Source: raw/I_let_Codex_control_my_browser_so_I_don_t_have_to.md Creator: Claire Vo (ChatPRD founder) — How I AI URL: https://www.youtube.com/watch?v=lk63Sl-LRKE Platform: YouTube (yt-podcast feed how-i-ai)

Browser and computer use is a capability the wiki documents at the harness level (Computer Use, browser-harness) but has thin coverage of as a daily working habit. This episode is the habit: Vo runs Codex’s browser and computer control “every minute of every day” and walks through four concrete use cases — agentic QA, persona-driven UX research, inbox triage, and shopping — plus the invocation model and the one prompting rule she says matters most. Her verdict on tool choice is blunt and worth recording as an operator data point rather than a benchmark: Codex (the ChatGPT desktop app) is “just the best at controlling my computer.”

Key Takeaways

  • Codex exposes three distinct control surfaces, and picking the right one matters. @browser opens Codex’s own side browser window; @chrome drives your Chrome through the extension (so it inherits your logged-in sessions); @computer controls the whole machine — mouse, keyboard, any app. Prerequisites are the desktop app plus the official Chrome extension. Vo reaches for @browser when the target is already open on localhost and @chrome when the task needs her real logged-in state.
  • Under-prompt frontier models on browser tasks. Her explicit tip: “QA the onboarding flow” produces a better result than “QA these 25 things in the onboarding flow.” Over-specification caps the agent at your checklist; letting it plan gets you coverage you wouldn’t have written down. ^[Vo’s stated finding from repeated use, not a controlled comparison]
  • Agentic QA’s real advantage is exhaustiveness, not speed. Humans QA the happy path — fill every required field, click next, don’t get tired. The agent tests failure states, error states, both signup paths, viewport resizes, accessibility, overflow, and touch targets. On her own long-shipped onboarding flow it surfaced 11 issues including one high-severity navigation blocker — a required field with no validation code, invisible to her because she always clicked the required buttons.
  • Make the QA output a spreadsheet, and the spreadsheet becomes the fix queue. The prompt shape is “test [flow] for usability and mobile responsiveness, screenshot as you go, and make a Google spreadsheet of any issues you identify.” The resulting sheet carries repro steps, viewport, severity, remediation, and the embedded screenshots — everything a fixing subagent needs, and a tracker she can compare against after-fix screenshots.
  • Persona-driven UX research is the more unusual use case. Rather than “you are a genius PM” in a text prompt, put the persona in the browser: have the agent walk the real product as a PM, then an engineer, then a team lead, and write a research-style critique with friction points, moments of delight, and improvements. It surfaced a genuine structural gap in ChatPRD (a document created in one thread can’t be referenced from another) that the team knew about abstractly but hadn’t felt from the user’s side.
  • Browser use is the channel for surfaces with no API or MCP. LinkedIn has no official API or MCP, so Vo drives it through the browser: “go through my recent unread messages and reply to any that are super critical for ChatPRD or the podcast; leave me notes on the others about how I should reply.”
  • Calibrate effort down for mechanical browser work. She noted she was running LinkedIn triage on GPT-5.6 Sol at high effort and that Terra at medium effort would likely be both sufficient and faster — the same model-vs-effort discipline the wiki documents for coding applies to browser agents, where the bottleneck is usually page latency rather than reasoning.
  • Bot detection is the hard wall, and the human becomes the verification layer. A retail shopping run tripped an are-you-a-bot check. Her framing: “sometimes we put AI to use for us, and then sometimes we need AI to put us to use” — expect to be the hands for the agent at the CAPTCHA step rather than expecting the run to be fully unattended.
  • Computer use can drive your phone via macOS iPhone mirroring. Her worked example: out of state, needing to reach Macs at home, she had Codex use iPhone mirroring to open her Wi-Fi app, change router settings to open firewall ports, SSH into the machines, and close the ports again when done. This is the sharpest capability claim in the episode — the phone becomes just another window computer use can operate.
  • Computer use is also the fallback when an MCP or plugin path fails. When an integration hits an access error, “please just open my browser and do it” often works — she uses this routinely for creating Google Docs and Sheets.
  • The standing cost is latency. Browser use is slow; her habit is to launch it and work on something else in parallel rather than watch it.

The four prompt shapes

Use caseSurfacePrompt shapeOutput
Agentic QA@browser (localhost)“Test the [flow] for usability and mobile responsiveness. Screenshot as you go and make a Google spreadsheet of any issues you identify.”Prioritized issue sheet w/ repro steps, viewport, screenshots — feeds a subagent fix loop
Persona UX research@browser”Go through [product] as three personas: [A], [B], [C]. I’m giving you no more instructions than that. Then create a doc with a research-style critique — friction points, places of delight, improvements for each persona.”Research doc grounded in an actual session, not a synthetic-user simulation
Inbox triage (no API)@chrome (logged-in session)“Go through my recent unread messages and reply to any that are critical for [X]. Leave me notes on the others about how I should reply. If it’s just a friendly thank-you, reply in a friendly way.”Sent replies + a decision queue for the rest
Web chores@chrome”[Constraints]. Put N items in my cart that I can pick from and buy — don’t check out.”A staged, reviewable end state rather than a completed transaction

The pattern across all four: give the agent a goal and a deliverable artifact, not a step list — and for anything consequential, stop it one step short of the irreversible action (cart not checkout, draft not send).

Try It

  1. Run the QA prompt against a flow you’ve shipped and stopped looking at. The bug Vo found existed precisely because she had tested it a hundred times as a human on the happy path.
  2. A/B your own prompting. Run one flow with a 25-item checklist and one with “QA the onboarding flow,” and compare issue counts. Her under-prompting claim is cheap to falsify on your own product.
  3. Pick the right @ surface deliberately. If the task needs your logins, @chrome; if it’s localhost or a throwaway session, @browser; only reach for @computer when the target isn’t a web page.
  4. Wire the spreadsheet into a fix loop. Have the QA run emit the sheet, then dispatch subagents against rows and write status back — that’s the loop the artifact is designed for.
  5. Drop an effort tier on mechanical runs. Page loads dominate the wall clock on browser work; see Model vs. Effort for the general heuristic.
  6. Stage, don’t commit. Add “don’t check out / don’t send / stop before the final step” to any browser-use prompt that touches money or someone else’s inbox.

Open Questions

  • No head-to-head data behind “Codex is best at controlling my computer.” It’s a practitioner preference stated without a comparison run against Claude’s Computer Use or the AI-native browsers (Comet, Atlas) she mentions but doesn’t test.
  • Bot-detection failure rate is unquantified — one observed trip on one retailer. How often browser use stalls on anti-bot checks across common sites isn’t established here; the wiki’s CloakBrowser coverage is the adjacent data point on that problem.
  • Security and privacy get raised and not answered. Vo explicitly invites audience questions about both at the end. Driving a logged-in Chrome session and a mirrored phone is a wide blast radius, and this episode offers no guardrail practice.
  • Does the persona-research output hold up against real user research? She flags herself as “not a big synthetic user girl” — the critique it produced was directionally useful, but nothing validates it against findings from actual users.

Field corroboration — Berman’s browser-use applications (2026-08-07 addition)

Source: raw/Master_Codex_with_these_15_Tips.md — Matthew Berman, “Master Codex with these 15 Tips,” auto-caption transcript. A second creator working the same capability, useful mainly for the breadth of non-coding applications it names.

His worked example: one prompt sent Codex to research studio cameras comparable to his current one and return a spreadsheet — price, delta against the camera he owns, ratings, and decision-relevant specs. Completed in ~9 minutes, output as an Excel file.

The applications list is the transferable part, because it is all consumer/admin work rather than development: getting refunds, negotiating with customer service, going through email and archiving what does not need reading, plus file organization and disk cleanup once computer control is enabled rather than browser-only.

This corroborates the article’s central claim — that browser use is the higher-leverage half of the browser-vs-computer-use pair for most tasks — from an independent creator, and extends the documented use cases past the ones covered here. No timings beyond the 9-minute figure, no failure rate, and no comparison against doing the same work by hand.

Field corroboration — Vo on GPT-6 Astra, plus QA agents at Google and Ramp (2026-09-29 addition)

Sources: raw/GPT-6_Astra_blew_away_every_one_of_my_benchmarks.md (How I AI, Claire Vo, youtube.com/watch?v=AniiF8rOu9c); raw/What_it_takes_to_be_a_top_PM_today_Robby_Stein_Google_Search.md (Lenny’s Podcast, Robby Stein of Google Search, youtube.com/watch?v=sTgM_sbLMNg); raw/What_product_looks_like_when_coding_is_solved_Geoff_Charles_Ramp_CPO.md (Lenny’s Podcast, Geoff Charles of Ramp, youtube.com/watch?v=ZG8Mf3P9xzI). Auto-caption transcripts; company figures are self-reported.

Vo, two months on, with GPT-6 Astra in Codex. Her new examples are node-based tools that are slow to operate by hand:

  • CRM workflow builder. Her CRM (caption “Adio”, probably Attio, unconfirmed) routes inbound leads to her or her business partner through a node-based workflow builder. She had spent an hour clicking through it herself. She then asked Codex to “update the workflow in Chrome” so an AI step drafts a custom email from whichever partner the lead is routed to, with their calendar link. Codex added the nodes and prompts itself, and the draft goes to Slack for a human to review and send. That is the “stage, don’t commit” rule from Try It in practice.
  • Flora thumbnails. In Flora, a node-based image and video tool, Astra read the existing thumbnail workflow and built generation nodes for three new photos. Her designer normally does this by hand. It then assembled the thumbnails in Figma, which Vo says computer use had been “quite terrible at” until now.
  • Her usage claim: “I do spend about 90% of my day sitting here watching Astra use my computer and specifically use Chrome.”
  • Long-running browser QA. After shipping a fix for race conditions in ChatPRD’s streaming chat, she asked Astra to “test this in Chrome.” It opened the preview branch, sent chats, inspected the console for errors, and refreshed Chrome to reproduce the race conditions. It “ran for an hour and 45 minutes,” found a couple of navigation issues and fixed some. Her advice: “if you are not using browser use for QA, please do.”

Two company-scale versions of agentic QA:

  • Google Search (Robby Stein). Stein built an internal tool with Antigravity that uses “the product as an agent.” It types a question into Google, asks the AI a follow-up, screenshots the result and grades it against “a really comprehensive rubric.” Two example checks: does a maths answer render correctly in LaTeX, and does a visual topic such as bioluminescence get a visual tray rather than plain text. He says the team “has now invested in a bunch of agents” that find broken or off-spec answers and, increasingly, fix them.
  • Ramp (Geoff Charles). Testo (caption spelling) is “a browser-based QA agent.” It spins up the product “in 100 different combinations based on the actual production data.” It takes instructions such as “pay an invoice, but amortize it” and returns blocking bugs plus design and quality feedback. “Testo in the last 30 days has caught 425 bugs.”

Compared with Vo’s persona QA, both add two things: a written rubric the agent grades against, and test configurations drawn from production data rather than one happy-path account.