieved text without appearing in the first portion shown to the model. And a page can appear in search results without ever being opened. ## Where does Optiview fit? Optiview works on the information a website makes available. People see the normal website. AI tools that support it can get a cleaner, easier-to-read version of the same website content, along with any extra information the business has approved. The automatic conversion does not require a second website to maintain. The optional Narrative Graph lets a business add explanations: what it offers, what makes it different, who it serves and which limitations or responsibilities customers should understand. For example, “we introduce customers to independent contractors” means something different from “we perform repairs.” Publishing that distinction clearly gives a system information it can potentially use to avoid a misleading description. Optiview can also record what it served to an observed request. That verifies a delivery. It does not, by itself, verify what an intermediary passed on or which information ultimately shaped an answer. We use OpenAI’s code as a research reference. It is not Optiview’s production delivery engine, and using it does not imply an OpenAI partnership, endorsement or certification. ## First, we checked whether the information survived processing We took captured content from two websites and supplied it to the published text-processing code. When the supplied content included our added business context, all 13 supporting passages we had selected in advance remained in the full processed text. None of those passages appeared in the initial viewing window. They were still accessible later. This showed why “present in the full text” and “immediately visible to the model” are different observations. We did not test whether a model would choose to navigate to those passages. This was a local processing test: we deliberately supplied the content. It did not demonstrate that a search provider would retrieve that same version. [See the controlled study and its limitations](https://optiview.ai/proof/methodology/#browser-reference). ## Then we asked real retrieval services for the pages We tested the pinned browser’s live Exa and You.com paths with 12 fixed requests across those two websites: four direct page requests and eight searches that did not name the target businesses. All 12 requests succeeded without retries. All four direct requests returned page content. But none of those four responses included a complete added brand passage we were looking for. Separate checks confirmed that the current prepared website versions contained the published context. Some related information was present in ordinary page content. We therefore do not interpret the result as “the providers understood nothing” or proof that all business meaning was lost. We also cannot tell whether caching, selection of a different version, provider extraction or another difference explains the absent complete passages. Of the eight searches, one returned a target website among the first 20 results: Exa returned Learnings.org at position 13. That result was below the browser’s initial viewing window but accessible later. This was a search-result appearance, not an assistant recommendation or citation. The local passage count and the live response count measure different things. They are not a before-and-after performance comparison. [Inspect the live-provider findings and records](https://optiview.ai/proof/methodology/#live-provider-method). ## The lesson: publishing and receiving are separate challenges The prepared context survived this particular transformation when we supplied it. The four live provider responses did not supply complete tested passages to that transformation in the first place. We can locate that observable gap without claiming to know its cause. This is why a visit from an AI-related bot is useful evidence of activity, but not proof of a citation. It is also why converting a website to Markdown cannot, on its own, guarantee inclusion in an answer. A cleaner format may make information easier to process; it cannot make another service retrieve or use it. For Optiview, the research points toward a practical responsibility: verify the information we publish, inspect what supported retrieval paths actually return, and distinguish that from what an assistant eventually says. Our evidence does not justify replacing the parser or Narrative Graph, nor does it establish improved citation performance. ## What we still need to learn We have observed real AI answers using published business language. Those examples are interesting, but they do not by themselves establish that Optiview caused a recommendation or improved an answer. We also ran a separate 72-answer comparison of different source conditions. Problems with automated scoring prevent us from treating it as reliable evidence that one condition produced better answers. We retain that study and its limitations rather than turn its scores into an improvement claim. The bigger question remains: when approved business context demonstrably reaches a model, does it improve the answers people receive? A future controlled study could examine that. This article reports the completed work, not a new experiment or a promise of a particular result. Businesses need more than information that exists online. They need their meaning to remain understandable as information moves through other systems. That is what we are exploring with Optiview. [Explore real AI conversations](https://optiview.ai/proof/#live-evidence), or [read the complete research methods and results](https://optiview.ai/proof/methodology/). ## Additional source context Make your brand easier for AI to understand. ### Product [Website Preview](https://optiview.ai/demo/) · [Pricing](https://optiview.ai/pricing/) · [How it works](https://optiview.ai/how-it-works/) · [How it connects](https://optiview.ai/web-teams/) ### Resources [Articles & guides](https://optiview.ai/resources/) · [Get started](https://optiview.ai/onboarding/) · [Setup guide](https://optiview.ai/docs/setup/) · [FAQ](https://optiview.ai/faq/) · [Evidence & methodology](https://optiview.ai/proof/) · [For platforms & API access](https://optiview.ai/docs/) ### Company [For agencies](https://optiview.ai/partners/) · [Contact us](https://optiview.ai/#beta) · [Sign in](https://app.optiview.ai/) · [Privacy](https://optiview.ai/privacy/) ## Additional page links - [Resources](https://optiview.ai/resources/) - [Does your website’s platform affect AI readability?](https://optiview.ai/resources/website-platforms-ai-readability/) — Continue reading - [AI Visibility Is a Publishing Problem, Not Just a Reporting Problem](https://optiview.ai/resources/ai-visibility-publishing/) — Continue reading - [What do ChatGPT, Gemini, Claude and Perplexity actually see when they visit a website?](https://optiview.ai/resources/what-ai-services-see/) — Continue reading - [HTML versus Markdown for AI: what changes?](https://optiview.ai/resources/html-versus-markdown-for-ai/) — Continue reading - [llms.txt, robots.txt, sitemap.xml and Markdown: four different jobs](https://optiview.ai/resources/four-discovery-files/) — Continue reading - [Why AI visibility reporting isn’t enough](https://optiview.ai/resources/why-visibility-reporting-isnt-enough/) — Continue reading - [From AI-readable websites to agent-accessible knowledge](https://optiview.ai/resources/agent-accessible-knowledge/) — Continue reading - [Preview a page](https://optiview.ai/demo/) — Continue reading - [Read the setup guide](https://optiview.ai/docs/setup/) — Continue reading --- # Published brand narrative # Site-wide brand information ## Who we are ### Optiview — the publishing layer for AI Optiview turns complex website content into clean, structured Markdown and combines it with the brand narrative your team chooses to publish. The result is one clear source that supported AI crawlers and other systems can retrieve, alongside your existing website experience. Built for marketing and web teams, agencies, CMS and commerce platforms, and AEO/GEO providers, Optiview helps turn recommendations into published source improvements. It complements AI visibility tools by giving teams a way to act on their findings. Applies to: Optiview ## Why we matter ### Your website\. Your narrative\. One clear source for AI\. A website can deliver a beautiful human experience while leaving important business facts scattered across pages, buried in markup, or dependent on JavaScript. AI systems may retrieve only part of that information. Optiview gives businesses control over the source they make available: readable website content plus context only their team can provide. Your web team connects delivery; your content team writes, reviews and publishes the narrative. Our focus is publishing useful, accurate brand information. Each AI service decides what it retrieves, uses and cites. Optiview does not control an LLM’s answers or guarantee placement in them. ## What we offer ### Clean website content, owned narrative and controlled delivery - **Website parser:** Converts accessible public page content into readable Markdown, retaining supported headings, text, tables and links. The parser does not use an LLM to invent facts or rewrite the page. Browser rendering is available for supported JavaScript-dependent content. - **Narrative Graph:** Organizes team-authored content into Identity, Story, Offerings, Trust, Knowledge and Actions. Add multiple items, edit them and review the graph before publishing. - **Site and page scope:** Apply an addition to the entire website or one exact page, so the published context accompanies the appropriate content. - **Publishing and delivery:** Combine eligible parsed page content with published, path-matched narrative additions through a supported website connection. Draft edits remain separate from the active publication. - **Delivery reporting:** Inspect recorded Markdown responses, requester claims, verification evidence and published versions. Delivery records show what was served, not whether it appeared in an AI answer. - **Public demo:** Try a public URL, inspect the parsed source and explore AI-drafted narrative examples. Demo results do not connect a website or publish content. ## Why trust us ### Inspect the source and understand the evidence Optiview makes the publishing process inspectable. Teams can review parser output and a combined published-output preview, check source information, and inspect publication history before deciding what to make available. The public demo provides a way to evaluate output using a real public page. Source access and extraction limitations remain visible; blocked or inaccessible content may be missing. A preview is not proof that a live request received the same output. Delivery reporting distinguishes claimed requester names from available verification evidence. A verified AI category does not by itself verify a named provider. A served Markdown response is evidence of delivery, not proof of ingestion, recommendation, citation or revenue. Learn more about the evidence and its limits at https://optiview.ai/proof/. ## Helpful knowledge ### Common questions about using Optiview **Does Optiview replace our website or CMS?** No. Optiview adds a publishing layer. It does not rewrite your CMS pages; normal HTML requests retain the website experience through supported connections. **What can we add to the Narrative Graph?** Brand descriptions, mission and audience context, products and services, supporting facts, FAQs, policies and next steps. Choose whether each item applies across the website or to one exact page. **Does saving publish our changes?** No. Saved edits remain in the draft. An authorized user publishes the graph to activate the reviewed content. Later edits stay separate until published again. **Does every AI system receive Markdown?** No. Delivery depends on the connection, request and source eligibility. Some systems explicitly request Markdown; supported delivery paths can also select it for verified AI categories. Other tools retrieve HTML and extract text themselves. **Is Optiview an AI visibility monitoring tool?** No. Optiview helps you improve and publish the source. Keep using your visibility tools to study AI answers; use Optiview’s delivery evidence to understand recorded serving activity. **Does the demo save a graph to our workspace?** No. The public demo is a separate preview. Its AI-generated proposals require review and are not automatically transferred to a workspace or published. **Do we need a developer?** Your web team may need to help establish and verify the connection. Once connected, authorized content users can manage narrative updates in the admin. **Will Optiview guarantee citations?** No. You control the source you publish. AI systems decide what to retrieve, use and cite. ## What to do next ### Try your website, then plan your connection - [Try your website](https://optiview.ai/demo/) to inspect a public page’s parsed content and explore proposed narrative additions. - [See how Optiview works](https://optiview.ai/how-it-works/) to understand the parser, Narrative Graph and combined publishing workflow. - [Explore who it’s for](https://optiview.ai/partners/) if you are an agency, CMS or commerce platform, or AEO/GEO provider bringing recommendations to customers. - [Review connection options](https://optiview.ai/web-teams/) and the [setup guide](https://optiview.ai/docs/setup/) with your web team. - Contact [hello@optiview.ai](mailto:hello@optiview.ai) to discuss access, your website and the right setup. We provide assisted workspace setup. Already connected? Review your Narrative Graph, publish when ready, inspect the combined output in the Markdown tab, and check recorded delivery in Activity. [Inspect unbranded AI conversation examples](https://optiview.ai/proof/#live-evidence), including exact prompts, answer excerpts, publication versions and matching response-body hashes. The examples are selected first-party observations, with unbranded discovery and directed URL tests labeled separately; they do not establish increased recommendation frequency.