🏠❤️ Care Services Discovery & AI Recommender

This actor discovers, crawls, and analyzes publicly available care service information from a UK-based care services provider website (e.g. Agincare).

It automatically discovers service and sub-service URLs, extracts structured content from each service page, and then uses AI to recommend up to 3 relevant care services based on the user’s needs.

The actor combines web scraping + AI reasoning into a single workflow.

---

How It Works

The actor operates in three logical steps:

1. Discover
   Finds all service and sub-service URLs from the provider’s website.

2. Crawl
   Extracts structured data from each service page, including reviews, contact details, and content sections.

3. Recommend (AI)
   Uses AI to analyze the crawled data and recommends the top 3 services based on the user’s input.

What kind of data is collected?

The scraper works in three logical phases:

1️⃣ Service Discovery

Discovers all service and sub-service URLs from Agincare’s navigation structure.

Collected data:

• Service name (top-level category)
• Sub-service name (menu item / care type)
• Service URL

---

2️⃣ Service Page Crawling

Crawls each discovered service page and extracts detailed information.

Collected data:

• Page URL
• Service / Sub-service
• Contact phone numbers
• Customer reviews
  • Review text
  • Review date (when available)
• Content blocks
  • Headings (H2 / H3)
  • Paragraph text grouped by section

3️⃣ Recommendations

Uses AI to recommend up to 3 services based on the user’s needs:

Included fields:

• name of service
• Url of the recommended service
• Explaination about why this service is best for your needs
• contact no

Each record is stored as

{
"type": "recommendation",
"userInput": "Elderly dementia home care with night support",
"recommendedServices": [
{
"name": "Dementia Care at Home",
"url": "https://www.agincare.com/services/dementia-care-at-home/",
"explanation": "This service specializes in dementia care with flexible night support.",
"contact": "01305 238360"
}
],
"createdAt": "2026-01-25T11:20:00.000Z"
}

---

Datasets Produced

This actor creates two datasets:

📁 service_urls

• Stores discovered service URLs.

Example:

{
  "service": "Care Services",
  "subService": "Cancer Care",
  "url": "https://www.agincare.com/care-types/cancer-care/"
}

📁 service_pages

• Stores extracted content from each service page.

Example:

{
  "url": "https://www.agincare.com/care-types/cancer-care/",
  "service": "Care Services",
  "subService": "Cancer Care",
  "phoneNumbers": ["01234 567890"],
  "reviews": [
    {
      "text": "Excellent and compassionate service...",
      "date": "April 2025"
    }
  ],
  "contentBlocks": [
    {
      "heading": "Why choose our cancer care?",
      "paragraphs": [
        "We provide specialist care...",
        "Our carers are fully trained..."
      ]
    }
  ]
}

---

📦 Storage & Output Structure

All data is stored in a single Apify Dataset (Default dataset)

There are no separate datasets for URLs, pages, or recommendations.

Instead, every item in the dataset contains a type field:

value Description

service_url Discovered service URLs

---

service_page Crawled service page data

---

recommendation AI-based service recommendations

🔍 How to Find Data in the Storage Tab

Go to Actor run → Storage → Dataset (Default).

You will see the Export dataset section with predefined views:

• Discovered Service URLs → type = service_url

• Service Pages Data → type = service_page

• AI Recommendations → type = recommendation

All items → everything together

These views do not create separate files — they simply filter the same dataset.

Downloading data

1. Select the view you want (e.g. Service Pages Data).

2 Choose a format (JSON, CSV, Excel, etc.).

3 Click Download.

Each view downloads only the filtered records, even though they all come from the same dataset.

Why everything downloads the same dataset?

This is expected behavior.

Apify allows only one dataset per Actor run by default. The UI tabs (Discovered URLs, Service Pages, AI Recommendations) are filtered views of the same dataset, based on the type field.

This design:

• Keeps storage simple

• Avoids duplicated data

• Makes filtering easy via UI or API

Example:

Input	Type	Required	Description
mode	enum	✅ Yes	Select execution mode: discover, crawl, or discover_and_crawl.
singleServiceUrl	string	❌ No	Crawl a single service page instead of all discovered URLs. Required when mode is crawl without discovery.
maxRequestsPerCrawl	integer	❌ No	Maximum number of service pages to crawl. Default: 3000.
dryRun	boolean	❌ No	If true, data is logged but not saved to datasets. Useful for testing.
proxyConfig	object	❌ No	Apify proxy configuration (recommended for production runs).

What are the advantages of using this scraper?

🧭 Automatically discovers all service URLs

🧱 Extracts structured WordPress content

⭐ Captures customer reviews without JavaScript execution

⚡ Fast and cost-efficient (no browser)

📊 Export-ready datasets

🔁 Reusable for monitoring content changes

This data can be used for:

• Market research
• Service comparison
• Content analysis
• Lead enrichment
• Data pipelines and reporting
• Elderly care advisory startups
• Assisted living comparison websites
• Care marketplaces

Developer Integration

You can get started quickly by installing the official client libraries for your preferred environment:

Node.js: Add the apify-client package via NPM.

Python: Grab the apify-client package from PyPI.

Check this API documentation:

Is it legal to scrape Agincare data?

This actor scrapes publicly available content from Agincare’s website.

However:

• Always review Agincare’s Terms of Service
• Ensure compliance with applicable data protection laws (e.g. GDPR)
• The responsibility for usage lies with the user.