Case file — A0CFF498
Launch roast · Launch HN
Launch HN: Intuned (YC S22) – Build and run reliable browser automations as code by fkilaiwi · HN thread
Unsolicited, from public launch info only. Founder? Reply or ask us to take it down.
The idea
“Launch HN: Intuned (YC S22) – Build and run reliable browser automations as code https://intunedhq.com Founder's Show HN post: Hey HN, we're Faisal and Ahmad from Intuned (https://intunedhq.com). We’re building a platform for building, deploying, and maintaining browser automations. Customers primarily use the Intuned AI agent to automate websites that don't expose APIs. Common use-cases include scraping data, pulling reports, and submitting forms. As the website changes, our agent also helps automatically heal the automation. On Intuned, browser automations are created by an AI agent and run as code. Our infra captures the context of every run, allowing our agent to debug and maintain the underlying code - to keep the automations working over time. This way, we’re able to offer the predictability, speed, and cost of code, without the painful parts of writing and maintaining it. Here’s a demo of building a scraper on Intuned: https://youtu.be/ruZP73bK4FU Here’s a demo of using AI to maintain a project: https://youtu.be/e4R4hLdHBro Backstory: we were accepted into YC for a completely different idea. During the batch, because of Faisal's background at UiPath, several batchmates asked us whether RPA tools could fill API gaps in their products by automating websites without APIs. When it was time to pivot, we went back to those founders to dig deeper. (RPA in this context is referring to using UI automation to do complete non-testing tasks) We discovered that the actual hard problem in browser automation is maintenance. Websites change, selectors break, and failures can be painful to reproduce and fix. So in early 2024, we decided to take a crack at this problem with a handful of customers. It needed a fair number of iterations before we landed on our current code-first approach. How it works: Intuned is infra + agent, deeply integrated. On the infrastructure side, Intuned is a managed runtime for browser automation code. Projects are usually Playwright-based TypeScript or Python. Users can write them directly in our online IDE, or hand the work off to the agent. Either way, once deploy Landing page content (Reliable Integrations for Services with No APIs | Intuned): Reliable Integrations for Services with No APIs | Intuned # Build integrations for services with no APIs Intuned Agent builds and runs integrations for services that don't have APIs. It does this by either copying the service's network calls or automating a browser - whatever works best. Its platform then runs these integrations at scale and handles logins, captchas, and monitoring for you. ## Scrapers Create a scraper for any website - e-commerce, government portals, job boards, or anything else. Intuned Agent is the first AI Agent designed to build and maintain Scrapers. Intuned Agent generates playwright code that will be fast, cheap and predictable when it runs. Key Features: Intuned Agent - Generate any scraper from a prompt and schema, fix issues with AI when sites change. Enable self-healing and Intuned will monitor and fix your scrapers when they break. Support for authenticated scrapers - Scrape any website even if its behind authentication. Intuned Infra support helps you manage the authentication state as needed. Use Jobs to schedule the scraper runs and send the data to your backend, can be done via webhooks or another sink. Intuned observability saves records and traces for all your runs so you can debug them when something goes wrong. ## Crawlers Discover, navigate, and collect data across thousands of pages. Build crawlers that follow links, parse sitemaps, and index content - with built-in stealth, scheduling, and scale. Key Features: Use TypeScript or Python, use any framework you want. With Intuned, you can deploy any code you want. Crawl4AI support - First-class support for Crawl4AI - deploy and scale your crawlers without managing infrastructure Stealth mode - Anti-detection, proxies, and captcha solving built in Scheduled jobs & monitoring - Recurring crawls with full logs and session recordings Scalable infrastructure - Control how many machines to run on, we handle the rest ## Back Office Automation Automate internal operations in back-office systems that have no APIs. Intuned runs reliable browser automations for data entry, status checks, form submissions, report downloads, and account updates - with auth, stealth, concurrency, and monitoring built in. Key Features: Back-office workflows - Automate repetitive work inside CRMs, vendor portals, admin panels, and legacy systems. Authenticated operations - Intuned manages login sessions and auth state for systems your team already uses. Read, write, and submit -”
The bull case
Coding models only recently got good enough to write and repair Playwright scripts, and teams running LLM agents on every execution are hitting walls on speed, cost, and nondeterminism. Faisal's UiPath background means he has watched RPA maintenance burden up close. A skeptical investor could back this on the bet that every AI product needs to act on websites with no APIs, and that the vendor who owns "this integration keeps working" captures recurring, sticky infrastructure spend. If the run-trace and fix history compounds into a real healing advantage, it becomes a data flywheel as well as a tool.
The panel
Grounded in live search01 Market
mixedNo market-size or growth figures appeared in the live data, so I can't say whether the category is growing or saturated. The only competitor name that appears verbatim is Reworkd, which the founder calls "the closest one." The snippet gives no funding, review, or activity signals for it. The founder also says "many of the companies mentioned" focus on powering agents via APIs or on runtime AI, but those names aren't in the data, so I'm leaving them out. The live data shows no prior launches of similar ideas, only Intuned's own Show HN (117 points, 58 comments), which is modest validation, not proof of demand. The most serious risk is differentiation in a crowded space. Runtime AI browser agents and scraping platforms all pitch "reliable automation," and the "maintenance as code" wedge is hard to explain and sell. The strongest advantage is that the wedge targets a real, repeated pain: selectors breaking when sites change, which the founders found by talking to YC batchmates.
02 Tech
mixedThe hardest problem is self-healing: an agent patching Playwright code after a site changes, then proving the fix is correct. The dangerous failure is a scraper that still runs but silently returns wrong fields, and schema validation only catches part of that. Anti-bot stealth and captcha solving is a second arms race. Faisal's UiPath background and the year of iteration on maintenance suggest they understand the problem. Generating deterministic code instead of running an LLM on every execution is the right call. It gives fast, cheap, predictable runs. Bundling stealth, proxies and captchas is the build-vs-buy risk, since vendors move quickly there. The moat is thin. Playwright and the LLMs are commodity. The only real asset is the accumulated run traces and fix history across customers' sites.
03 Finance
mixedPricing isn't in the material, so this is built on inference. Intuned appears to sell usage-based infrastructure plus agent credits to developer teams and startups that need integrations with sites that have no API. A Show HN (117 points) gives cheap developer-led top-of-funnel, but the back-office and enterprise use cases need sales-assisted CAC, and no conversion or revenue data was given. What breaks first at scale is gross margin. Each run carries browser machine time, proxies, and captcha solving. Each self-heal carries LLM tokens, and anti-bot sites will trigger heals more often. If pricing is flat per project, the worst-behaved sites will erode margin. Churn is also a risk, because a finished scraper project is easy to cancel or move in-house. In its favor: running generated code instead of an LLM on every run keeps the marginal cost per run low, which is a real cost advantage over agent-at-runtime competitors.
04 Timing
strongWell-timed, but not early. Coding models are now good enough to write and repair Playwright scripts, which was the missing piece when RPA's maintenance burden killed the last wave. Intuned's "AI at build time, code at runtime" bet fits current pain, since teams running LLM agents on every page load are finding them slow, costly, and nondeterministic. Macro trend: Demand for web data and actions from AI agents is growing, while sites are tightening anti-bot defenses. Stealth and captcha handling will keep escalating, and that cost may decide who survives. Window: Open but narrowing. Browser-agent platforms and the closest competitor the founders named, Reworkd, can converge on code generation with self-healing. Favoring factor: Code-generation reliability has crossed a usable threshold.
Competitors found during analysis
Live dataReworkd
Founder's closest comparable
Risks to manage
The positioning is a three-headed product with no defined buyer
The landing page pitches scrapers, crawlers, and back-office automation. The target market isn't specified, and the Market Agent flagged that the "maintenance as code" wedge is hard to explain and sell in a crowded field. A developer scraping job boards and an ops team updating a vendor portal buy differently, have different budgets, and need different security assurances. Chasing all three early means a diluted message and a sales motion that fits none of them.
Gross margin is exposed to the worst-behaved sites
Each run carries browser machine time, proxies, and captcha solving, and each heal carries LLM tokens. Anti-bot-heavy sites trigger more heals, so your most expensive customers are the ones whose sites fight back hardest. Under flat project pricing, those customers erode margin. A finished scraper is also easy to cancel or pull in-house, so retention is unproven.
The moat is thin and the window is narrowing
Playwright and the underlying models are commodity. The only real asset is accumulated run traces and fix history across customer sites, and that flywheel is unproven. Reworkd, which the founder names as the closest comparable (its current activity is unknown from the data), and other browser-agent platforms can converge on code generation with self-healing. Bundling stealth and captcha solving also means competing with specialist vendors that move fast.
Blind spot
Back-office automation means holding customers' credentials for third-party systems and logging in as them, on sites whose terms of service often prohibit automated access. The first serious buyer will ask where the credentials live, who can see session recordings (which may contain sensitive data), and who is liable when an automated form submission goes wrong. Your observability feature, which records every run, is also a data-retention liability. Founders tend to treat this as a later compliance checkbox, but it can decide which segments are sellable at all.
What would need to be true
Customers must value verified, maintained integrations enough to pay recurring fees, rather than treating a working scraper as a one-time build they can cancel or bring in-house.
Heal cost per project must stay low enough, including on anti-bot-heavy sites, for gross margin to hold under whatever pricing model replaces flat per-project fees.
Accumulated run traces and fix history must measurably improve healing accuracy over time, so a competitor adding code generation and self-healing still can't match your repair quality.
Actions to take this week
Pick one buyer this week: product teams that need to fill API gaps in their own product, the same profile as the YC batchmates who started this. Call 10 of them and ask what they pay now (in engineer hours or vendors) to keep one site integration alive. A positive signal is a named dollar or hours-per-month figure, not "that sounds useful."
Rewrite the landing page hero for that single segment and move scrapers, crawlers, and general back-office use below the fold. Track whether demo requests from that segment go up over the next two weeks.
Pull cost per project from your own data: machine time, proxy, captcha, and heal tokens, split by site type. Find the ten worst-margin projects and decide whether pricing should shift to per-successful-run or per-maintained-integration.
Build and publish a "silent failure" demo: break a site's field mapping so the scraper still runs, then show your agent catching the wrong data and proposing a verified patch with an output diff. Use it as the lead proof asset.
Write a one-page security and data-handling sheet covering credential storage, recording retention, and redaction. Send it to the three most enterprise-leaning leads and note which questions they raise.
Work through these, then roast the idea again with what you learned.
No account needed. One email, no follow-ups.
Your idea is next
What would the panel say about yours?
You just read what four AI examiners found in someone else's idea.
Your startup has a fatal flaw. Find it before you build.