Automation that does the boring work.

Microsoft's modern browser-control tool. It drives Chrome, Firefox and Safari's engine, waits for pages automatically and is very reliable.

Use it for

Almost every browser bot we build

Trade-off

Pages that change often still need maintenance.

Google's browser-control library for Chrome, popular and well documented.

Use it for

Chrome-only tasks and existing Puppeteer code

The oldest browser-automation tool, supporting many languages and browsers.

Use it for

Older projects and special browser setups

A small add-on (Manifest V3) that adds buttons or automates pages right inside your own browser.

Use it for

Helping a person work faster in their own tabs

Browsers that run in the cloud on demand (for example Browserbase or Cloudflare Browser Rendering), so bots do not need your computer.

Use it for

Bots that run on a schedule, at scale

Agents that look at a page and decide what to click, which helps on sites that change often.

Use it for

Flexible tasks that are hard to script

Trade-off

Slower and costlier than a scripted bot, so used where it pays.

Running a browser with no visible window, so bots run quietly on a server.

Use it for

Every scheduled bot

When a site offers an API, RSS feed, sitemap or export, we use it first. It is faster, steadier and clearly allowed.

Use it for

Always the first choice

We read robots.txt and terms, add delays, cache pages and limit requests, so we never overload a site.

Use it for

Every collection job

Trade-off

We are not lawyers, so we flag risks and you decide.

A Python framework built for crawling many pages quickly, with scheduling and retries built in.

Use it for

Large crawls across many pages

Python tools that read a page's HTML and pull out the exact pieces you need.

Use it for

Extracting clean data from pages

A crawling toolkit for JavaScript and Python that manages browsers, queues and retries together.

Use it for

Browser-based crawls with strong reliability

Pandas and Polars tidy messy data: fix formats, remove duplicates and join sources.

Use it for

Turning raw pages into usable tables

Read text from scanned documents, images and PDFs, such as invoices and forms.

Use it for

Document processing

An AI model reads messy text or a screenshot and returns neat, structured data.

Use it for

Pages and documents with no fixed layout

A visual workflow builder you can host yourself, with hundreds of ready connections. Good for teams that want to see and tweak the flow.

Use it for

Connected workflows your team can read

Hosted no-code tools that connect apps with ready-made steps.

Use it for

Simple, quick connections between popular apps

Trade-off

Cost grows with volume, and complex logic gets awkward.

A workflow engine for long, important jobs that must never lose their place, even if a server restarts.

Use it for

Multi-step jobs that run for hours or days

Tools for scheduling data pipelines, with dashboards, retries and dependencies.

Use it for

Regular data pipelines

The simplest scheduler: run this at 6am every day. GitHub Actions can do it too.

Use it for

Straightforward daily and weekly jobs

Jobs wait in a queue and workers process them in turn, with BullMQ, Celery or SQS.

Use it for

Spiky workloads and heavy tasks

Another tool calls the bot the moment something happens, so it reacts instantly.

Use it for

Event-driven automations

UiPath and Power Automate control desktop programs by mimicking clicks.

Use it for

Old desktop software with no API

The standard way software talks to software. Almost every modern tool has one.

Use it for

Connecting your tools directly

Familiar spreadsheet-style homes for results that your team can read and edit.

Use it for

Simple data sharing and lists

HubSpot, Salesforce, Zoho and others. Bots create and update contacts, deals and notes automatically.

Use it for

Keeping sales records complete

Messages through Meta's official WhatsApp Business Platform: confirmations, alerts and support.

Use it for

Customer messages and team alerts

Trade-off

Official only: unofficial WhatsApp automation can get a number banned.

A shared standard (the Model Context Protocol) that lets AI assistants safely use your tools and data, such as files, databases and calendars.

Use it for

Letting an AI agent work with your own systems

Bots that post updates and answer commands inside the chat tools your team already uses.

Use it for

Team alerts and quick actions

Send and read email through Gmail's API, Resend or SendGrid, with templates and tracking.

Use it for

Reports, notices and inbox automation

Branded, ready-to-send documents built from data with libraries such as ReportLab, WeasyPrint and openpyxl.

Use it for

Invoices, summaries and client reports

Metabase, Grafana or Looker Studio turn collected data into live charts.

Use it for

Seeing trends at a glance

Webhooks from Stripe or Razorpay start the next step as soon as money moves.

Use it for

Receipts, access and accounting

An AI that plans steps and uses tools, such as a browser or your CRM, to finish a task for you.

Use it for

Multi-step tasks that vary each time

Trade-off

We add limits, approvals and logs of every action.

AI sorts emails, tags tickets and pulls fields out of documents, far faster than a person.

Use it for

High-volume reading and sorting

AI summarises long threads and drafts replies or reports for a person to review.

Use it for

Saving time on writing

A bot that answers from your own documents, quoting them, instead of guessing.

Use it for

Support and internal Q&A

Risky actions pause and ask a person to approve first, in Slack, WhatsApp or email.

Use it for

Anything involving money or customers

AI that looks at a screenshot or photo and understands it, useful where there is no text to read.

Use it for

Visual checks and unusual layouts

Tools that click through your own website like a visitor and check it works, every time you change it.

Use it for

Automated website testing (QA)

Postman and k6 check that your APIs answer correctly and hold up under many users.

Use it for

Checking back ends before launch

Compares screenshots before and after a change, to catch layout slips no test would notice.

Use it for

Design-sensitive sites

Checks pages and services on a schedule, and alerts you when something changes or stops answering.

Use it for

Prices, stock, competitors and uptime

Tools such as OpenTelemetry and Langfuse record every step a bot or AI agent takes, so you can see why it did what it did.

Use it for

Trusting and debugging AI-powered automations

Every run is logged, failed steps retry on their own, and Sentry or a chat alert tells us when a person must look.

Use it for

Reliable, observable bots

Logins and keys are kept in a secure vault, never in code, with access limited to what each bot needs.

Use it for

Bots that log in to accounts

Trade-off

We only log in with your permission, and we advise when a site forbids bots.

Docker on a small server, Cloud Run, AWS Lambda, GitHub Actions or your own computer, whichever suits the job and the budget.

Use it for

Running on time, at the lowest cost

Websites change and bots can break. A care plan means we notice and fix it before you do.

Use it for

Bots you rely on every day