ProgrammableMarketing/steel.dev (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 6
  • This quarter · 9
  • Big bets · 3
  • Wildcards · 4
  • How we can help
Generated Oct 26, 2026
  1. Programmable
  2. Playbooks
  3. Steel
Steel logo

growth playbook · Oct 26, 2026

steel.dev (opens in a new tab)

Steel growth playbook

Diagnosis

Steel grew 50x in 2025 through word of mouth and open source, with nearly 7,900 GitHub stars, millions of sessions and 16 logo customers. Its first growth hire is asked to turn that momentum into a system: developers finding Steel when they have the problem, getting value fast, and telling others.

The site is not built for that yet: 12 use cases on the homepage with no pages, comparisons only as old blog posts, no customer stories, metered pricing with no calculator, and a blog that went quiet for ten weeks. This playbook builds those discovery pages first, then activation, agent-led setup, integrations, the leaderboard audience, open-source conversion and growth loops.

Growth Score

7/10

Biggest Opportunity

Be the page developers find when they search a specific browser job, with working code and a cost estimate, and the default provider coding agents set up.

Biggest Risk

A crowded category (Browserbase, Kernel, Hyperbrowser, Browser Use, Anchor) where rivals compete on price and stealth; without discovery pages and proof, word of mouth alone may not keep up.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    Give each of the 12 jobs on the homepage its own page, built from the cookbook recipes you already have

    Pick the four jobs with recipes and logos and draft the price-monitoring page first.

    High impact First result in 1 month
  2. 2

    A comparison hub: turn the two "Steel vs" blog posts into pages, and add the rivals you have not covered

    Collect each rival's current pricing page.

    High impact First result in 1 month
  3. 3

    Turn 16 logos into five customer stories on steel.dev

    Email the five logo customers who use Steel most.

    High impact First result in 1 month

02

Your idea portfolio

22 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 7 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 2Big bets · 9Fill-ins · 9Reconsider · 2
Effort →

Showing 22 of 22 ideas

This week 6

Quick wins with a clear first action.

  • 01

    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.

    High impactMedium effort First result in 1 month Founding Growth Lead

    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

    1. 1

      Rank the 12 jobs

      By search demand and how many current users do them.

      Founding Growth Lead 1 day

    2. 2

      Draft a template

      Problem, code, features, cost, proof, button.

      Founding Growth Lead 1 day

    3. 3

      Write four pages

      Scraping, deep research, price monitoring, QA.

      Founding Growth Lead 1 week

    4. 4

      Pull in code

      Trim each cookbook recipe to a runnable sample.

      Founding Engineer 2 days

    5. 5

      Add proof

      One logo or quote per page from current users.

      Founding Growth Lead 1 week

    6. 6

      Link the tiles

      Homepage use-case tiles open the pages.

      Founding Engineer 2 hours

    7. 7

      Track

      Signups and first sessions by landing page.

      Founding Growth Lead weekly

    8. 8

      Finish the set

      The remaining eight jobs.

      Marketing Intern 1 month

    First result
    1 month
    Cost
    Staff time only
    Owner
    Founding Growth Lead

    Smallest test

    Publish the price-monitoring page and compare its signups with the blog average for a 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
  • 02

    Steel compared itself with Browserbase and Kernel in January blog posts, but there is no comparison page, and nothing on Hyperbrowser, Browser Use or Anchor, which all pitch cloud browsers for agents. Browser Use now advertises stealth browsers at $0.02 an hour on its homepage, a claim buyers will weigh against Steel's rates.

    High impactLow effort First result in 1 month Founding Growth Lead

    First action

    Collect each rival's current pricing page.

    How to measure it

    Comparison-page signups

    Now
    –
    Target
    100 a month
    By
    90 days

    Steps

    1. 1

      Collect facts

      Rivals' pricing and features from their own pages.

      Founding Growth Lead 2 days

    2. 2

      Set a format

      Same table on every page.

      Founding Growth Lead 1 day

    3. 3

      Convert the posts

      Browserbase and Kernel.

      Founding Growth Lead 2 days

    4. 4

      Write three more

      Hyperbrowser, Browser Use, Anchor.

      Founding Growth Lead 1 week

    5. 5

      Add a benchmark link

      Point to Steel's published remote-browser benchmark.

      Founding Engineer 1 hour

    6. 6

      Publish the hub

      /compare linked from the footer and pricing.

      Founding Engineer 1 day

    7. 7

      Refresh

      Re-check every rival quarterly.

      Founding Growth Lead quarterly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Founding Growth Lead

    Smallest test

    Publish the Browserbase comparison as a page and watch its signups for a month.

    Why it works

    Create /compare with one page per rival: what each does, pricing and limits side by side, open-source and self-hosting, session start time, observability, and when the rival is the better fit. Move the two blog posts into this format and add Hyperbrowser, Browser Use and Anchor. Date every page and refresh quarterly.

    Comparison searches are the highest-intent searches in a crowded category, and fair pages are what AI assistants quote when developers ask which provider to use.

    Risks and how to handle them

    • Claims go stale

      Date every page and cite each rival's own page.

    • Looks hostile

      Say plainly when a rival fits better.

    Evidence

    • 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
    • Competitor researchRivals' own homepages: Browserbase, "Give your agents access to the whole web"; Kernel, "browser infrastructure for web agents and automations" with a customers page; Hyperbrowser, "Cloud browsers, isolated sandboxes, and web data APIs for AI agents"; Browser Use, "connect your own agent to stealth browsers at $0.02 an hour"; Anchor, "Secure Infrastructure for Computer Use 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
    • 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
  • 03

    The homepage shows 16 customer logos and says Steel is "loved by 30,000+ engineers", yet steel.dev has no customer story at all. The only story about Steel is on Fly.io's site, about Steel as Fly's customer. Product-led sales needs proof a buyer can forward to their team.

    High impactMedium effort First result in 1 month Founding Growth Lead

    First action

    Email the five logo customers who use Steel most.

    How to measure it

    Enterprise calls citing a story

    Now
    –
    Target
    30%
    By
    2 quarters

    Steps

    1. 1

      Pick five

      Different jobs and sizes.

      Co-founder and CEO 2 days

    2. 2

      Interview

      30 minutes each.

      Founding Growth Lead 2 weeks

    3. 3

      Write

      One page each with numbers.

      Founding Growth Lead 1 week

    4. 4

      Approve

      Customer sign-off.

      Founding Growth Lead 1 week

    5. 5

      Publish

      /customers and quotes on landing pages.

      Founding Engineer 1 day

    6. 6

      Cut clips

      Short videos from the interviews.

      Marketing Intern 2 weeks

    First result
    1 month
    Cost
    Staff time only
    Owner
    Founding Growth Lead

    Smallest test

    Publish one story and share it on X and LinkedIn.

    Why it works

    Interview five logo customers across different jobs (research, sales automation, QA, shopping, customer service). Write short stories: what they built, why they chose Steel over building browsers themselves, sessions or hours they run, and what changed. Publish at /customers, put one quote on each use-case and comparison page, and send the stories to enterprise leads.

    Peer proof moves self-serve users to paid plans and shortens enterprise evaluations.

    Risks and how to handle them

    • Customers decline to be named

      Offer anonymised stories with numbers.

    Evidence

    • 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
    • Third-party storyFly.io publishes a customer story about Steel, "How Steel.dev Turns AI Agents Into True Browser Ninjas", on its own site; steel.dev links to no customer stories of its own.Source
    • 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
  • 04

    Steel's rates are metered: $0.10 or $0.08 per browser hour, CAPTCHA solves, proxy gigabytes and scrape calls. A developer has to do the math alone to see when Scale's $250 pays off. The homepage gives stealth its own section, but the pricing table includes "Stealth browser" only on Enterprise, which a self-serve developer finds out late.

    High impactLow effort First result in 2 weeks Founding Growth Lead

    First action

    Write the cost formula with three example workloads.

    How to measure it

    Launch to Scale upgrade rate

    Now
    –
    Target
    +20%
    By
    90 days

    Steps

    1. 1

      Write the formula

      Hours, proxies, CAPTCHAs, scrape calls.

      Founding Growth Lead 1 day

    2. 2

      Build the calculator

      On /pricing.

      Founding Engineer 3 days

    3. 3

      Label stealth

      Plan line on the homepage section.

      Founding Growth Lead 1 hour

    4. 4

      Prefill sales

      Pass estimates to the enterprise form.

      Founding Engineer 1 day

    5. 5

      Test

      Upgrades before and after.

      Founding Growth Lead 1 month

    6. 6

      Review

      Adjust example workloads.

      Founding Growth Lead monthly

    First result
    2 weeks
    Cost
    Engineering time
    Owner
    Founding Growth Lead

    Smallest test

    Publish a static table of three example workloads this week.

    Why it works

    Add a calculator: sessions per day, minutes per session, share with proxies and CAPTCHAs. Show the monthly cost on Launch and Scale, the break-even point, and when Enterprise is worth a call. On the homepage stealth section, add a line saying which plan it comes with and what Launch and Scale users get instead (fingerprints, proxies, CAPTCHA solving).

    Showing the math removes the main reason a developer stalls between a free plan and a $250 one, and clear plan labels prevent disappointed upgrades.

    Risks and how to handle them

    • Estimates look high

      Show cost per task as well as per month.

    Evidence

    • 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
    • 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
  • 05

    The growth role starts with "Building our content machine", but the blog went ten weeks without a post between July 30 and October 7. Steel's best posts come from its own data and engineering, such as the finding that customer-supplied proxies failed 22 times more often than managed ones.

    Medium impactLow effort First result in 1 quarter Founding Growth Lead

    First action

    List ten post ideas from last month's changelog.

    How to measure it

    Weekly posts kept up

    Now
    –
    Target
    12 in a row
    By
    1 quarter

    Steps

    1. 1

      Build a list

      Ten ideas from the changelog.

      Founding Growth Lead 1 day

    2. 2

      Assign drafts

      One engineer per post.

      Co-founder and CEO 1 day

    3. 3

      Edit and publish

      Weekly.

      Founding Growth Lead weekly

    4. 4

      Cut social posts

      X, LinkedIn, Reddit.

      Marketing Intern weekly

    5. 5

      Track

      Signups by post.

      Founding Growth Lead monthly

    6. 6

      Prune

      Drop formats that do not convert.

      Founding Growth Lead quarterly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Founding Growth Lead

    Smallest test

    Publish one post from the changelog this week.

    Why it works

    Set one post a week: alternate an engineering story (a bug, a benchmark, a design decision), a recipe from the cookbook turned into a tutorial, and a data note from aggregated session stats. Keep a running list from the engineers' Slack and changelog; the growth lead edits, engineers draft.

    A steady stream of real technical posts compounds in search and gives every social channel something to share.

    Risks and how to handle them

    • Engineers short of time

      Interview them for 20 minutes and write it up.

    Evidence

    • HiringFounding Growth Lead sample projects: "Building our content machine: turn AI agent experiments and research into high-value blog posts, demos, and technical content that establishes Steel as the authority on browser automation"; influencer campaigns "with AI engineers and developer advocates"; fast experiments on "channels, messaging, value props, and best-fit customer profile"; launches for features and open-source releases; optimizing "activation, conversion, and retention to turn freemium into paying customers"; "events, dinners, and community gatherings"; presence on "X, Reddit, HN, Discord". It asks for someone who prioritizes "growth loops over linear channels" and "compounding acquisition where one cohort brings the next".Source
    • Your websiteBlog dates: launch weeks in October 2025 (v2) and June 2026 (v3, with Stealth Browser, Dedicated IPs, Rust and Go SDKs, Atlas and new pricing); posts on July 15, 22 and 30, 2026, then nothing until "Build the agent. Don't build its computer." on October 7, 2026, a ten-week gap. A July post reported that "Customer-supplied proxies failed 22x more often than managed ones".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
  • 06

    The growth post asks for developers "experiencing immediate value", and the homepage already has a CLI one-liner and a copyable agent prompt. What matters is how many signups reach a first working session and how long it takes.

    Medium impactLow effort First result in 2 weeks Founding Engineer

    First action

    Query last month's signups for time to first session.

    How to measure it

    First-hour activation

    Now
    –
    Target
    +25%
    By
    60 days

    Steps

    1. 1

      Define events

      Signup to first real session.

      Founding Engineer 1 day

    2. 2

      Baseline

      Last 30 days.

      Founding Growth Lead 1 day

    3. 3

      Empty-state snippet

      Dashboard first-run code.

      Founding Engineer 2 days

    4. 4

      Nudge email

      No session after an hour.

      Founding Growth Lead 1 day

    5. 5

      Watch replays

      Ten stuck signups.

      Founding Growth Lead 1 week

    6. 6

      Repeat

      Fix the biggest drop each month.

      Founding Engineer monthly

    First result
    2 weeks
    Cost
    Engineering time
    Owner
    Founding Engineer

    Smallest test

    Add the dashboard snippet and compare a week of signups.

    Why it works

    Instrument signup, API key created, first session started and first session over one minute. Put a copy-paste first-session snippet on the empty dashboard, pick the user's language, and send a nudge email to anyone without a session after an hour.

    Activation is the lever for a free plan with one-time credits; small gains compound across every channel.

    Risks and how to handle them

    • Over-nudging

      One email, then stop.

    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 developer section: "Use Steel's Python and Node.js SDKs or its REST API", or "connect Playwright, Puppeteer or Selenium directly, no agent needed. Coding agents can set Steel up for you too", with a one-line CLI install, "steel login", "steel browser start" and a copyable agent prompt in the hero. Launch needs "No credit card".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

This quarter 9

The core plays for the next three months.

  • 07

    The homepage says "Coding agents can set Steel up for you too", and the docs ship an llms-full.txt and five Steel skills. Developers increasingly ask their coding agent to add a browser, and the agent picks whichever provider its context makes easiest.

    High impactMedium effort First result in 1 quarter Founding Growth Lead

    First action

    Run the setup prompt in three coding agents and note every failure.

    How to measure it

    Agent-led signups

    Now
    –
    Target
    25% of signups
    By
    2 quarters

    Steps

    1. 1

      Tag the paths

      CLI, skills, llms-full.txt.

      Founding Engineer 1 week

    2. 2

      Test agents

      One prompt to a working session.

      Software Engineer Intern 2 weeks

    3. 3

      Fix friction

      Docs and error messages agents read.

      Founding Engineer 1 month

    4. 4

      Publish

      A one-prompt setup page.

      Founding Growth Lead 1 week

    5. 5

      Video

      Screen recording of the setup.

      Marketing Intern 1 week

    6. 6

      Track

      Share of signups.

      Founding Growth Lead monthly

    First result
    1 quarter
    Cost
    Engineering time
    Owner
    Founding Growth Lead

    Smallest test

    Test the setup prompt in one agent and fix the first failure.

    Why it works

    Tag signups and API keys that come through the CLI, skills and llms-full.txt; test that the main coding agents can go from "add a cloud browser" to a working session with Steel in one prompt; fix what trips them. Publish the one-prompt setup as a page and a short video.

    Being the default answer inside coding agents is a distribution channel that rivals cannot buy, and measuring it shows whether it is growing.

    Risks and how to handle them

    • Agents change often

      Re-test monthly.

    Evidence

    • Your websiteHomepage developer section: "Use Steel's Python and Node.js SDKs or its REST API", or "connect Playwright, Puppeteer or Selenium directly, no agent needed. Coding agents can set Steel up for you too", with a one-line CLI install, "steel login", "steel browser start" and a copyable agent prompt in the hero. Launch needs "No credit card".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
    • 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
  • 08

    docs.steel.dev has about 30 integration pages (Playwright, Stagehand, Browser Use, CrewAI, LangGraph, Mastra, Replit, Bolt.new and more), but steel.dev has no integrations directory, and the partner form already offers "Co-marketing". Stripe Projects and the Hermes plugin show distribution through other products works.

    High impactMedium effort First result in 1 quarter Founding Growth Lead

    First action

    Rank integrations by active users.

    How to measure it

    Partner-sourced signups

    Now
    –
    Target
    200 a month
    By
    2 quarters

    Steps

    1. 1

      Rank integrations

      By usage.

      Founding Growth Lead 1 week

    2. 2

      Build the directory

      /integrations on steel.dev.

      Founding Engineer 1 week

    3. 3

      Pitch five

      Listing, tutorial, launch post.

      Founding Growth Lead 1 month

    4. 4

      Write tutorials

      Joint posts.

      Founding Growth Lead 1 month

    5. 5

      Track

      Referral sources.

      Founding Growth Lead monthly

    6. 6

      Add more

      Next five frameworks.

      Founding Growth Lead quarterly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Founding Growth Lead

    Smallest test

    Ask one framework for a joint tutorial.

    Why it works

    Build /integrations on steel.dev with one card per integration linking to its docs. Ask the five frameworks with the most Steel users for a listing in their docs or marketplace, a joint tutorial and a shared launch post.

    Each framework listing puts Steel in front of developers at the moment they choose a browser, and joint posts reach audiences Steel does not have.

    Risks and how to handle them

    • Partners list rivals too

      Offer the best tutorial and fast support.

    Evidence

    • 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 websitePartner page form options: "Integrate Steel into your product", "Resell or refer Steel", "Build with Steel for your clients", "Co-marketing". Blog posts announce distribution through others' products: "Steel Browser is live in Stripe Projects" (July 15, 2026) and a browser plugin for Hermes Agent (July 22, 2026).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
  • 09

    leaderboard.steel.dev tracks agents on WebVoyager, BrowseComp, DRACO, WebArena and OSWorld with source-linked scores, an audience of exactly the AI engineers Steel wants. But the main site has no newsletter sign-up and the leaderboard only links "Try Steel".

    Medium impactLow effort First result in 1 quarter Founding Growth Lead

    First action

    Add the sign-up form to the leaderboard.

    How to measure it

    Subscribers

    Now
    –
    Target
    3,000
    By
    2 quarters

    Steps

    1. 1

      Add sign-up

      Leaderboard and blog.

      Founding Engineer 1 day

    2. 2

      Write issue one

      What moved, plus a recipe.

      Founding Growth Lead 1 week

    3. 3

      Benchmark guides

      Run WebVoyager on Steel.

      Software Engineer Intern 1 month

    4. 4

      Send monthly

      Fixed date.

      Founding Growth Lead monthly

    5. 5

      Share

      X and Reddit summaries.

      Marketing Intern monthly

    6. 6

      Measure

      Signups from the email.

      Founding Growth Lead monthly

    First result
    1 quarter
    Cost
    $0
    Owner
    Founding Growth Lead

    Smallest test

    Send one issue to current users.

    Why it works

    Add a sign-up to the leaderboard and blog for a monthly "what moved on the leaderboards" email, with one Steel recipe per issue. Add "run this benchmark on Steel" guides for the browser benchmarks.

    An owned list turns one-off visitors into a channel Steel controls, and the leaderboard gives a reason to subscribe that is not a sales pitch.

    Risks and how to handle them

    • Small list early

      Keep it monthly and short.

    Evidence

    • Your websiteSteel's own channels: Discord (discord.gg/steel-dev), X (@steeldotdev), GitHub (steel-browser, Apache-2.0, 7,859 stars and 997 forks on October 23, 2026, with commits the same day) and a public leaderboard site, leaderboard.steel.dev, comparing agents on WebVoyager, BrowseComp, DRACO, WebArena and OSWorld with source-linked scores. The site has no newsletter sign-up.Source
    • HiringFounding Growth Lead sample projects: "Building our content machine: turn AI agent experiments and research into high-value blog posts, demos, and technical content that establishes Steel as the authority on browser automation"; influencer campaigns "with AI engineers and developer advocates"; fast experiments on "channels, messaging, value props, and best-fit customer profile"; launches for features and open-source releases; optimizing "activation, conversion, and retention to turn freemium into paying customers"; "events, dinners, and community gatherings"; presence on "X, Reddit, HN, Discord". It asks for someone who prioritizes "growth loops over linear channels" and "compounding acquisition where one cohort brings the next".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
  • 10

    Steel Browser has nearly 7,900 GitHub stars and almost 1,000 forks, and Steel says it grew through its open-source community. The FAQ explains that Cloud adds scale, stealth, proxies and CAPTCHA solving, but self-hosters have no clear moment or tool to move.

    High impactMedium effort First result in 1 quarter Founding Engineer

    First action

    Write the migration guide.

    How to measure it

    Cloud signups from the README

    Now
    –
    Target
    150 a month
    By
    2 quarters

    Steps

    1. 1

      Write the guide

      Same code, Cloud endpoint.

      Founding Engineer 1 week

    2. 2

      CLI command

      Switch an existing script.

      Founding Engineer 2 weeks

    3. 3

      README note

      Cloud features and credits.

      Founding Growth Lead 1 day

    4. 4

      Contributor credits

      For active contributors.

      Founding Growth Lead 1 week

    5. 5

      Track

      Tagged signups.

      Founding Growth Lead monthly

    6. 6

      Interview

      Why self-hosters stay.

      Founding Growth Lead 1 month

    First result
    1 quarter
    Cost
    Credit costs
    Owner
    Founding Engineer

    Smallest test

    Add the README link with a tracked URL.

    Why it works

    Add a "move to Cloud" guide and a CLI command that points an existing script at Steel Cloud with the same code. Show a short note in the self-hosted server's startup logs and README about Cloud features, with credits for active contributors.

    Self-hosters are the warmest leads Steel has; making the move a one-line change turns open-source reach into revenue.

    Risks and how to handle them

    • Community dislikes upsell

      Keep the note short and useful.

    Evidence

    • Your websiteSteel's own channels: Discord (discord.gg/steel-dev), X (@steeldotdev), GitHub (steel-browser, Apache-2.0, 7,859 stars and 997 forks on October 23, 2026, with commits the same day) and a public leaderboard site, leaderboard.steel.dev, comparing agents on WebVoyager, BrowseComp, DRACO, WebArena and OSWorld with source-linked scores. The site has no newsletter sign-up.Source
    • HiringEvery Steel job post: "over 7,000 GitHub stars, dozens of paying customers, and millions of sessions served monthly, we grew our platform 50x in 2025 purely through word-of-mouth and our open-source community." The growth role offers "strong organic momentum (7K+ stars, 70% MoM growth)" and work "with innovative AI companies transforming healthcare, supply chain, accounting, and more".Source
    • Your websiteHomepage developer section: "Use Steel's Python and Node.js SDKs or its REST API", or "connect Playwright, Puppeteer or Selenium directly, no agent needed. Coding agents can set Steel up for you too", with a one-line CLI install, "steel login", "steel browser start" and a copyable agent prompt in the hero. Launch needs "No credit card".Source
  • 11

    The growth role plans influencer campaigns "with AI engineers and developer advocates" and asks for growth loops where "one cohort brings the next".

    Medium impactMedium effort First result in 1 quarter Founding Growth Lead

    First action

    List twenty builders who have posted about Steel.

    How to measure it

    Advocate-sourced signups

    Now
    –
    Target
    300 a month
    By
    2 quarters

    Steps

    1. 1

      Find builders

      From GitHub and X mentions.

      Founding Growth Lead 1 week

    2. 2

      Design the deal

      Credits, access, fees.

      Founding Growth Lead 1 week

    3. 3

      Sign ten

      Clear disclosure rules.

      Founding Growth Lead 1 month

    4. 4

      Feature builds

      Blog and cookbook.

      Marketing Intern monthly

    5. 5

      Track

      Tagged links.

      Founding Growth Lead monthly

    6. 6

      Renew

      Keep the best five.

      Founding Growth Lead quarterly

    First result
    1 quarter
    Cost
    Fees and credits
    Owner
    Founding Growth Lead

    Smallest test

    Sponsor one builder's project.

    Why it works

    Recruit ten builders with real audiences who already use Steel; give them credits, early access and a small fee per project; ask for one public build a month with code. Feature the builds on the blog and in the cookbook.

    Developers trust builders more than ads, and each build becomes a reusable recipe.

    Risks and how to handle them

    • Undisclosed sponsorship

      Require clear disclosure.

    Evidence

    • HiringFounding Growth Lead sample projects: "Building our content machine: turn AI agent experiments and research into high-value blog posts, demos, and technical content that establishes Steel as the authority on browser automation"; influencer campaigns "with AI engineers and developer advocates"; fast experiments on "channels, messaging, value props, and best-fit customer profile"; launches for features and open-source releases; optimizing "activation, conversion, and retention to turn freemium into paying customers"; "events, dinners, and community gatherings"; presence on "X, Reddit, HN, Discord". It asks for someone who prioritizes "growth loops over linear channels" and "compounding acquisition where one cohort brings the next".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 websiteSteel's own channels: Discord (discord.gg/steel-dev), X (@steeldotdev), GitHub (steel-browser, Apache-2.0, 7,859 stars and 997 forks on October 23, 2026, with commits the same day) and a public leaderboard site, leaderboard.steel.dev, comparing agents on WebVoyager, BrowseComp, DRACO, WebArena and OSWorld with source-linked scores. The site has no newsletter sign-up.Source
  • 12

    Steel records every session in 1080p with a live viewer, human takeover and readable traces, and the new Marketing Intern will shoot "short-form videos, product demos". Agents doing real web tasks are the most shareable thing Steel has.

    Medium impactLow effort First result in 1 month Marketing Intern

    First action

    Record one run of the homepage demo prompt.

    How to measure it

    Clip-sourced signups

    Now
    –
    Target
    100 a month
    By
    2 quarters

    Steps

    1. 1

      Pick runs

      Team-built demo agents only.

      Marketing Intern weekly

    2. 2

      Cut clips

      30 seconds with trace.

      Marketing Intern weekly

    3. 3

      Add code

      Link to the recipe.

      Marketing Intern weekly

    4. 4

      Post

      X, LinkedIn, Shorts.

      Marketing Intern weekly

    5. 5

      Measure

      Views to signups.

      Founding Growth Lead monthly

    6. 6

      Double down

      Formats that convert.

      Founding Growth Lead monthly

    First result
    1 month
    Cost
    $0
    Owner
    Marketing Intern

    Smallest test

    Post one replay clip this week.

    Why it works

    Each week, pick one striking run from the team's own agents (an agent booking, shopping or researching), cut the replay to 30 seconds with the trace on screen, and post it with the code. Never use customer sessions.

    It shows the product working, costs almost nothing to make and suits X, LinkedIn and YouTube Shorts.

    Risks and how to handle them

    • Privacy

      Use only Steel's own demo sessions.

    Evidence

    • HiringMarketing Intern (Winter 2027, Toronto, posted October 21, 2026): "Shooting and editing short-form videos, product demos, and behind-the-scenes content", "Designing graphics, social assets" in Figma, "Helping launch new features, open-source projects, and creative campaigns" and "Working with engineers to uncover interesting stories". A Software Engineer Intern (Summer 2027) was posted on October 20.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 websiteHomepage developer section: "Use Steel's Python and Node.js SDKs or its REST API", or "connect Playwright, Puppeteer or Selenium directly, no agent needed. Coding agents can set Steel up for you too", with a one-line CLI install, "steel login", "steel browser start" and a copyable agent prompt in the hero. Launch needs "No credit card".Source
  • 13

    Steel runs "product-led growth combined with product-led sales": most users upgrade alone, and the team steps in for evaluation and procurement. Launch caps users at 10 concurrent sessions and 15-minute sessions; accounts hitting those limits are telling you they are ready.

    High impactMedium effort First result in 1 quarter Founding Growth Lead

    First action

    List accounts that hit the 10-session limit last month.

    How to measure it

    Flagged-account upgrade rate

    Now
    –
    Target
    25%
    By
    2 quarters

    Steps

    1. 1

      Define signals

      Limits and usage.

      Founding Growth Lead 1 week

    2. 2

      Build alerts

      Daily list.

      Founding Engineer 1 week

    3. 3

      Write the note

      Personal, with a cost estimate.

      Founding Growth Lead 1 day

    4. 4

      Send

      Within a day of the signal.

      Co-founder and CEO daily

    5. 5

      Track

      Upgrade rate.

      Founding Growth Lead monthly

    6. 6

      Tune

      Thresholds.

      Founding Growth Lead monthly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Founding Growth Lead

    Smallest test

    Email ten accounts at the limit this week.

    Why it works

    Score accounts on concurrency hits, session-length cuts, hours per week and seats. When a score passes a threshold, the founder or growth lead sends a short personal note with a cost estimate for Scale, or an offer to talk about Enterprise.

    Reaching accounts at the moment they hit a limit converts far better than generic upgrade emails.

    Risks and how to handle them

    • Feels like surveillance

      Keep the note helpful and specific.

    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 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
    • Your websiteEnterprise page: a two-step form asking for target sites, estimated monthly browser hours, concurrency, "What do you use today? Another provider / Self-built / Nothing yet / Steel", go-live timing and monthly budget. Homepage enterprise block: 1,000+ concurrent sessions, SSO, "HIPAA signed BAAs" and Steel Outpost for self-hosting "inside your network". A Vanta trust center is at trust.steel.dev.Source
  • 14

    Launch gives $30 of credits once, then usage billing. The moment credits run low is when a developer decides whether to add a card, and it is invisible unless Steel speaks up.

    Medium impactLow effort First result in 1 month Founding Growth Lead

    First action

    Count how many Launch accounts used all credits last quarter.

    How to measure it

    Card added after credits run out

    Now
    –
    Target
    +30%
    By
    90 days

    Steps

    1. 1

      Find thresholds

      50%, 90%, 0%.

      Founding Growth Lead 1 day

    2. 2

      Write emails

      Three short emails.

      Founding Growth Lead 2 days

    3. 3

      Build triggers

      From usage data.

      Founding Engineer 1 week

    4. 4

      Dashboard banner

      Same messages.

      Founding Engineer 3 days

    5. 5

      Track

      Billing added.

      Founding Growth Lead monthly

    6. 6

      Test copy

      Two versions.

      Founding Growth Lead 1 month

    First result
    1 month
    Cost
    Engineering time
    Owner
    Founding Growth Lead

    Smallest test

    Send the 90% email by hand to this week's accounts.

    Why it works

    Send three emails: at 50% used (what you have built, what it would cost monthly), at 90% (add a card, or what Scale includes), and at zero (a one-click way to keep going). Show the same in the dashboard.

    Clear timing at the credit cliff keeps active builders from churning by accident.

    Risks and how to handle them

    • Annoying

      Three emails at most.

    Evidence

    • 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
    • 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
  • 15

    The growth role includes "events, dinners, and community gatherings that turn users into advocates", and Steel hires in Toronto.

    Medium impactMedium effort First result in 1 month Founding Growth Lead

    First action

    Book a venue for the first Toronto night.

    How to measure it

    Event-sourced pipeline

    Now
    –
    Target
    5 enterprise talks a quarter
    By
    2 quarters

    Steps

    1. 1

      Pick dates

      Monthly Toronto, quarterly SF.

      Founding Growth Lead 1 week

    2. 2

      Find speakers

      Three users each time.

      Founding Growth Lead 2 weeks

    3. 3

      Invite

      Active users and prospects.

      Founding Growth Lead 2 weeks

    4. 4

      Record

      Clips for social.

      Marketing Intern monthly

    5. 5

      Follow up

      Credits and calls.

      Founding Growth Lead monthly

    6. 6

      Measure

      Upgrades and referrals.

      Founding Growth Lead quarterly

    First result
    1 month
    Cost
    Venue and food
    Owner
    Founding Growth Lead

    Smallest test

    Run one Toronto demo night.

    Why it works

    Host a monthly demo night in Toronto and a quarterly dinner in San Francisco: three Steel users show what they built, then open conversation. Invite active users and prospects from the enterprise form.

    Small, high-quality events create advocates and enterprise relationships that online channels do not.

    Risks and how to handle them

    • Low turnout

      Start small with 20 seats.

    Evidence

    • HiringFounding Growth Lead sample projects: "Building our content machine: turn AI agent experiments and research into high-value blog posts, demos, and technical content that establishes Steel as the authority on browser automation"; influencer campaigns "with AI engineers and developer advocates"; fast experiments on "channels, messaging, value props, and best-fit customer profile"; launches for features and open-source releases; optimizing "activation, conversion, and retention to turn freemium into paying customers"; "events, dinners, and community gatherings"; presence on "X, Reddit, HN, Discord". It asks for someone who prioritizes "growth loops over linear channels" and "compounding acquisition where one cohort brings the next".Source
    • HiringMarketing Intern (Winter 2027, Toronto, posted October 21, 2026): "Shooting and editing short-form videos, product demos, and behind-the-scenes content", "Designing graphics, social assets" in Figma, "Helping launch new features, open-source projects, and creative campaigns" and "Working with engineers to uncover interesting stories". A Software Engineer Intern (Summer 2027) was posted on October 20.Source
    • Your websiteEnterprise page: a two-step form asking for target sites, estimated monthly browser hours, concurrency, "What do you use today? Another provider / Self-built / Nothing yet / Steel", go-live timing and monthly budget. Homepage enterprise block: 1,000+ concurrent sessions, SSO, "HIPAA signed BAAs" and Steel Outpost for self-hosting "inside your network". A Vanta trust center is at trust.steel.dev.Source

Big bets 3

Larger plays that can change the trajectory.

  • 16

    Steel runs millions of browser sessions and has already published findings from its own data, like managed proxies failing far less often than customer-supplied ones. No one else in the category publishes regular, aggregated data on agent browsing.

    High impactHigh effort First result in 2 quarters Co founder and CEO

    First action

    Pick ten questions developers ask most.

    How to measure it

    Report downloads

    Now
    –
    Target
    5,000
    By
    1 year

    Steps

    1. 1

      Pick questions

      Ten developer questions.

      Founding Growth Lead 1 week

    2. 2

      Privacy rules

      Aggregate only, no customers or sites.

      Co-founder and CEO 1 week

    3. 3

      Analyze

      Session data.

      Founding Engineer 1 month

    4. 4

      Write

      Report and summary.

      Founding Growth Lead 1 month

    5. 5

      Launch

      Post, talk, press.

      Founding Growth Lead 1 month

    6. 6

      Repeat

      Every six months.

      Founding Growth Lead 6 months

    First result
    2 quarters
    Cost
    Analyst time
    Owner
    Co founder and CEO

    Smallest test

    Publish one chart from session data.

    Why it works

    Publish an aggregated, anonymised report: session lengths by job, failure causes, which page types slow agents down, how human takeover is used. Never name customers or target sites. Launch it with a post, a talk and press outreach.

    Original data earns links, press and AI citations, and makes Steel the authority on browser automation, as the growth role asks.

    Risks and how to handle them

    • Customer data concerns

      Aggregate and publish the method.

    Evidence

    • Your websiteBlog dates: launch weeks in October 2025 (v2) and June 2026 (v3, with Stealth Browser, Dedicated IPs, Rust and Go SDKs, Atlas and new pricing); posts on July 15, 22 and 30, 2026, then nothing until "Build the agent. Don't build its computer." on October 7, 2026, a ten-week gap. A July post reported that "Customer-supplied proxies failed 22x more often than managed ones".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
    • HiringFounding Growth Lead sample projects: "Building our content machine: turn AI agent experiments and research into high-value blog posts, demos, and technical content that establishes Steel as the authority on browser automation"; influencer campaigns "with AI engineers and developer advocates"; fast experiments on "channels, messaging, value props, and best-fit customer profile"; launches for features and open-source releases; optimizing "activation, conversion, and retention to turn freemium into paying customers"; "events, dinners, and community gatherings"; presence on "X, Reddit, HN, Discord". It asks for someone who prioritizes "growth loops over linear channels" and "compounding acquisition where one cohort brings the next".Source
  • 17

    Steel offers HIPAA BAAs, Steel Outpost for self-hosting inside a customer's network, and a Vanta trust center, and says it works with AI companies "transforming healthcare". Few rivals lead with regulated use.

    High impactHigh effort First result in 2 quarters Co founder and CEO

    First action

    List current users in healthcare.

    How to measure it

    Healthcare pipeline

    Now
    –
    Target
    10 qualified deals
    By
    3 quarters

    Steps

    1. 1

      Find users

      Healthcare accounts today.

      Founding Growth Lead 1 week

    2. 2

      Build the page

      Jobs, security, BAA, Outpost.

      Founding Growth Lead 1 month

    3. 3

      Security pack

      From the trust center.

      Co-founder and CEO 2 weeks

    4. 4

      Two stories

      Healthcare customers.

      Founding Growth Lead 2 months

    5. 5

      Events

      Two healthcare AI events.

      Co-founder and CEO 2 quarters

    6. 6

      Measure

      Pipeline.

      Co-founder and CEO monthly

    First result
    2 quarters
    Cost
    Event costs
    Owner
    Co founder and CEO

    Smallest test

    Interview three healthcare users.

    Why it works

    Package a healthcare page: common jobs (payer portals, prior authorization forms, records lookups), the security pack, a BAA path and Outpost. Find two healthcare customers for stories and go to two healthcare AI events.

    A vertical with compliance needs pays more, churns less and favours vendors that already have the paperwork.

    Risks and how to handle them

    • Long sales cycles

      Keep self-serve for small teams.

    Evidence

    • Your websiteEnterprise page: a two-step form asking for target sites, estimated monthly browser hours, concurrency, "What do you use today? Another provider / Self-built / Nothing yet / Steel", go-live timing and monthly budget. Homepage enterprise block: 1,000+ concurrent sessions, SSO, "HIPAA signed BAAs" and Steel Outpost for self-hosting "inside your network". A Vanta trust center is at trust.steel.dev.Source
    • HiringEvery Steel job post: "over 7,000 GitHub stars, dozens of paying customers, and millions of sessions served monthly, we grew our platform 50x in 2025 purely through word-of-mouth and our open-source community." The growth role offers "strong organic momentum (7K+ stars, 70% MoM growth)" and work "with innovative AI companies transforming healthcare, supply chain, accounting, and more".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
  • 18

    The growth role asks for "growth loops over linear channels" and "compounding acquisition where one cohort brings the next". Steel's cookbook already has about 45 recipes written by the team and contributors.

    High impactHigh effort First result in 3 quarters Founding Growth Lead

    First action

    Turn ten cookbook recipes into one-click templates.

    How to measure it

    Template-sourced signups

    Now
    –
    Target
    20% of signups
    By
    3 quarters

    Steps

    1. 1

      Pick ten

      Recipes to templates.

      Founding Engineer 1 month

    2. 2

      Build the gallery

      Publish and deploy.

      Founding Engineer 1 quarter

    3. 3

      Author credits

      Per deploy.

      Founding Growth Lead 1 month

    4. 4

      Invite authors

      Contributors and advocates.

      Founding Growth Lead 1 month

    5. 5

      Feature

      Monthly picks.

      Marketing Intern monthly

    6. 6

      Measure

      Loop rate.

      Founding Growth Lead monthly

    First result
    3 quarters
    Cost
    Engineering time and credits
    Owner
    Founding Growth Lead

    Smallest test

    Ship one template with an author credit.

    Why it works

    Open a gallery of runnable templates (price watcher, researcher, form filler) that anyone can publish. Each deploy from a template credits its author and signs the new user up to Steel. Feature top templates monthly.

    Every published template brings the next users, and the authors are motivated to promote their own work.

    Risks and how to handle them

    • Low-quality templates

      Review before featuring.

    Evidence

    • HiringFounding Growth Lead sample projects: "Building our content machine: turn AI agent experiments and research into high-value blog posts, demos, and technical content that establishes Steel as the authority on browser automation"; influencer campaigns "with AI engineers and developer advocates"; fast experiments on "channels, messaging, value props, and best-fit customer profile"; launches for features and open-source releases; optimizing "activation, conversion, and retention to turn freemium into paying customers"; "events, dinners, and community gatherings"; presence on "X, Reddit, HN, Discord". It asks for someone who prioritizes "growth loops over linear channels" and "compounding acquisition where one cohort brings the next".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 websiteSteel's own channels: Discord (discord.gg/steel-dev), X (@steeldotdev), GitHub (steel-browser, Apache-2.0, 7,859 stars and 997 forks on October 23, 2026, with commits the same day) and a public leaderboard site, leaderboard.steel.dev, comparing agents on WebVoyager, BrowseComp, DRACO, WebArena and OSWorld with source-linked scores. The site has no newsletter sign-up.Source

Wildcards 4

Unexpected experiments, each with a cheap way to test it.

  • 19

    Steel grew through word of mouth, and Launch already gives $30 of credits. A referral link makes the word of mouth trackable and rewards it.

    Medium impactLow effort First result in 1 month Founding Growth Lead

    First action

    Build a simple referral link.

    How to measure it

    Referred signups

    Now
    –
    Target
    10% of signups
    By
    2 quarters

    Steps

    1. 1

      Set rules

      Ten hours to qualify.

      Founding Growth Lead 1 day

    2. 2

      Build the link

      Dashboard.

      Founding Engineer 1 week

    3. 3

      Announce

      Email and Discord.

      Founding Growth Lead 1 day

    4. 4

      Track

      Referrals.

      Founding Growth Lead monthly

    5. 5

      Stop abuse

      One reward per account.

      Founding Engineer 1 week

    6. 6

      Review

      Cost per referral.

      Founding Growth Lead quarterly

    First result
    1 month
    Cost
    Credit costs
    Owner
    Founding Growth Lead

    Smallest test

    Offer referral credits to Discord members by hand.

    Why it works

    Add a referral link in the dashboard: both developers get $30 of extra credits when the new one runs ten browser hours. Show referrals in the account.

    Rewarding what already happens lets Steel see and grow it.

    Risks and how to handle them

    • Abuse

      Require real usage before rewards.

    Evidence

    • HiringEvery Steel job post: "over 7,000 GitHub stars, dozens of paying customers, and millions of sessions served monthly, we grew our platform 50x in 2025 purely through word-of-mouth and our open-source community." The growth role offers "strong organic momentum (7K+ stars, 70% MoM growth)" and work "with innovative AI companies transforming healthcare, supply chain, accounting, and more".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
  • 20

    Steel's Discord is where developers ask how to do things, but answers there are not searchable. The FAQ page covers a few questions.

    Medium impactLow effort First result in 1 month Marketing Intern

    First action

    Pick last week's five best answers.

    How to measure it

    New FAQ entries

    Now
    –
    Target
    60
    By
    1 quarter

    Steps

    1. 1

      Pick answers

      Five a week.

      Marketing Intern weekly

    2. 2

      Rewrite

      Short entries.

      Marketing Intern weekly

    3. 3

      Review

      Engineer check.

      Founding Engineer weekly

    4. 4

      Publish

      FAQ and docs.

      Marketing Intern weekly

    5. 5

      Link back

      Discord thread.

      Marketing Intern weekly

    6. 6

      Measure

      Search traffic.

      Founding Growth Lead monthly

    First result
    1 month
    Cost
    $0
    Owner
    Marketing Intern

    Smallest test

    Publish five entries this week.

    Why it works

    Every week, pick the five best Discord answers, rewrite them as FAQ entries or docs pages, and link back to the thread. Track which entries get search traffic.

    It turns support work into search content and cuts repeat questions.

    Risks and how to handle them

    • Out-of-date answers

      Date and review quarterly.

    Evidence

    • Your websiteSteel's own channels: Discord (discord.gg/steel-dev), X (@steeldotdev), GitHub (steel-browser, Apache-2.0, 7,859 stars and 997 forks on October 23, 2026, with commits the same day) and a public leaderboard site, leaderboard.steel.dev, comparing agents on WebVoyager, BrowseComp, DRACO, WebArena and OSWorld with source-linked scores. The site has no newsletter sign-up.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
  • 21

    Job posts say "over 7,000 GitHub stars" and "millions of sessions served monthly"; the homepage says 7.9K stars, "millions of browser sessions every week" and 35,000,000+ sessions. Different numbers in different places make a strong story look less precise.

    Low impactLow effort First result in 1 week Founding Growth Lead

    First action

    Write the current stats block.

    How to measure it

    Pages updated

    Now
    –
    Target
    All public pages
    By
    1 month

    Steps

    1. 1

      Agree numbers

      One source.

      Co-founder and CEO 1 day

    2. 2

      Update the homepage

      Dated block.

      Founding Engineer 1 hour

    3. 3

      Update job posts

      Same numbers.

      Founding Growth Lead 1 hour

    4. 4

      Update the brand kit

      Press stats.

      Marketing Intern 1 hour

    5. 5

      Set a reminder

      Monthly refresh.

      Founding Growth Lead monthly

    6. 6

      Check decks

      Sales materials.

      Founding Growth Lead 1 day

    First result
    1 week
    Cost
    $0
    Owner
    Founding Growth Lead

    Smallest test

    Update the job posts to the homepage numbers.

    Why it works

    Keep one dated stats block (stars, sessions, hours, engineers) and use it in the homepage, job posts, decks and press kit. Update it monthly.

    Consistent, current numbers are more believable and easier for press to quote.

    Risks and how to handle them

    • Numbers go stale

      Monthly update.

    Evidence

    • HiringEvery Steel job post: "over 7,000 GitHub stars, dozens of paying customers, and millions of sessions served monthly, we grew our platform 50x in 2025 purely through word-of-mouth and our open-source community." The growth role offers "strong organic momentum (7K+ stars, 70% MoM growth)" and work "with innovative AI companies transforming healthcare, supply chain, accounting, and more".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 websiteSteel's own channels: Discord (discord.gg/steel-dev), X (@steeldotdev), GitHub (steel-browser, Apache-2.0, 7,859 stars and 997 forks on October 23, 2026, with commits the same day) and a public leaderboard site, leaderboard.steel.dev, comparing agents on WebVoyager, BrowseComp, DRACO, WebArena and OSWorld with source-linked scores. The site has no newsletter sign-up.Source
  • 22

    Real visitors pass Core Web Vitals on steel.dev, but largest paint is trending about 10% slower and the mobile lab test scores 41 with 1.4 seconds of main-thread blocking, mostly from the animated hero.

    Low impactLow effort First result in 2 weeks Founding Engineer

    First action

    Profile the homepage on a mid-range phone.

    How to measure it

    Mobile performance score

    Now
    –
    Target
    70
    By
    1 quarter

    Steps

    1. 1

      Profile

      Find the blocking scripts.

      Founding Engineer 1 day

    2. 2

      Defer animation

      After first paint.

      Founding Engineer 2 days

    3. 3

      Lazy sections

      Below the fold.

      Founding Engineer 2 days

    4. 4

      Set a budget

      CI check.

      Founding Engineer 1 day

    5. 5

      Watch field data

      Monthly.

      Founding Engineer monthly

    6. 6

      Review

      New sections.

      Founding Engineer ongoing

    First result
    2 weeks
    Cost
    Engineering time
    Owner
    Founding Engineer

    Smallest test

    Defer the hero animation and re-run the test.

    Why it works

    Defer the hero animation until after first paint, split the long homepage into lazily loaded sections, and set a performance budget in CI.

    Developers judge infrastructure companies by their own speed, and a budget stops the slide before real visitors feel it.

    Risks and how to handle them

    • Less impressive hero

      Keep the animation, start it later.

    Evidence

    • Google PageSpeedsteel.dev homepage: real Chrome visitors pass Core Web Vitals (largest paint 1.4 seconds at the 75th percentile, interaction 87 ms, no layout shift), though largest paint is trending about 10% slower. The mobile lab test scores 41 with a 6.1-second largest paint (the hero sub-heading) and 1.4 seconds of main-thread blocking; desktop scores 68.Source

03

How Programmable can help

Where an idea in this playbook is something Programmable builds or runs, it's listed here with what we'd do. Everything else is yours to run.

ProgrammableTalk

Fill the events in this plan

These ideas are events. Talk gets registrants for them and tells you who's coming and what their company does before the day.

Plan an event (opens programmable.talk in a new tab)
ProgrammableDigital

Show it, don't just describe it

These ideas need something to show: a demo, a video, an animation, graphics or a deck. Digital turns screenshots or the site into all of them, in the company's own brand.

Make a demo (opens programmable.digital in a new tab)
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • Icarus growth strategyBuilds APOLLO, a solar-powered autonomous aircraft that flies above 60,000 ft for persistent sensing, communications relay and data…22 growth ideas
  • Prose growth strategyCustom, made-to-order haircare and skincare: an online consultation feeds Prose's Singular platform, which formulates each product and…21 growth ideas
  • Cofertility growth strategyAn egg-sharing platform: through Split, women freeze their eggs for free when they donate some to intended parents; Keep helps women…20 growth ideas
  • Sparrow growth strategyEmployee leave management for US and Canadian employers: software plus leave specialists who handle policies, claims filing, payroll…22 growth ideas
  • Sweetgreen growth strategyA US fast-casual chain of about 300 restaurants serving salads, warm bowls, wraps and plates made from scratch, with its own app and SG…22 growth ideas
  • Jenni Kayne growth strategyA California lifestyle brand selling women's apparel (cashmere, knitwear and now denim), furniture and home decor, Oak Essentials…22 growth ideas
  • VEIR growth strategyDesigns and manufactures superconducting power delivery systems for AI data centers: the SL Series for low-voltage (480 VAC and 800 VDC)…21 growth ideas
  • Gamma growth strategyAn AI design partner for presentations, documents, websites and social posts: an agent that plans and builds a whole gamma from a prompt…21 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on October 26, 2026, and cites them throughout. Steel did not commission, review or endorse it, and Programmable is not affiliated with Steel. Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how Steel runs its business.

Something wrong, or do you represent Steel? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 307 ideas

    Use-case pages, a comparison hub, customer stories, a cost calculator, a weekly blog and activation tracking; one set of traction numbers.

  2. Days 31 to 6012 ideas

    Agent-led setup, integrations, the leaderboard newsletter, open-source to Cloud, advocates, replay clips, account signals, credit emails and events; Discord answers, referrals and homepage speed.

  3. Days 61 to 903 ideas

    Commit to the big bets: an agent-browsing data report, a healthcare offer and a template gallery.

The research behind the plan

3 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value sessions that start in under a second. The most cited problem: bot detection and CAPTCHAs.

  • Open source and self-hosting
  • Running browser fleets yourself
  • In their words: “cloud browsers”

Market

Competitive landscape

5 competitors compared: Browserbase, Kernel, Hyperbrowser, Browser Use and 1 more.

  • Gap to exploit: Use-case pages with working code and real costs
  • Browserbase: Not open source at the core
  • Kernel: Younger ecosystem

Hiring signals

Steel's first growth hire is a Founding Growth Lead to turn word-of-mouth momentum into a repeatable system across discovery, activation and referral, with product-led sales for enterprise (ideas 1, 2, 4, 6, 13).

  • Founding Growth Lead: Discovery when developers have the problem, activation, loops. Ideas 1-6, 13, 18.
  • Marketing Intern (Winter 2027): Short-form video, demos, launch assets. Ideas 12, 20.
  • Software Engineer Intern (Summer 2027): Developer tools and agent experiments. Ideas 7, 9.

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks first (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence