Why we built a browser

Every SEO has run this experiment, usually by accident. Your rank tracker reports position 3. You search the term yourself and find your listing below two ads, a map pack, and an AI Overview, somewhere near the bottom of the first viewport. Neither number is wrong. They're measuring different webs.

The tracker measures the clean one: a datacentre IP, no cookies, no history, no meaningful location, often no JavaScript. Nobody you're trying to reach searches from there.

The web your tools stopped seeing

SEO tooling followed the rest of software into the cloud, mostly for good reasons. Crawls scale better on someone else's servers. Dashboards are easier to share with clients. Recurring data collection belongs in infrastructure, not on a laptop.

But something got lost in the move. Cloud tools see the web the way a lab sees a patient: useful readings, taken under conditions that never occur in real life. A crawler fetches your page with a clean user agent from an IP range Google has known about for a decade. A rank tracker scrapes a SERP assembled for a user who does not exist. An API dashboard aggregates both and presents the result as ground truth.

Meanwhile, the web your market uses is logged in, localised, personalised, and rendered by JavaScript. The SERP a customer in Manchester sees on their phone at 8pm shares a keyword with your tracker's export and not much else. The page Googlebot renders can differ from the page your crawler fetched, which differs again from the page a real session receives after consent banners, A/B tests, and client-side hydration have done their work.

We spent years as consultants working around this. You already know the workaround, because you use it too: a VPN, an incognito window, a user-agent switcher, and four extensions, all to reconstruct what a normal person sees when they search. The most sophisticated tool stack in marketing, and the last mile still runs on duct tape.

Lab data has its place. We built dashboards on it for years and will keep using it for trend lines and scale. But when the question is "what does my market see, and what's wrong with the page they land on," lab data answers a different question and calls it the same one.

So we built the browser

There is one place where the real web exists: a live browser session, with cookies, a location, a viewport, and a rendered DOM. Not a snapshot, not an API's memory of one.

That's an awkward conclusion for two SEO consultants to reach, because the browser is the least glamorous tool anyone owns. It's also the only honest one. So instead of building another dashboard that approximates the browser, we built the browser: a desktop app for macOS and Windows, purpose-built for SEO work.

What that means in practice:

On-page audits run against the rendered DOM, live, as the page updates. The 300+ rule engine checks what the session received, not what the server sent. Structured data gets validated on the page it lives on.

SERPs are real SERPs. SERP feature detection runs in your own session, in your own location, or in any of 50+ emulated geolocations. Switch city and device without losing where you were. No VPN.

The pixel-width preview shows where your title truncates before your client screenshots it and asks why.

Google Search Console data sits next to the page itself. Queries, clicks, and impressions for the URL you're looking at, in the same window as the DOM they describe, instead of three tabs and a lot of alt-tabbing during quarterly reviews.

None of this is exotic. Most of it existed in fragments across extensions, bookmarklets, and subscriptions you resent paying for. What didn't exist was one place where the rendered page, the live SERP, and your Search Console data share a window. The industry decided the browser was a dumb client and the intelligence should live in the cloud. We think the intelligence belongs where the evidence is.

AI can only fix what it can see

Then AI arrived, and the gap between lab data and the real web stopped being an annoyance and became a liability.

An assistant is only as good as its context. Bolt one onto a cloud API and it reasons about a data export: a crawl from last Tuesday, rankings from a datacentre, HTML that may bear little resemblance to what users receive. It will produce confident recommendations about a web it has never seen. We've read enough AI-generated audit summaries to know how that goes.

Put the same intelligence inside the browser and the raw material changes. The new AI panel in SERP Lens is a streaming chat that carries the current page's analysis in its context from the first message. Ask it why a template is misfiring and it executes real tools: runs an audit, reads the rule results, reads the SERP data, checks a live element in the DOM in front of you both.

We were deliberate about the boundaries. Every tool call is visible in the panel as it happens, so you can watch the reasoning instead of auditing it after the fact. Runs are step-capped and stoppable. The AI proposes fixes; it never silently changes anything. If you've watched an agent wander off unsupervised in a codebase, you'll understand why those constraints are load-bearing.

And the AI doesn't have to be ours. SERP Lens runs as an MCP server, one npx @serplens/mcp away, so Claude, Cursor, or whatever assistant you already work in can run SERP Lens audits and read the results. There's a CLI for the same reason. The browser supplies the eyes. You choose the brain.

This is why we keep saying the browser is the right home for AI-era SEO. The argument was never affection for desktop software. An agent with access to the rendered page, the live SERP, and your real session can observe and act on the web as it exists. An agent reading an export is doing lab work, at higher speed, with better prose.

Where this goes

Everything on our roadmap follows from the same premise: if the work happens in the session, the tooling should too. There is more to see from this vantage point than we've shipped, and more an agent can do with what it sees. We'd rather ship it than tease it, so that's all we'll say for now.

Open it when you work

We're not asking you to switch browsers. Keep Chrome or Arc or whatever you live in for email and Slack. SERP Lens is the browser you open when you do SEO work, the way you open your editor to write code.

It's now free to start: the core browser, live page analysis, a 30-rule audit set, SERP feature detection, and 500 AI credits per cycle. No card required. Paid tiers add the full 300+ ruleset, projects, screenshots, and more. Your data stays local-first. We're UK-registered and GDPR-aligned, and the browser is private by design and quiet by default.

Open a page you're responsible for. Run the audit. Ask the panel what it sees. That's the whole pitch, and the product makes it better than we can.