Claude for salespeople
We describe what Claude does in B2B sales work, on which plan and under what conditions, and go no further than what Anthropic confirms in its documentation. Every fact on this page has a source at the bottom.
As of August 12, 2026
In 30 seconds
What Claude does in a salesperson’s work, and which of those things only the organization’s Owner can unlock.
- The name “Claude” alone doesn’t tell you what you get. Features depend on the plan, and on business plans also on what the organization’s Owner decides.
- Continuity on a single client comes from a project: its own knowledge base, its own instructions and separate memory. Projects work on every plan, with a limit of five on the free plan.
- A project shared with the whole team requires a business license. On individual plans, a project stays with one person.
- Pre-meeting briefs come from your work inbox and calendar. The Microsoft 365 and Google Workspace connectors pull email, documents and calendar entries within the permissions you have in those systems.
- Follow-ups come out as drafts. The Gmail connector and the Outlook add-in only create drafts; only the Microsoft 365 connector can send, after an admin turns on its write tools.
- Your team’s standard can be saved as a procedure. Skills are packages with process instructions, and on business plans the Owner can roll them out to everyone at once.
- Voice mode is in beta and can’t reach projects or Cowork skills.
Plans: individual and business tiers
What you get depends on the plan, and on business plans also on what the admin decides. We keep the two layers apart below, because they’re aimed at different people and affect the team differently. We don’t quote prices, because pricing and plan contents changed at all four vendors in the last six months. Current rates are on Anthropic’s site.
Individual plan
Free, Pro and Max: a salesperson’s personal account
A personal account is enough for research, call prep and work on a proposal. Chat runs on the web, iOS, Android and in the desktop app.
The free plan holds up to five projects. Paid plans lift that limit and add a mechanism that expands a project’s knowledge capacity up to ten times as you approach the context limit.
Two things to know before you buy: the Pro plan doesn’t include API access in the Claude Console (that’s billed separately), and the Max plan is available with monthly billing only. Prices in the mobile app stores may differ from a subscription bought in the browser.
| What you get | Free | Pro | Max |
|---|---|---|---|
| Projects | up to five | unlimited | unlimited |
| Expanded project knowledge for a large knowledge base | - | yes | yes |
| Search across past chats | - | yes | yes |
| Skills | yes | yes | yes |
| Slack and Google Workspace connectors | yes | yes | yes |
| Custom connectors via remote MCP | one | yes | yes |
| Cowork, Research and Office add-ins | - | yes | yes |
| Claude in Chrome extension (beta) | - | yes | yes |
| Voice mode (beta) | yes | yes | yes |
| Usage per session window | baseline | at least 5x the Free plan | a multiple of Pro, in two tiers |
| Annual billing | - | yes | monthly only |
A dash means Anthropic’s documentation doesn’t list that item for the plan. Anthropic doesn’t publish limits as a number of messages, only multipliers relative to the Pro plan and reset windows.
Business plan
Team and Enterprise: a team with an organization Owner
The Team plan starts at two people and tops out at 150 seats; above that you have to move to Enterprise. Self-serve Enterprise starts at 20 seats, and Enterprise bought through Anthropic’s sales team starts at 50 seats, with invoicing.
A business license changes three things at once: shared projects and skills for the whole team, admin control over connectors and Cowork, and the default setting for training models on the organization’s content.
A billing note for Enterprise: the seat fee covers access only, and all usage is billed separately, with no per-plan or per-seat limits. The admin sets spending controls for the organization and for individual users.
| What you get | Team | Enterprise |
|---|---|---|
| Model training on the organization’s content | off by default | off by default |
| Shared projects with view or edit access | yes | yes |
| Skills rolled out to the whole team by the Owner | yes | yes |
| SSO, domain capture and just-in-time provisioning | yes | yes |
| SCIM provisioning, audit logs, Compliance API | - | yes |
| Custom retention policy and role-based permissions | role-based permissions | yes |
| Network-level access control, customer-managed encryption keys, US-only inference, HIPAA-ready configuration | - | yes |
| Claude in Slack through Claude Tag (public beta) | yes | yes |
| Seats | from 2 to 150 | from 20 when bought self-serve, from 50 with invoicing |
| Usage limits | per person, with a session reset and a separate weekly limit | no per-plan or per-seat limits, usage billed separately |
| Context window | 200K tokens | - |
A dash means Anthropic’s documentation doesn’t list that item for the plan.
A business license doesn’t switch everything on by itself. Organization-level connectors, Cowork cloud sessions and working across Office apps are separate decisions for the Owner, and on Enterprise some of them are off by default. Team and Enterprise also stay on the older memory experience; the migration has been announced without a date.
Anthropic doesn’t publish limits as a number of messages. It gives multipliers relative to the Pro plan, a session window that resets every five hours and a weekly limit that resets at a fixed time assigned to your account. Usage counts across all surfaces combined and shows up in your account settings. The message counts that circulate on blogs have no basis in the documentation.
Five use cases in sales
Many of the hard requirements below come down to a decision by the organization’s Owner, so read them before you talk to IT. Each use case also names its plan.
Individual and business plans
A project per key client or segment
- What you do
- You keep product materials, pricing, case studies and meeting notes for a given client in one place, plus standing instructions on how to write and talk about them. You come back to the conversation two weeks later without explaining the context all over again.
- Where you click
- The Projects section, “+ New Project”, a name and description, files in the knowledge section, standing instructions through “Set project instructions”. Existing chats can be moved into a project, and projects can be starred and archived.
- Which plan
- Every plan, with a limit of five projects on the free plan. Sharing a project with your team with view or edit access requires the Team or Enterprise plan.
- Hard requirement
- A single file can be up to 30 MB and the number of files is unlimited, but the real constraint is the context window. Each project has its own memory space, separate from the others, and Anthropic doesn’t publish an upper limit for the knowledge base in tokens.
Project content is cached, so reusing it doesn’t count against your usage limit. That makes a client dossier cheaper to keep than pasting the same materials into every chat.
Individual and business plans
Call prep from your work inbox and calendar
- What you do
- Before a meeting you ask for a brief built from past email, documents and calendar entries. The Microsoft 365 connector covers SharePoint, OneDrive, Outlook mail and calendar, plus the Teams messages and channels you’re part of. The Google Workspace connector covers searching and reading email, the calendar including finding a time that works for everyone, and documents in Drive.
- Where you click
- Connector settings in Claude, on the web and in the desktop app.
- Which plan
- The Microsoft 365 and Google Workspace connectors work on all plans. On Team and Enterprise, the Owner has to enable them for the organization first.
- Hard requirement
- Microsoft 365 needs one-time consent from a Global Administrator in Microsoft Entra for the whole tenant. Claude sees only what that salesperson can see in company systems, doesn’t search in the background and pulls data only when you ask. From Google Drive it extracts text only, and Gmail attachments come through as metadata only.
A route without a connector: the Claude for Outlook add-in writes a one-page pre-meeting brief from recent threads and attachments, reads document and spreadsheet attachments without opening them, and sorts your unread inbox into three buckets: needs you, Claude can handle, noise. The add-in is in beta, needs a paid plan and works only on the message or event that’s currently open.
Individual and business plans
Follow-ups and proposal input as drafts for approval
- What you do
- After the meeting you put together a summary and a draft message, then read and fix it before sending. The three routes differ in who clicks “Send”: the Google Workspace connector only creates drafts in your inbox, the Outlook add-in does the same and doesn’t even ask for send permission, and the Microsoft 365 connector can send the email and manage the calendar once an admin turns on its write tools.
- Where you click
- A connector in the chat, or the add-in installed from Microsoft AppSource, in Outlook and Excel on the web, Windows and Mac.
- Which plan
- Office add-ins require a paid plan. The Microsoft 365 connector works on all plans, and an admin turns on the write tools.
- Hard requirement
- Neither the Microsoft 365 connector nor the Outlook add-in can handle an email with an attachment: the connector won’t send, forward or draft emails with attachments, and the add-in doesn’t read PDF attachments inline.
Longer materials (a summary of what was agreed, a one-pager, a comparison) are built in the side panel as an artifact and need the code execution and file creation setting turned on. Claude can also read from and write to Excel, PowerPoint, Word and Outlook in one conversation, but working across files is off by default on business plans.
Individual and business plans
Your team’s standard as a procedure, plus recurring reports
- What you do
- You write down once how your company does meeting notes, builds a proposal and qualifies an opportunity, and from then on it works the same way for every salesperson. On top of that you set up a recurring task that puts together a pipeline review or the day’s briefing every morning.
- Where you click
- Skills are added under Customize, recurring tasks in Cowork through “Scheduled” and “New task”, where you choose the approval mode and frequency.
- Which plan
- Skills are available on all plans, including Free. Cowork and recurring tasks require a paid plan, and rolling a skill out to the whole organization requires the Team or Enterprise plan and the Owner role.
- Hard requirement
- Skills need code execution turned on, and the package is a ZIP file with the skill folder and a required Markdown instructions file (name up to 64 characters, description up to 200). Recurring tasks run remotely, so they fire even when your computer is asleep, but they can use only cloud tools and files stored in your account.
Cowork uses more of your limit than a regular chat because it works in multiple steps. Since August 6, 2026, Enterprise has a beta that automatically scans third-party skills and plugins for malicious content.
Individual and business plans
Working from your phone and finishing at your desk
- What you do
- After a client meeting you dictate your notes and what was agreed, and at your desk you turn that into a follow-up and a CRM entry. Cowork sessions run in the cloud and sync across devices, so a task you start on your laptop can be finished on your phone.
- Where you click
- On the phone, the sound-wave icon next to the microphone; on the web and in the desktop app, the wave symbol in the chat window. Two modes: hands-free conversation, and push-to-talk for noisy places.
- Which plan
- Voice mode is in beta on all plans. Cowork on web and mobile is in beta for Max and the business plans, and is rolling out to Pro.
- Hard requirement
- Voice conversations count toward your plan’s regular usage limits, and Anthropic doesn’t give a limit in minutes. The transcript stays in the chat history, so you can work with it at your computer.
A custom assistant for the team
In Claude, “your own assistant” means three different things, and mixing them up is the most common reason people are disappointed after the first week.
Project, skill and connector: three different layers
A project holds static background knowledge that loads every time a chat in that project starts. That’s where the client dossier and your rules for writing proposals go.
A skill kicks in dynamically when a task calls for it, and it works everywhere in Claude. That’s where a procedure goes: what meeting notes look like, how you qualify an opportunity, what proposal input has to include.
A connector links Claude to an outside system and its data: your inbox, calendar, drive or CRM.
How it reaches the whole team
On Team and Enterprise plans, the person running the team has two routes. A project can be shared with specific people or the whole organization, with view or edit access. A skill is uploaded once by the organization’s Owner and then shows up automatically in every team member’s list, either on or off by default.
Free, Pro and Max have no project sharing. Each salesperson builds their own, and nobody else will see it.
Where to build it and what goes in the instructions
You put the assistant together at a computer. A phone is fine for using a finished project and for conversation, but building skills and working with local files are comfortable, or possible at all, only in the desktop app.
Rules that should apply to every conversation in a project belong in its instructions. The Excel, PowerPoint and Word add-ins have their own field for preferences applied to every conversation, separate for each app.
You build a custom assistant at a computer. None of the four tools lets you build one from a phone.
On the road: phone and voice mode
Between meetings, in the car or in a reception area, you work from your phone. The lists below show which Claude features work there.
Works on the phone
- The iOS and Android apps are listed on the pricing page for every plan, including Free.
- Voice mode in beta, on all plans, with two recording modes.
- Using ready-made projects and skills in a text chat.
- Cowork cloud sessions synced with your computer, in beta on Max and the business plans.
- On iOS, drafting a message or email with the content filled in, in Messages, Mail, WhatsApp, Slack, Messenger, Signal, Gmail and Outlook.
- Reading calendar events, showing locations on a map and adding items to existing reminder lists.
Doesn’t work on the phone
- Access to existing messages, emails and the contact list on iOS. Claude won’t read your email history with a client there.
- Creating or editing the reminder lists themselves; Claude only adds items to lists that already exist.
- Location tools for people on Team and Enterprise plans.
- Local files, browser control and computer use: these need the desktop app running.
- Live previews of materials as they’re created; on the phone and on the web you see the finished files.
Voice mode: what it covers
Voice mode is in beta and works on all plans, on the phone, in the desktop app and on the web. In a conversation it uses connected tools (the documentation names Gmail, Google Calendar, Google Docs and Slack) and web search. You can switch between voice and text in the same conversation, and the transcript stays in your history.
There are three limitations: voice conversations count toward your plan’s regular usage limits, Anthropic doesn’t give a limit in minutes, and one of the models isn’t supported in this mode.
Cowork away from your desk
Cowork sessions run in the cloud and sync across devices. Access to local files, browser control and computer use require the desktop app to be running, and live previews of materials being created work only there.
Projects and Cowork skills don’t work by voice: in a voice conversation you can’t call on them, and voice mode itself doesn’t work in Cowork or in Claude Code.
Voice mode works well for a long note after a meeting, because the transcript stays in the chat history and you pick it up again at your computer. To run a company procedure, switch back to text.
Company data and admin decisions
A salesperson can’t unlock anything in this section alone. It’s written for the person who signs the license and for IT.
Model training and retention
On Team and Enterprise plans, Anthropic states that by default it doesn’t train models on the organization’s content, and on Enterprise the admin configures the retention policy.
On Free, Pro and Max, training depends on the model improvement setting. With it on, retention extends to five years and covers new and resumed chats; with it off, retention stays at 30 days. You can change the setting at any time in your privacy settings.
- On individual plans, with the model improvement setting on, content pulled into a chat from Gmail, Google Drive and Google Calendar may be used to improve models. The commitment not to train on data from these connectors covers Team and Enterprise.
- The consumer terms don’t apply to services under commercial terms or to the API.
What the admin has to turn on
A business license gives access, and the organization’s Owner does the configuration. Go through these settings with IT first.
- Organization-level connectors: the Owner has to enable them before anyone can connect. The Owner can also limit what actions a connector can take across the organization and allow connectors with a verified domain only for accounts in your own organization.
- Microsoft 365 also needs one-time consent from a Global Administrator in Microsoft Entra for the whole tenant, and the write tools (sending email, managing the calendar, updating files) are a separate decision.
- On the Team plan, Cowork has one switch for the whole organization, with no split by team. Cloud sessions are on by default on Team; on Enterprise they’re off by default and need the Owner to enable them and grant the permission to a group. On Enterprise the most restrictive layer wins, and role-based grants can’t override the organization setting.
- Office add-ins are installed from Microsoft AppSource, and working across apps is off by default on business plans and controlled globally from the organization settings.
- A custom MCP server has to be reachable from the public internet. A server behind a firewall or on a private network won’t connect; the fix is to allowlist Anthropic’s IP address ranges.
Connecting your CRM
HubSpot has an official connector, available to HubSpot customers on every tier and requiring a paid plan on the Claude side. It’s read-only and covers standard records: contacts, companies, deals, tickets and the associations between them, without custom properties marked as sensitive. HubSpot permissions apply automatically, so a salesperson sees only the deals they have access to.
Salesforce has an official, documented integration, but only as a connection inside Claude Tag, Claude’s presence in Slack, which is in public beta for Team and Enterprise plans. Claude reads accounts, contacts, opportunities and cases; writing is optional and depends on the permissions granted. Setup requires a connected app on the Salesforce side and a dedicated integration user.
With Salesforce, the credential belongs to the agent, so an individual salesperson’s permissions don’t automatically filter what Claude shows in a channel. Visibility is set by the integration user’s permission set, and Anthropic recommends starting with read-only permissions.
Claude in Slack
Claude Tag is in public beta for Team and Enterprise plans. Only the Primary Owner or an Owner can set it up (the Admin role isn’t enough), and users don’t configure anything on their side. The Owner sets a hard spending cap for the organization and per-channel limits, gets alerts at 75 and 95 percent, and sees usage analytics broken down by channel.
Work in a channel is billed to the organization’s account, direct messages to the individual’s account. The Owner can review, edit and delete channel and workspace memory, and after the integration is disconnected, conversations are deleted within 30 days.
Audit and compliance for Office add-ins
Inputs and outputs are deleted on Anthropic’s side within 30 days, and the add-in’s chat history is stored locally in the browser and doesn’t sync across devices. The add-ins don’t inherit the organization’s custom retention settings, and their activity doesn’t reach Enterprise audit logs, the Compliance API or data exports. Enterprise can make up for that by sending telemetry to its own collector; on Pro, Max and Team, audit export isn’t available.
The documentation also warns explicitly about prompt injection in emails and files from outside sources and recommends verifying every action. The add-ins work in Outlook and Excel on the web, Windows and Mac with a Microsoft 365 subscription. They don’t work on the perpetual 2016 and 2019 versions, Outlook for iOS and Android, Excel on iPad and Android, or on-premises Exchange mailboxes.
What Claude won't do
Eleven things Claude can’t do today. Anthropic spells them out in its documentation, so it’s better to run into this list now than in week three of the rollout.
Won’t send email from Gmail or the Outlook add-in
The Google Workspace connector only creates drafts, and the Outlook add-in doesn’t even ask for send permission. Only the Microsoft 365 connector can send, after an admin turns on the write tools, and even then without attachments.
Won’t post a message in a Teams channel
The connector reads the messages and channels you’re part of, and that’s where its scope ends. It doesn’t post messages or change Teams settings.
Won’t run your assistant by voice
Voice mode doesn’t work in Cowork or in Claude Code and can’t reach projects or Cowork skills. Voice conversations also count toward your plan’s regular usage limits.
Won’t share a project with your team on an individual plan
Project sharing starts with the Team plan. On Free, Pro and Max, each salesperson builds their own, and nobody else will see it.
Won’t do everything from your phone
Local files, browser control and computer use need the desktop app, and live previews of materials work only there. Cowork on web and mobile is in beta and doesn’t cover the Pro plan, and recurring tasks can’t reach local folders.
Won’t work out of the box when you buy a business license
Connectors, Cowork, Claude Tag in Slack and the write tools need the Owner’s decision, Microsoft 365 also needs a Global Administrator’s consent, and the Office add-ins need a rollout on the Microsoft 365 side.
Won’t connect Salesforce in one click from the chat window
The integration exists only inside Claude Tag in Slack and needs setup on the Salesforce side, and an individual salesperson’s permissions don’t filter visibility the way they do in HubSpot.
Doesn’t work on all of Excel or your whole inbox
The add-ins read and write only the files that are currently open and the message or event that’s currently open. They don’t open, close or switch files, and the Outlook add-in doesn’t read PDF attachments inline.
Won’t bring work in the Office add-ins under company audit
Add-in activity doesn’t reach Enterprise audit logs, the Compliance API or data exports. Separately, on individual plans with the model improvement setting on, content pulled from Gmail, Google Drive and Google Calendar may be used to improve models, and retention then extends to five years.
Won’t give business plans the new memory experience
Team and Enterprise stay on the older version of memory, and the migration has been announced without a date. A company gets no organization-level memory control, because Team members manage it themselves, and memory doesn’t work in Cowork.
Won’t give you a message limit, and isn’t ready in the browser
Anthropic publishes multipliers and reset windows, and the specific message counts that circulate online come from blogs. The browser extension is in beta and needs a paid plan, and Anthropic says plainly that despite safeguards the feature remains risky.
Frequently asked questions
Will Claude send the follow-up after a meeting on its own?
Not in the setup you’ll see the first time you log in. The Gmail connector only creates drafts, and so does the Outlook add-in, which doesn’t even ask for send permission.
The Microsoft 365 connector can send, but only after an admin turns on the write tools, and it can’t attach files. In practice we teach working with a draft and an approval step, because that’s how it works with all four vendors.
We bought a business license. Does that mean everything works now?
No. Connectors have to be enabled at the organization level, and Microsoft 365 also needs one-time consent from a Global Administrator in Microsoft Entra.
On top of that, Cowork on the Team plan is switched on for everyone at once or for no one, cloud sessions on Enterprise are off by default, and working across Office apps is off by default precisely on business plans.
Can we connect our CRM?
HubSpot has an official read-only connector that respects each user’s HubSpot permissions, and it requires a paid plan on the Claude side.
Salesforce has an official integration, but only inside Claude Tag in Slack, in public beta for Team and Enterprise plans, with the credential belonging to the agent. Other systems connect through your own MCP server, which has to be reachable from the public internet.
Are our conversations used to train models?
On Team and Enterprise plans, Anthropic states that by default it doesn’t train models on the organization’s content.
On individual plans it depends on the model improvement setting: on means five-year retention, off means 30 days. With the setting on, this also applies to content pulled into a chat from Gmail, Google Drive and Google Calendar.
How we teach this in the workshop
We don't try to talk anyone into switching tools. Workshop exercises run in whatever the company already has: ChatGPT, Gemini, Copilot or Claude.
We start by asking what tasks your salespeople handle and which tools they use. If the team works in Claude, the workshop revolves around what we described above: a project per key client, a procedure written down so everyone produces proposals to one standard, and working with a follow-up draft that goes through a person before it’s sent.
After the workshop, the team has finished projects and skills to use from day one. The core program has 11 modules covering the whole sales process, from fundamentals and research, through building assistants, prospecting and call prep, to proposals, AI agents and account management.
Sources
We read the documentation from Anthropic and its partners linked below on August 12, 2026, and describe plans, limits and integration scope as of that day. Open the links again before you buy.
- Claude plans and pricing
- What is the Pro plan?
- What is the Max plan?
- What is the Team plan?
- What is the Enterprise plan?
- What are projects?
- How can I create and manage projects?
- What are Skills?
- How to create custom Skills
- What are artifacts and how do I use them?
- Use Claude with iOS apps
- Use Claude with Android apps
- Use voice mode
- Use connectors to extend Claude’s capabilities
- Use Google Workspace connectors
- Connect to Microsoft 365
- HubSpot connector for Claude
- Claude Tag: Salesforce connection
- What is Claude Tag?
- Get started with Claude Cowork
- Use Claude Cowork on web, desktop, and mobile
- Use Claude Cowork on Team and Enterprise plans
- Schedule recurring tasks in Claude Cowork
- Get started with Claude in Chrome
- Claude for Excel
- Claude for Outlook
- Office agents: work across apps
- Use Claude’s chat search and memory to build on previous context
- How do usage and length limits work?
- Usage limit best practices
- Upload files to Claude
- Updates to our consumer terms
- Claude release notes
All sources accessed on August 12, 2026. We update this page when features that matter for sales work change.