Cycling through all six. Tap any point to stop.

Measured on six things

Ask for the list no one sells, in plain words. Get every match, ready to use in your sheet, CRM or agent, and new ones as they appear.

Speed

Desktop Lighthouse lands at 89/100, but mobile is 60/100 with a 6.3s LCP — well past the 2.5s good threshold. The heavy hero media is the likely culprit. Desktop users are fine; phone users wait.

Screenshots

1 / 7

SpecStream

About this app

Ask for the list no one sells: every roofer in Columbus rated 4.5 or better with fewer than 20 reviews, every food bank in Texas with its revenue, every company that filed to raise $5M or more this week. You get all of it, the small and new ones too, from government systems, registries, maps, each result's own website, PDFs and scans. Open any row and see why it's there. Save it and new matches arrive in Google Sheets, Excel, HubSpot or your agent. 10¢ a match.

Key features

1Natural-language list building from the live web: describe what you want in plain English with and/or/but-not, get back up to 10,000 matching rows per request.
2Per-row pricing with provenance: 10¢/30¢/$1 per matching row depending on depth, and every result shows the page and line that put it there.
3Agent and workflow integrations: MCP server for Claude/ChatGPT/Cursor, plus HubSpot, Salesforce, Sheets, Excel, Airtable, Notion, Zapier, Make, n8n, Slack.

Target customer profile

Drawn from the product itself, not from a survey.

Scout Review

Written by AI from measured evidence, scored out of 100.

SpecStream is a live-web list builder with a genuinely differentiated pitch: ask in plain English for the list no one sells, pay 10¢ per matching row, see the page that put each result there. The site is well-designed, the docs are clear, the security headers are flawless, and the accessibility audit is perfect. Where it stumbles is mobile speed (6.3s LCP) and first-run polish — the hero demo table is obscured by a cookie banner and renders empty. Pricing is aggressive against a $662/mo stack comparison, and the $10 no-card trial is a smart on-ramp. The open questions are the ones the site doesn't answer: data handling for a scraping product, and what actually stops a bigger crawler from copying the wedge.

Measure by measure

Usability

Docs walk you from plain-English request to quote to results, with templates, voice input, and a $10 no-card trial. The natural-language parsing of and/or/but-not is genuinely thoughtful. First-run demo is undercut by the overlapping modal.

Design

Strong hero typography, clean quote modal, and colorful category icons give the site real polish. But the cookie banner covers the live demo table and the demo rows render empty — the best proof of the product is hidden behind a consent popup.

Speed

Desktop Lighthouse lands at 89/100, but mobile is 60/100 with a 6.3s LCP — well past the 2.5s good threshold. The heavy hero media is the likely culprit. Desktop users are fine; phone users wait.

Security

Perfect 12/12 on the header probe — HSTS, CSP, frame protection, Referrer-Policy, SPF, DMARC all in place. Spend-cap approval before work starts is a real safety control. No SOC 2, ISO, or bug bounty published, which is normal at this stage.

Accessibility

Lighthouse accessibility audit returns a perfect 100/100, and the site reads well at mobile widths. No WCAG conformance statement or a11y page published, and keyboard/screen-reader behavior isn't independently verified — so this sits at the top of the measured band, not above it.

Growth

A real, specific problem — the small, local, and new records that data vendors structurally miss — with a clear wedge: build the list on request from live sources, charge per matching row, show provenance. The $662/mo stack comparison is a sharp pitch. Headroom exists, but the moat is thin: scraping quality and coverage are copyable, and the alternatives list is empty enough that the competitive read is unclear.

Improvement areas

Mobile performance: 6.3s LCP is more than double the 'good' threshold — the hero media needs compression or lazy-loading.
First-run demo: the hero table is obscured by the cookie banner and renders empty, hiding the product's strongest proof point.
Trust surface: no data-handling, retention, or provenance policy published for a product whose core function is reading third-party sites.
Accessibility disclosure: 100/100 automated score but no WCAG conformance statement, keyboard notes, or a11y page.

Design & Usability

SpecStream's homepage is genuinely well-built: bold hero type on a dark city-street overlay, a clean quote modal that shows 'Food banks in Texas' at $0.10/result with an 'Approve $2.50' button, and eight colorful category tiles that instantly communicate what you can ask for. The problem is that the best asset — the live demo table — is half-covered by a cookie consent banner and its rows render empty in the captured render, so the proof of the product is the weakest part of the page. Usability is stronger: the docs walk you from a plain-English request through and/or/but-not parsing, templates, and voice input, to a quote you approve before any work runs. A $10 no-card trial lowers the entry cost meaningfully. The gap is first-run polish — the demo that should sell the product is the thing getting in its way.

Speed & Security

Speed is a tale of two devices. Desktop Lighthouse lands at 89/100, which is genuinely good. Mobile lands at 60/100 with a 6.3s LCP, 3.1s FCP, and 282ms TBT — the LCP is more than double the 2.5s 'good' threshold, and the heavy hero media is the obvious suspect. CLS at 0.004 is excellent, so the layout is stable; it's just slow to paint on phones. Security is the stronger half of this pair: the HTTP probe returns a perfect 12/12 — HTTPS enforced, HSTS, CSP, frame protection, X-Content-Type-Options, Referrer-Policy, Permissions-Policy, plus SPF and DMARC. The pricing page's 'you approve the maximum before any work starts' is a real spend-cap control, not marketing. What's missing is disclosure: no SOC 2, ISO, pen-test summary, or bug bounty published, and no page explaining data handling for a product whose entire job is reading other people's sites.

Accessibility & Growth

Accessibility scores a perfect 100/100 on the Lighthouse axe-core audit, and the site holds up at mobile widths. That's a strong automated result, but automated checks cover only part of WCAG — there's no accessibility statement, no WCAG conformance claim, and keyboard and screen-reader behavior aren't independently verified, so this sits at the top of the measured band rather than above it. Multi-language support is partial: docs say you can ask in any language and sources in other languages are read. Growth is the more interesting question. The problem SpecStream names is real and specific — the shop with one Google listing, the company that filed with the state last week, the tender buried in a PDF — records that data vendors structurally miss because they sell what they collected earlier. The wedge is clear: build the list on request from live sources, charge per matching row, show provenance for every result. The $662/mo 'usual stack' comparison is a sharp, concrete pitch. The limit is the moat: coverage and extraction quality are copyable, and with no alternatives listed in the app details, the competitive read is genuinely unclear.

Conclusion

SpecStream is aiming at a real gap — the small, local, and new records that data vendors miss because they sell yesterday's collection. The execution on the marketing site is above average, the docs are honest about limits, and the per-row pricing with provenance is a genuinely good idea. If the mobile experience gets fixed and the product publishes even a basic data-handling page, this becomes easy to recommend to anyone who currently pays for a list vendor plus an enrichment tool plus an email finder. Right now it's a strong pitch with a slow phone experience and a thin public trust surface. Worth a $10 trial; worth watching.

Suggestions for improvement

Fix the mobile LCP — 6.3s is more than double the good threshold. Compress or lazy-load the hero media and re-run Lighthouse.
Move or delay the cookie consent banner so it stops covering the live demo table in the hero.
Populate the hero demo table with real rows, or replace it with a short looping video of a real request running.
Publish a data-handling and provenance page: what gets read, what gets stored, retention, and how to request removal.

How we score

Alternatives comparison

Named competitors, point by point. Nobody paid to appear here or to be left out.

	SpecStream
Pricing model	Per matching row: 10¢ standard, 30¢ expanded, $1 deep. Plans from $0 (with $10 trial) to $499/mo.	-
Data freshness	Built on request from the live web — small, local, and newly filed records included.	-
Provenance / why-this-row	Every result shows a one-line reason per requirement, the source line, and when it was read.	-
Agent / API access	MCP server for Claude, ChatGPT, Cursor; REST API with Python and TypeScript clients.	-
Integrations	HubSpot, Salesforce, Pipedrive, Attio, Google Sheets, Excel, Airtable, Notion, monday, Zapier, Make, n8n, Slack, Teams.	-
Free trial	$10 of work on us for the first 30 days, no credit card required.	-