Analyze
Break down any public URL — summary, structure critique, SEO surface signals, and risk flags — using fetched context, not imagination.
Structured reads for research, sales prep, and content audits.
Drop a URL, pick a lens — executive summary, SEO content marketing, light security posture, UX copy critique, or balanced multi-lens — and get a memo that separates what the page actually says from what you might infer. When fetch fails, the tool admits it instead of improvising marketing claims; when the page is paywalled, you get a clear blocker note instead of a fabricated review. The output is shaped for downstream use: sales discovery, competitive briefs, content audits, and quick due-diligence reads.
Six small choices that change the depth and accuracy.
Each lens optimizes a different downstream use.
Five bullets, fast
What the page is, who it is for, and what it asks the reader to do — paste into a Slack or call note.
Content marketing read
Title, meta, heading hierarchy, internal links, content gaps — the shape competitors actually scan for.
Trust signals only
Visible compliance badges, support contacts, and signal flags — not a pen-test, just visible posture.
Critique mode
CTA clarity, form friction, copy voice consistency — the things designers and PMs argue about.
All angles, lighter depth
Quick read that touches summary, SEO, and UX without going deep on any one lens.
Workflows where one URL is the start of a lot of work.
Most bad URL summaries are confident hallucinations of features that don't exist.
Generic AI summarizers will gladly invent pricing tiers, compliance badges, and customer logos because the page "feels like" it should have them. This template forces a different discipline: every claim is either evidence (quoted or paraphrased from fetched content) or inference (clearly labeled as your reviewer's interpretation). When fetch fails — paywalls, geo-blocks, robots.txt — the output names the blocker instead of guessing what the page would have said. The result is a memo your team can quote without losing credibility.
Habits that compound across competitive scans.
No — paywalled and gated content may be inaccessible. The output will reflect the blocker honestly instead of inventing what the page would have said.
Depends on the underlying fetch and any caching layer in the model pipeline. Re-run after deploys when you need the freshest read.
Best results are on server-rendered pages. Heavy client-side rendering may be partially or fully invisible to the fetcher.
It is instructed not to. If a number or logo appears in the output, verify it against the source page before quoting.
No — security posture lens is a visible-signals read only. Real penetration testing requires authorized professional tools.
Default streaming text models with web fetch tooling. Switch to deeper reasoning models for complex multi-page analyses.
Re-run with a single-lens choice and deep depth. Multi-lens always trades breadth for per-section detail.
Know what a page really argues — before the meeting.
Use it before sales calls, content briefs, due diligence, and competitive scans. Evidence first, inference second, blockers honestly named — your team makes faster decisions when the input memo is shaped for them.