Key takeaways
- Claude runs as a web app, desktop apps, mobile apps and Claude Code; all store conversations in your Anthropic account, not only on your device.
- The API is the HTTP interface to Claude; the SDKs are official client libraries that call it for you, with retries and types built in.
- An 'overloaded' error is temporary capacity pressure; retry with exponential backoff, which the SDKs already do.
- Apify connects to Claude through an MCP server, so Claude can run Apify's scrapers as tools.
- Let Claude tailor applications, but submit them yourself; automated mass applying breaks site rules and hurts your chances.
The Claude app is Anthropic's assistant on the web at claude.ai, on desktop for macOS and Windows, and on iOS and Android; Claude Code is its coding tool for the terminal and IDEs. Developers reach the same models through the Claude API, usually with Anthropic's official SDKs. This page covers the apps and where your data lives, the API and SDKs and how they differ, what the "overloaded" error means, how to connect Apify's scrapers to Claude, whether that is free, and whether Claude can apply to jobs for you.
What is the Claude app?
The Claude app is Anthropic's AI assistant, available in a browser at claude.ai and as desktop and mobile apps. You chat with it, upload files and images, create documents and small apps in a side panel, and connect it to tools.
Also asked as: what is claude app
All versions share one account, so conversations and projects appear everywhere you sign in. The desktop app adds conveniences such as a quick-launch shortcut and connecting local tools through the Model Context Protocol. The mobile app is good for voice and quick questions. Claude Code is a separate tool for developers that works directly on code in your terminal or IDE. The same family of Claude models powers all of them, with plans that set your usage limits.
What are the Claude apps?
The official Claude apps are the web app at claude.ai, desktop apps for macOS and Windows, mobile apps for iOS and Android, and Claude Code for the terminal and IDEs such as VS Code and JetBrains. All are published by Anthropic.
Also asked as: what are claude apps
Each suits a different job:
- Web and desktop: writing, research, analysing documents and spreadsheets, projects with shared files.
- Mobile: voice conversations, photos of notes or whiteboards, quick questions.
- Claude Code: reading a codebase, editing files, running tests and commands.
Inside the chat apps, "artifacts" are small documents or apps Claude creates that you can preview, edit and share. Be careful with third-party apps that use the Claude name: unless the publisher is Anthropic or a known IDE vendor, avoid entering your credentials.
Does the Claude app keep all data locally?
No. The Claude apps store your conversations in your Anthropic account on Anthropic's servers, so they sync across devices. The desktop app is a client, not an offline model; nothing runs fully on your computer.
Also asked as: does claude app keep all data locally
What stays local depends on what you connect: files on your computer that you expose through a local MCP server are read on your machine and only the parts Claude uses are sent with the conversation. Anthropic's privacy settings let you control whether your chats may be used to improve models, and you can delete conversations. Check the privacy settings once and choose deliberately. If you need data to stay entirely on your own hardware, you need a local open-weight model, not Claude.
What are the Claude APIs?
The Claude API is Anthropic's HTTP interface for using Claude models in your own software. The core is the Messages API: you send a list of messages and settings, and get Claude's reply, with options for streaming, tool use, images, PDFs and structured outputs.
Also asked as: what are claude apis
Around it are supporting APIs: token counting, the Message Batches API for large asynchronous jobs at lower cost, a Files API for reusing uploaded documents, and features such as prompt caching to cut cost on repeated context. You authenticate with an API key from the Anthropic Console and pay per token used, separately from any chat subscription. The API is what you use to build a chatbot into a college project, add document Q&A to an app, or run an agent on a server.
What is the Claude SDK?
The Claude SDKs are Anthropic's official client libraries for calling the API from code, available for Python, TypeScript and several other languages. There is also the Claude Agent SDK, for building agents on the same foundations as Claude Code.
Also asked as: what is claude sdk
A minimal Python example:
from anthropic import Anthropic
client = Anthropic() # reads ANTHROPIC_API_KEY from the environment
msg = client.messages.create(
model="claude-sonnet-5-5",
max_tokens=500,
messages=[{"role": "user", "content": "Explain BM25 in two sentences."}],
)
print(msg.content[0].text)
Install it with pip install anthropic. Check the model overview in the docs for the current model names, since new versions replace old ones regularly.
What is the difference between the Claude API and SDK?
The API is the service: HTTP endpoints that accept requests and return Claude's responses. An SDK is a library in your programming language that calls those endpoints for you, handling authentication, retries, streaming and typed responses.
Also asked as: difference between claude api and sdk
You can call the API directly with any HTTP client, which is useful for languages without an SDK or for quick tests with curl. In practice, the SDK saves effort and prevents mistakes: it retries temporary errors automatically, parses streaming events correctly, and gives editor autocompletion for parameters. Both reach the same models with the same limits and prices. Use the SDK for real projects and raw HTTP only when you have a reason.
What does "Claude API is overloaded" mean?
It means Anthropic's servers are temporarily under heavy load and could not process your request. The API returns an overloaded_error, with HTTP status 529, and the right response is to wait briefly and retry.
Also asked as: what does claude api is overloaded mean
It is not a problem with your key, code or account, and it is different from a rate-limit error, which means you exceeded your own limits. Handle it with retries using exponential backoff and a little randomness, so many clients do not retry at the same instant. The official SDKs already retry such errors a couple of times by default, and you can raise the retry count. In production, also add a fallback, such as a smaller model or a different provider, so users see a slower answer rather than an error. RAG.NextUpgrad uses multi-provider fallback for exactly this reason.
How do I connect Apify to Claude?
Apify provides an MCP server that exposes its scrapers, called Actors, as tools. Add the Apify MCP server to Claude, through the connectors settings or the desktop app's MCP configuration, with your Apify API token, and Claude can run Actors and read their results.
Also asked as: how to connect apify to claude
Once connected, you can ask Claude, for example, to collect the titles and links of recent posts from a public page using a suitable Actor, then summarise them. Keep three things in mind: scrape only public data you are allowed to collect and respect each site's terms; store your Apify token like a password; and limit which Actors are exposed so Claude cannot run expensive jobs by accident. Apify's documentation has current setup instructions for its MCP server.
Is Apify for Claude free?
Connecting Apify to Claude through its MCP server does not need a special paid add-on, but running Apify Actors uses your Apify account's platform usage. Apify has a free plan with a monthly usage allowance, and heavier use needs a paid plan.
Also asked as: is claude apify free
Your Claude plan applies on the Claude side as usual. Costs depend on the Actors you run, how many pages they process and how often; some Actors are maintained by third parties and have their own pricing. Start with small test runs and check usage in the Apify console before scheduling anything recurring. Apify's pricing page lists the current free allowance and plans.
Can Claude apply to jobs for me?
Claude can do most of the work around applying: tailoring your resume to each job description, drafting cover letters and referral messages, and preparing you for interviews. You should still submit applications yourself.
Agent tools that control a browser can technically fill in forms, but automated applications on LinkedIn, Indeed and similar sites can break their terms of service and get accounts restricted. Mass-applying also produces poor matches that recruiters ignore. A better use of Claude: "Here is the job description and my resume. List the gaps, rewrite my project bullets using only facts from my resume, and draft a 100-word note to an engineer at the company." Then read, edit and send it yourself. Quality and a referral beat volume.
Sources
GenAI engineer and AI product builder with 2+ years shipping production-grade AI systems: RAG pipelines, fine-tuned LLMs, hybrid retrieval and multi-modal apps across vision, speech and OCR, architected end to end from ingestion to deployment. Leads engineering for SCULT INDIA's 14-member team, founded the 500+ member TechVerse Enclave community and has mentored 200+ students. Three hackathon first prizes: Changethon 2025 (IIT Roorkee), Product Genesis at Vividhotsava 2025 (CSJMU) and BYTEBATTLE (MeetKats).
Open to GenAI roles, hackathon judging, mentorship sessions and guest talks at colleges. For campus sessions the inviting institution covers the visit end to end: travel (a First AC train ticket for journeys up to five hours, a flight for anything longer), accommodation, meals, and pickup and drop; the honorarium and workshop fees are negotiable on WhatsApp or on the call, and requests are welcome from institutions that can cover the visit this way. Email pranjulrathour41@gmail.com.
