Someone on your team says “use Claude.” You open a browser, land on a chat box, and feel fine for twenty minutes. Then a coworker says they shipped a pull request with Claude, someone else mentions Cowork reorganized a folder of vendor PDFs, and a third person forwards a Console invoice that is not the same bill as Pro. Same brand. Four different doors. That mismatch is the whole reason this series exists.
This is Part 1 of Claude product map. If you finished Learn Claude from scratch, you already know company vs model vs product, Free vs paid plans in broad strokes, and when chat is the wrong tool. Now we map the product family so you stop treating every Claude surface like one chat window with extra marketing.
What you’ll learn
- How Claude chat on web, desktop, and mobile differs from agent-style products
- Where Claude Code, Claude Cowork, Claude Design, and Claude Science fit in plain jobs language
- Why the API Console is a separate money and risk story
- A chooser you can use in a meeting when someone says “Claude” without naming a surface
- Common mix-ups that waste seats, burn limits, or dump bad work into production
Product names and plan bundling change. Treat what follows as a July 2026-era map, and re-check claude.com/pricing and product pages before you buy or write policy. As of writing, the public product list includes chat surfaces, Claude Code, Claude Cowork, specialized products such as Design and Science, and a developer platform with its own Console.
Why one brand needs a map
Anthropic ships models. You almost never buy “a model” as a civilian. You open a product that uses one. That is good for UX and bad for vocabulary. People collapse everything into “Claude” the way they once collapsed everything into “the cloud.”
At work the collapse costs real money and real trust. IT may allow browser chat under a Team plan and still block agent tools that can edit local folders. Finance may approve a Pro seat and reject a surprise Console spend. An analyst may paste a CSV into chat while an engineer expects multi-file edits in a repo. If you only know the chat box, you will either underuse useful tools or overtrust them in the wrong place.
The fix is boring and durable: name the door. “Claude chat on claude.ai.” “Claude Code in VS Code.” “Cowork on a folder on my Mac.” “API calls billed in Console.” Once the door is named, plan fit, permission risk, and review habits become discussable.
Claude chat: the default door most people mean
When someone says they “asked Claude,” they usually mean a conversational assistant. As of writing, that lives at claude.ai in the browser, in desktop apps (Mac, Windows, Linux entry points on Anthropic’s download pages), and in mobile apps for iOS and Android. Same account family in many setups. Same general habit: you write a message, maybe attach a file, maybe use a project or memory feature, and Claude replies in the thread.
What chat is good at
Chat is the right tool when the unit of work is a conversation. Draft an email. Critique a slide outline. Explain a policy PDF. Turn messy notes into a checklist. Brainstorm five options, then ask for the strongest two with risks listed. Translate jargon for a specific audience. Rewrite for tone. Help you think before you open a spreadsheet or a ticket.
Chat is also where most people should start if they are new. You can learn prompting, verification, and “paste less private data” habits without giving an agent write access to a repo or a folder of contracts. That sequencing matters more than any feature comparison chart.
What chat is bad at (or awkward for)
Chat is awkward when the unit of work is a tree of files that must stay consistent. You can paste snippets, but you are the glue. You copy, you apply, you re-paste errors, you hope you did not miss a second file that imports the first. Chat is also awkward when the job is “open these twelve spreadsheets, rename them, build a deck, and leave artifacts on disk.” You can force it with downloads and uploads. You will feel the friction.
Chat is still not a source of record. If the answer becomes a board number, a customer commitment, or a SQL change that hits production, a human owns the check. That rule does not change because the product has a prettier UI.
Web, desktop, mobile: pick by where the work lives
Web is the lowest-friction start: any machine with a browser, IT can often allow-list the domain, and you are not installing software. Desktop is useful when you want a dedicated window, local file habits that the app supports, or less browser tab chaos. Mobile is for capture and light review: dictate a thought on the train, reread a draft, kick something that continues elsewhere when the product supports that. As of writing, Anthropic markets chat across those surfaces; exact feature parity still moves, so treat mobile as “great for many tasks, not always identical to desktop.”
Rule of thumb: If the job is thinking, drafting, or explaining, open chat first. If the job is changing a codebase or completing multi-step work inside files and tools you choose, you are probably looking at an agent product, not a longer chat thread.
Chat vs agents: the split that actually matters
Marketing will sell you “smarter Claude.” Your workflow cares about agency. In chat, Claude mostly responds inside the thread. In agent-style products, Claude plans steps, uses tools, reads and writes files (with permission), and can keep going while you step away. That is a different risk profile and a different review habit.
Think of chat as a sharp colleague on a video call. Think of agent products as that colleague sitting at your keyboard under rules you set. You still review. You just review outcomes and diffs, not only paragraphs.
Two agent-shaped products dominate everyday confusion right now: Claude Code and Claude Cowork. Official product pages describe them as siblings with different primary jobs. Code is built for software work in repositories and developer tooling. Cowork is built for knowledge work on folders, documents, spreadsheets, research, and office tools. Same family idea of “hand off a goal and review the result.” Different default worlds.
Claude Code in one honest paragraph
Claude Code is Anthropic’s coding agent. As of writing, it shows up in the terminal, in VS Code and JetBrains, in Claude’s desktop app, on the web at claude.ai/code, in Slack integrations, and on mobile entry points Anthropic documents for coding work. Plan access commonly ties to Pro, Max, and team/enterprise seating (verify live pricing pages; bundling has shifted more than once). In practice people use it to explore a repo, make multi-file edits, run tests, and open pull requests, with permission prompts before it changes things or runs commands. Part 2 of this series goes deep on “what it is for.” The install and first-run walkthrough lives in the later Claude Code tutorial series, not here.
If you never touch a codebase, you can still understand Code as “Claude with hands inside software projects.” You do not need to become a full-time engineer to benefit from that mental model when an engineer asks you to review a PR description they drafted with Code.
Claude Cowork in one honest paragraph
Claude Cowork is the agent for knowledge work. As of writing, Anthropic positions it for goals across files and tools you choose: research, decks, spreadsheets, organizing folders, multi-step office tasks. Official materials describe paid-plan access (Pro and above in consumer language; team and enterprise stories exist too) and emphasize review, permissions, and visibility of steps. Cowork is not the coding-first product. If your job is mostly docs, analysis packs, ops admin, and “make this pile into something I can send,” Cowork is the door to learn next after chat. Part 3 of this map series is dedicated to it.
Design and Science: specialized products, not chat skins
As of writing, Anthropic also lists specialized products such as Claude Design and Claude Science on claude.com. Treat them as purpose-built surfaces for visual design collaboration and science-oriented workflows, not as “chat with a different wallpaper.” Exact feature sets move quickly (Labs-style launches and renames happen). For this map, the useful rule is: if your work is mainly design artifacts or scientific research pipelines, check whether a specialized product is available on your plan before you force everything through generic chat.
You do not need to master every specialized product on day one. You need to know they exist so a teammate’s “Claude did the mock” does not get misread as “someone typed into claude.ai.” Later parts of this series (Design and Science, model chooser, API) fill those corners. The landscape diagram above is enough for orientation.
API Console: same models, different store
Developers and product teams call Claude through an API. Anthropic’s developer home and docs live under the Claude platform story (Console login and docs on Anthropic’s developer sites; product navigation on claude.com points at Platform / API). Billing for API usage is not the same mental model as a consumer Pro chat subscription. You pay for tokens and platform usage under Console billing, rate limits, and org settings. You also take on engineering responsibilities: keys, logging, retention, evals, and user-facing safety.
If you only write emails and decks, you can ignore Console until IT or engineering invites you into that world. If someone asks you to “just put Claude in the product,” you are no longer in chat-map territory. You are in build territory. Part 6 of this series covers when the API path is worth it. For now, keep one hard line in your head: chat seat ≠ API meter.
A chooser table for real Mondays
Use this when a meeting gets fuzzy. Pick the row that matches the job, not the hype.
| Job you actually have | Default door (as of writing) | Why |
|---|---|---|
| Draft, rewrite, explain, brainstorm in a thread | Claude chat (web, desktop, or mobile) | Conversation is the unit of work; low setup |
| Explore a repo, edit many files, run tests, open a PR | Claude Code | Built for software projects and developer surfaces |
| Multi-step office work on folders, docs, sheets, research packs | Claude Cowork | Agent for knowledge work, not coding-first |
| Design collaboration / science-specialized workflows | Design or Science (if available on your plan) | Specialized products; check current product pages |
| Ship Claude inside your app or backend | API + Console | Separate billing, keys, and engineering ownership |
| Unclear, sensitive data, or high liability | Slow down: policy + chat sandbox first | Do not grant folder or repo access by default |
Two people can share the same brand subscription and still need different doors for different hours of the day. An ops lead might live in chat and Cowork. An engineer might live in Code and only open chat for prose. That is normal. Shared vocabulary beats shared habits.
Plans, limits, and “included with” language
As of writing, Anthropic markets Claude Code and Claude Cowork as included with certain paid consumer plans (Pro and Max language appears on product pages) and with team/enterprise seating stories that have changed over 2025-2026. Usage limits still apply. Agentic work often burns capacity faster than a short chat. That is not a moral failing. It is a product reality: multi-step tool use costs more work than one reply.
Before you roll out a team habit, write down three facts on one page:
- Which products each seat type includes today (screenshot the pricing page; dates help later).
- Whether Console API spend is allowed, who owns keys, and what the monthly alert threshold is.
- What data is banned from personal accounts even if a feature would make the work faster.
If you skip that page, you will learn the rules from a surprise invoice or a compliance email. Neither is a fun teacher.
Worked example: four people, one phrase
Imagine a 30-minute standup. The product manager says, “Let’s use Claude on the launch.” Without a map, four people walk away with four plans.
Alex (PM) opens claude.ai, pastes the brief, and asks for a timeline critique. Correct door for that slice of work.
Jordan (engineer) opens Claude Code against the repo, asks for a dark-mode toggle that matches existing patterns, reviews the diff, runs tests, opens a PR. Correct door.
Sam (ops) points Cowork at a folder of partner PDFs and asks for a one-page risk table plus renamed files. Correct door if policy allows folder access.
Riley (data) starts wiring an internal tool to call the Messages API through Console because the team wants Claude inside a support console. Correct door only if engineering owns that path; wrong door if Riley meant “chat with me about the SQL.”
The phrase “use Claude” was incomplete. The fixed version is longer and clearer: “Alex: chat for the brief. Jordan: Code for the UI change. Sam: Cowork for the partner pack if legal says yes. Riley: no API work this week unless eng is in the thread.” That sentence saves two days of crossed wires.
Common mistakes
Treating Code as “chat that writes code”
You can paste code into chat. Claude Code is still a different product: repo context, multi-file edits, terminal/IDE surfaces, PR workflows. If you only ever paste snippets into chat, you are not “doing Code.” You are coding with chat. That can be fine. Call it what it is.
Treating Cowork as “chat with file upload”
Uploading a PDF to chat is not the same as granting an agent a folder and a goal. Permissions, autonomy, and review loops differ. If your org allows chat but bans local-folder agents, those are different policy lines for a reason.
Mixing Console spend with chat seats
A Pro subscription does not automatically mean unlimited API apps for your product. Console billing can show up as a separate line. Put ownership on a named human before anyone pastes an API key into a side project on a laptop.
Assuming mobile has every desktop agent feature
Mobile is excellent for many chat tasks and for checking agent progress when the product supports it. It is a weak place to invent your whole security model. Start policy from desktop/web realities, then layer mobile.
Skipping verification because the product feels “agentic”
More steps completed by software does not mean more truth. It means more surface area to review. Diffs, numbers, names, and “did this touch the wrong folder?” still sit with you.
What stays the same across every door
Maps can make products feel more different than they are. A few constants still hold wherever you click.
Models still predict text under product limits. A prettier agent UI does not turn Claude into a database of your private company facts. If the fact was not in the prompt, the files you attached, the tools you connected, or reliable retrieval your org built, treat detailed “memory of last quarter’s numbers” with suspicion until you verify.
Humans still own liability. Chat draft, Code PR, Cowork deck, API response in a customer-facing tool: if it ships wrong, your name is still on the outcome. Agents increase speed and surface area. They do not absorb responsibility.
Data handling still matters. Personal Free/Pro accounts, team seats, and enterprise controls are different trust boxes. Paste rules from Learn Claude and from your IT policy still apply when you open Code or Cowork. Folder access is not a free pass to drop payroll exports into a home laptop account.
Names still change. Model labels, plan bundles, and which seat includes which agent have moved multiple times. Date your notes. Prefer screenshots of the pricing page over tribal knowledge from a Slack thread last spring.
FAQ: short answers you will need in meetings
Is Claude Code “better Claude”?
No. It is a different product for software work. For writing a client email, chat is usually better. For multi-file repo changes, Code is usually better. “Better” only makes sense next to a job.
Do I need Cowork if I already upload files in chat?
Not always. Uploads in chat handle many one-off reads. Cowork earns its keep when the job is multi-step work on folders and tools, with an agent completing work for review rather than only answering in a thread. If your org has not approved folder agents, stay on chat.
Is the API just Pro with a developer skin?
No. Console billing, keys, logging, and product integration are a separate store and a separate risk model. Building “Claude into our app” is an engineering project, not a settings toggle on a chat seat.
Should every teammate install every product?
No. Start with chat fluency. Add Code for people who touch repositories. Add Cowork for people who run multi-step file and office workflows under policy. Specialized Design/Science products only if those jobs are real for your team. Unused agents still create permission questions and support load.
How to practice this week
- Write down the last three times someone said “Claude” at work. Label each with a door: chat, Code, Cowork, specialized, or API/unknown.
- Open Claude’s product overview and pricing. Note what your actual plan includes today. Date the note.
- Pick one real task you will do tomorrow. Force yourself to choose a door before you open a tab. If you choose chat, write one sentence on why Code or Cowork would be worse for that task.
- If you have never used anything beyond chat, stop here until Part 2 and Part 3. Do not install every agent on day one “just in case.”
Where this series goes next
Part 2 is Claude Code: what it is for (jobs, surfaces, and honest non-goals for people who will touch a project). Part 3 covers Cowork. Later parts cover Design and Science, model names without hype, and the API/Console path for builders. Deep install tutorials for Code and Cowork live in their own series after this map is clear.
If you still want everyday chat habits before you expand products, stay with Learn Claude from scratch. If you want broader AI workflow hygiene that is not vendor-specific, the Practical AI series sits beside this track.
Quick recap
- One brand, many doors: chat, Code, Cowork, specialized products, API Console.
- Chat is the default for conversation work on web, desktop, and mobile.
- Agent products change files and use tools under permissions; review outcomes, not only prose.
- Code is coding-first. Cowork is knowledge-work-first. Design and Science are specialized.
- Console billing is separate from chat seats. Name an owner before keys appear.
- When someone says “use Claude,” answer with a door name, not a vibe.
Sources
Official product pages and docs used for this map (re-check before purchase or policy; surfaces change):
- Claude product overview: https://claude.com/product/overview
- Claude Code product page: https://claude.com/product/claude-code
- Claude Code documentation overview: https://code.claude.com/docs/en/overview
- Claude Code on the web: https://claude.ai/code
- Claude Cowork product page: https://claude.com/product/cowork
- Claude Design product page: https://claude.com/product/design
- Claude Science product page: https://claude.com/product/claude-science
- Claude download (desktop and app entry points): https://claude.com/download
- Claude plans and pricing: https://claude.com/pricing
- Claude Platform / API overview: https://claude.com/platform/api
- Developer docs home: https://platform.claude.com/docs
- Anthropic company site: https://www.anthropic.com/
- AMS Learn Claude series: https://analyticsmadesimple.com/series/learn-claude/
- AMS Practical AI series: https://analyticsmadesimple.com/series/practical-ai/
