Steel's growth post says developer-tool growth starts with "developers discovering you when they have the problem". The homepage names 12 problems people bring to Steel, from price monitoring and QA testing to travel booking and deep research, yet steel.dev has 61 URLs and not one of them is a page for a use case. The working code already exists: docs.steel.dev has recipes such as "convex-price-watch", "deep-research", "scrape", "credentials" and "trigger-dev-browser-job". A developer searching "price monitoring with headless browsers" or "browser API for QA tests" finds a recipe in the docs at best, never a page that sells Steel for that job.
First action
Pick the four jobs with recipes and logos and draft the price-monitoring page first.
How to measure it
Signups attributed to use-case pages
- Now
- –
- Target
- 15% of new signups
- By
- 90 days
Steps
Rank the 12 jobs
By search demand and how many current users do them.
Founding Growth Lead 1 day
Draft a template
Problem, code, features, cost, proof, button.
Founding Growth Lead 1 day
Write four pages
Scraping, deep research, price monitoring, QA.
Founding Growth Lead 1 week
Pull in code
Trim each cookbook recipe to a runnable sample.
Founding Engineer 2 days
Add proof
One logo or quote per page from current users.
Founding Growth Lead 1 week
Link the tiles
Homepage use-case tiles open the pages.
Founding Engineer 2 hours
Track
Signups and first sessions by landing page.
Founding Growth Lead weekly
Finish the set
The remaining eight jobs.
Marketing Intern 1 month
Why it works
Publish /use-cases/<job> pages for the 12 jobs, starting with the four that have recipes and logos behind them (web scraping, deep research, price monitoring, QA). Each page: the problem in the searcher's words, a 20-line code sample from the recipe, which Steel features matter for that job (profiles, proxies, human-in-the-loop, traces), a real cost estimate at Launch and Scale rates, a customer logo or quote, and a "Start with $30 free credits" button. Link every page from the homepage tiles that already name the jobs.
Problem-shaped pages catch searches and AI-assistant answers at the moment of need, and turn existing docs into acquisition pages with very little new writing.
Ready to use
CopyUse-case page outline
H1: <Job> with real cloud browsers Problem: what breaks when you run this yourself (2-3 lines) Code: 20 lines from the cookbook recipe What Steel handles: profiles / proxies / human takeover / traces What it costs: example run at $0.10 and $0.08 per browser hour Proof: customer logo or quote CTA: Start with $30 free credits (no credit card)
Risks and how to handle them
Thin pages
Ship each page only with working code and a real cost example.
Docs and site drift apart
Pull code from the cookbook so one source stays current.
Evidence
- HiringFounding Growth Lead (Toronto, posted October 14, 2026): "turn Steel's organic momentum into systematic, repeatable user and revenue growth". Steel believes developer-tool growth "comes down to three things: developers discovering you when they have the problem, experiencing immediate value, and wanting to tell others." The GTM motion is "product-led growth combined with product-led sales. Most users discover Steel, sign up, and upgrade through self-service."Source
- Your websiteHomepage: H1 "Browser Infrastructure for AI Agents"; "Loved by 30,000+ engineers at" 16 logos (Readwise, Zo Computer, Avoca, Profound, Cognizant, Navattic, Magical and others); "Steel runs millions of browser sessions every week"; counters of 1,500,000+ browser hours, 35,000,000+ sessions and "4 Trillion+" tokens scraped; a GitHub link showing "7.9K stars". A "What people build on Steel" block names 12 jobs: AI Agents, Foundational Model Training, AI Shopping Assistants, Web Scraping, RPA Software, Sales Automation, QA Software, Customer Service Agents, Deep Research, Price Monitoring, Form Filling and Travel Booking.Source
- Your websiteSitemap (61 URLs, checked October 23, 2026): the homepage, 51 blog posts and nine other pages (pricing, enterprise, partners, FAQ, changelog, launch week, brand kit, terms, privacy). There is no page for any of the 12 use cases, no customer or case study page and no comparison page; head-to-head content exists only as two blog posts, "Steel vs Browserbase: a practical comparison" (January 14, 2026) and "Steel vs Kernel" (January 20, 2026).Source
- Your docsdocs.steel.dev (150 URLs): about 30 integration pages (Playwright, Puppeteer, Selenium, Stagehand, Browser Use, CrewAI, LangGraph, Mastra, Agno, Pydantic AI, Replit, Bolt.new, x402 and coding-agent setups) and about 45 cookbook recipes, including "convex-price-watch", "deep-research", "scrape", "credentials", "temporal-browser-workflow" and "trigger-dev-browser-job", plus an llms-full.txt and five Steel skills for agents.Source
- Your websitePricing: Launch "$0 per month + usage" with "$30 of credits, one time", 10 concurrent sessions, sessions up to 15 minutes; Scale "$250 per month + usage" with $100 of credits a month, 100 concurrent sessions and a dedicated Slack channel; Enterprise custom. Browser hours $0.10 (Launch) and $0.08 (Scale); CAPTCHA solves $3/1k and $1/1k; proxy bandwidth $10/GB and $6/GB. "Stealth browser" is included only on Enterprise, while the homepage gives stealth its own section ("Keep browsing unblocked").Source