The Claude Product Ecosystem¶

Level: Beginner · Time: ~45 min · Exam relevance: CCAO-F D3 (3.1), D4 (4.4) · CCAR-P D3 (3.7), D7 (7.1)

In plain English¶

"Claude" can mean several different things depending on who is talking:

A marketing manager usually means the Claude app at claude.ai or on their laptop or phone.
A software developer might mean Claude Code, the coding tool that works in a terminal or editor.
An application developer usually means the Claude API: sending requests from their own software.
A cloud architect might mean Claude running inside Amazon Bedrock, Google Cloud Vertex AI or Microsoft Foundry.

Underneath all of these sit the same Claude models (covered in Model families). The products are different doors into the same building, each with its own furniture: Projects, Artifacts, research, connectors, Skills and so on.

This page is a map. Each feature gets one or two paragraphs here and a full page elsewhere.

Facts on this page are as of 2026-10-24 and were checked against support.claude.com, claude.com/pricing, platform.claude.com and code.claude.com. Plans and features change; check those pages before relying on a detail.

Why it matters¶

CCAO-F objective 3.1 asks you to choose the right Claude product feature for a task (Projects, research, chat, Artifacts). Picking the wrong one wastes time and usage: running a quick question through research, or re-uploading the same handbook into every chat instead of using a Project.

For architects, CCAR-P objective 3.7 asks you to choose an integration mechanism, and 7.1 to configure Claude tools for teams. Knowing which surface fits which need, and what each one costs and controls, is the starting point for both.

How it works¶

The map¶

flowchart TB
    M[Claude models: Haiku, Sonnet, Opus, Fable]
    subgraph Apps["Claude apps (web, desktop, mobile)"]
        CH[Chat] --- PR[Projects]
        PR --- AR[Artifacts]
        AR --- RS[Research]
        RS --- CO[Connectors]
        CO --- SK[Skills]
        SK --- ME[Memory and chat search]
    end
    subgraph Dev["Developer surfaces"]
        CC[Claude Code: terminal, IDE, desktop, web]
        API[Claude API + Claude Console]
    end
    subgraph Cloud["Cloud platforms"]
        BR[Amazon Bedrock]
        VX[Google Cloud Vertex AI]
        FD[Microsoft Foundry]
        CPA[Claude Platform on AWS]
    end
    M --> Apps
    M --> Dev
    M --> Cloud

The Claude apps¶

The consumer and business apps are where most CCAO-F work happens.

Surface	Where	Notes
Web	claude.ai	Full feature set; nothing to install
Desktop	macOS and Windows (Linux in beta), from claude.com/download	Adds desktop extensions (local integrations) and includes Claude Code in a Code tab on paid plans
Mobile	iOS and Android	Chat, voice, photos of documents; some features behave differently (for example, research started on mobile searches only the web)

The same account, chats and Projects follow you across all three.

Plans¶

Plan	Who it is for	Notable differences (as of 2026-10-24)
Free	Trying Claude	Chat, web search, up to five Projects, memory, connectors; no research; no Claude Code
Pro	Individuals using Claude regularly	More usage, research, Claude Code, access to Opus models, larger Project knowledge via retrieval
Max	Heavy individual users	Several times Pro's usage, priority at busy times
Team	Groups of colleagues (minimum seats apply)	Central billing, single sign-on, admin controls for connectors, shared Projects
Enterprise	Larger organisations	Team features plus role-based access, SCIM provisioning, audit logs, compliance API, custom data retention and other controls

There are also education plans. See claude.com/pricing for prices and the full comparison.

As of 2026-10-24, the plan comparison on claude.com/pricing matches the table for Projects, research, Claude Code, Opus access, priority access, single sign-on, SCIM, audit logs, the compliance API, custom data retention and role-based access. That page lists Team for 2 to 150 seats and Enterprise with a 20-seat minimum. The split changes often, and some features are in beta or limited to selected users, so check the page before relying on a detail.

App features¶

Feature	What it does	Use it when	Details
Chat	A single conversation	Quick questions, one-off drafts	Your first conversation
Projects	A workspace with its own instructions, knowledge files and chat history	You return to the same topic, documents or style repeatedly	Projects and knowledge
Artifacts	Standalone outputs (documents, code, diagrams, small interactive tools) that open beside the chat and can be edited and shared	The output is substantial and you will edit, reuse or share it	Artifacts and output formats
Research	Claude runs a series of searches, each informed by the last, across the web and enabled connectors, and reports back	Questions that need many sources and some digging	Below
Web search	Claude searches the web during a normal chat	Current information, quick fact-finding	Toggle in the chat menu
Connectors	Let Claude read from and act in other apps (Google Drive, Gmail, Calendar, Slack, and many others) with your permissions	The information lives in another tool, or you want Claude to take an action there	Connectors and integrations
Skills	Folders of instructions, scripts and resources that Claude loads only when relevant	Repeatable specialised tasks (branded documents, a company analysis workflow)	Below
Memory and chat search	Carry context from earlier chats into new ones	You want continuity without re-explaining	Projects and knowledge
Incognito chats	Chats not saved to history or memory	One-off questions you don't want kept	Responsible use

Research in more detail. You turn it on from the + menu; web search must also be on. Claude plans a line of enquiry, searches, reads, and decides what to look for next, which takes longer and uses more of your plan's limits than a normal chat. As of 2026-10-24, the Help Center says research is available on Pro, Max, Team and Enterprise plans, runs on an Opus model regardless of the model selected for the chat, and on web and desktop can use Google Workspace and other enabled connectors automatically, without asking for approval each time (Using research). That last point matters for data governance: enable only the connectors you are happy for research to search.

Skills in more detail. A Skill is a folder containing instructions (written in Markdown) and optionally scripts and resources. Claude checks which Skills are available, loads only the ones relevant to the current task, and applies them, a pattern called progressive disclosure that keeps the context window uncluttered. Anthropic provides built-in Skills (for example, for working with Excel, Word, PowerPoint and PDF files); individuals and organisations can create custom Skills; Team and Enterprise owners can provision Skills to everyone. Skills need code execution to be enabled. See What are Skills? and, for the developer view, Progressive disclosure and Skills.

Claude Code¶

Claude Code is an agentic coding tool: it reads a codebase, edits files, runs commands and works with tools such as git, all under permissions you control. It runs in the terminal, in IDEs (VS Code and JetBrains), in the desktop app and on the web, and most surfaces need a paid Claude subscription or a Claude Console account. It is covered in Module 09.

The Claude API and Claude Console¶

Developers call Claude from their own software through the Claude API (the Messages API and related endpoints), using official SDKs for Python, TypeScript, Java, Go, Ruby, C#, PHP and others. The Claude Console at platform.claude.com is where you create API keys, manage workspaces, test prompts and view usage and billing. API usage is billed per token, separately from app subscriptions. Module 06 covers the API.

Claude on cloud platforms¶

Organisations that already run on a major cloud can use Claude through that provider, which keeps billing, identity, networking and data-residency controls in one place.

Platform	Notes (as of 2026-10-24)
Amazon Bedrock	AWS-operated. Model IDs carry an `anthropic.` prefix (for example `anthropic.claude-opus-5-5`). Pricing set by AWS
Google Cloud Vertex AI	Google-operated (listed in Anthropic's docs as Claude Models on Google Cloud, Google managed). Uses the plain model IDs. Pricing set by Google Cloud
Microsoft Foundry	Billed through the Microsoft Marketplace at Anthropic's standard API rates
Claude Platform on AWS	Anthropic-operated, using Claude API model IDs and features, accessed through AWS

Feature availability differs between the Claude API and each cloud platform: some newer or beta features reach the Claude API first. Architects should check per-feature availability before committing. See Cloud platforms and the models overview for platform IDs.

Choosing the right feature¶

flowchart TD
    A[What do you need?] --> B{Writing or code in your own software?}
    B -- Yes --> C{Coding in a repository?}
    C -- Yes --> CC[Claude Code]
    C -- No --> API[Claude API or a cloud platform]
    B -- No, using the app --> D{Will you return to the same documents, instructions or topic?}
    D -- Yes --> P[Create a Project]
    D -- No --> E{Needs many sources and investigation?}
    E -- Yes --> R[Research]
    E -- No --> F{Information lives in another app, or you want an action there?}
    F -- Yes --> CN[Enable the relevant connector]
    F -- No --> G[Plain chat, with web search if current facts are needed]
    P --> H{Substantial output to edit or share?}
    G --> H
    R --> H
    H -- Yes --> ART[Ask for an Artifact]

These choices combine: a research task inside a Project that produces an Artifact is common.

Example¶

Five tasks and the best starting point for each:

Task	Best feature	Why
"Reword this sentence to sound less formal"	Plain chat	One-off, small
Weekly status reports in the same format, using the same project charter	Project with the charter as knowledge and the format in instructions	Reused context and format every week
"What are competitors saying publicly about their pricing changes this quarter?"	Research	Many sources, current information, investigation
"Find last month's email from the supplier about delivery dates and draft a reply"	Gmail connector	The information lives in email
A one-page interactive budget calculator to share with the team	Artifact	Substantial, reusable, shareable output

Enterprise use case¶

Contoso Health is planning its Claude rollout. Different groups need different doors:

Administrative and communications staff use Claude Enterprise apps, with shared Projects for the style guide and policy library and the Google Drive connector enabled read-only for policy folders.
The software team uses Claude Code for internal tools, configured with organisation settings and permissions (CCAR-P 7.1).
The patient-portal team builds a scheduling assistant on the Claude API, accessed through the cloud platform the organisation already uses, so identity, logging and data residency follow existing controls.

One decision record explains the split, so later teams know which door to use (CCAO-F 4.4, integrating Claude into existing workflows).

Common mistakes¶

Using one feature for everything. A single endless chat instead of Projects; research for simple lookups.
Assuming every feature is on every plan or platform. Research, chat search, larger Project knowledge and some connectors depend on plan; API features vary across cloud platforms.
Confusing subscriptions with API billing. A Pro or Team subscription does not pay for API calls from your own software (Claude Code usage on paid plans is the exception; it shares the plan's usage).
Ignoring admin controls. On Team and Enterprise, owners decide which connectors, Skills and features members can use.

Security considerations¶

Each surface has different controls. Enterprise apps offer audit logs and retention settings; the API offers workspace-level keys and usage monitoring; cloud platforms integrate with existing identity and network controls. Choose the surface whose controls match the data.
Connectors and research can reach into other systems with the user's permissions. Least privilege applies (Connectors and integrations).
Custom Skills and custom connectors are code and configuration from someone else. Review them before enabling, just as you would a browser extension.

Troubleshooting¶

Symptom	Likely cause	What to do
A feature mentioned here is missing	Plan, region, beta access or an admin setting	Check the Help Center article for that feature; ask your administrator
Research didn't search your Google Drive	Started on mobile, or connector not enabled	Start research on web or desktop with the connector enabled
API feature works on the Claude API but not in your cloud account	Feature not yet available on that platform	Check the platform availability notes in the API docs

Check your understanding¶

Name the three app surfaces and two developer surfaces for Claude.
You produce the same monthly board summary from the same three reference documents. Which feature should you use first?
What does research do differently from a normal chat with web search?
Why might an organisation use Claude through a cloud platform rather than the Claude API directly?
What is progressive disclosure in the context of Skills?

Answers

Apps: web, desktop, mobile. Developer surfaces: Claude Code and the Claude API (with the Claude Console).
A Project, with the reference documents as knowledge and the summary format in the instructions.
It runs a series of searches, each shaped by what it found so far, across the web and enabled connectors, and compiles a report. It takes longer and uses more of your limits.
To keep billing, identity, networking, logging and data-residency controls inside the cloud environment it already uses.
Claude loads only the Skills relevant to the current task, rather than all of them, which keeps the context window focused.