Contents
Every page links to book a Workflow consultation — reach out whenever a question turns into a project.
The Basics
What is Claude?
Claude is an AI assistant made by Anthropic. Think of it like a very capable colleague you can talk to in plain English — you give it a task, it figures out how to do it.
Unlike a search engine (which finds information) or a basic chatbot (which just chats), Claude can actually do things: write documents, analyze data, build things in your tools, answer complex questions, and help you think through decisions.
How do I get started in Claude?
The most common ways to access Claude are:
claude.ai
— the web and desktop app, which is what you're likely using right now.
Claude API
— for developers building Claude into their own products.
Claude Code
— a terminal tool for coding tasks.
Plugins / connectors
— like the monday.com integration, which lets Claude take actions directly inside your tools.
Unsure how to get started with Claude for your organization? Reach out to Workiflow today.
How do I prompt in Claude?
Give context, not just a command
“Write me an email”
“Write a follow-up email to a prospect who attended our webinar but hasn't booked a demo. Keep it short, friendly, and end with a soft CTA.”
Tell Claude what role to play
“Act as a project manager reviewing this plan for gaps” gets better output than just “review this plan.”
Specify the format
Bullet points? A table? A draft you can copy-paste? A numbered list of options? Say so — Claude will match it.
Include constraints
“Under 150 words,” “no jargon,” “written for someone who's never used monday.com” — constraints sharpen the output dramatically.
Iterate, don't restart
If the first response isn't right, don't start over. Say what to adjust: “Make it more direct” or “Cut the second paragraph and add a stat.” Claude holds the full conversation in memory.
Share examples when you have them
“Here's an email we sent that performed well — write something in this style” is one of the most powerful prompts you can give.
The short version: Claude is only as specific as you are. The people who get the most out of it treat it like a briefing, not a search bar.
Should I keep the same chat for every conversation?
No — and this trips a lot of people up. Each chat has a context window, meaning there's a limit to how much Claude can hold in mind at once. Long chats eventually get unwieldy and Claude can lose track of earlier details.
New topic
new chat
Keep conversations focused.
Ongoing project
one chat
If you're iterating on the same thing — a doc, a plan, a campaign — staying in one thread keeps the full history in view.
Repetitive tasks
a reusable prompt
Build one you paste fresh each time rather than scrolling back through old chats.
Can Claude remember me across conversations?
By default, no. Each new chat starts completely fresh — Claude has no memory of previous conversations, your name, your preferences, or anything you've discussed before. It's like meeting a smart colleague who gets amnesia between every meeting.
However, some setups change this. In Cowork, Claude can be given persistent context through saved instructions, connected tools, and skills — so it knows who you are, what company you work for, and how you like to work without you re-explaining every time.
How do I make Claude remember past conversations and context?
Custom instructions / system prompts
tell Claude upfront who you are, your role, your preferences, and your company context.
Paste context at the start of a new chat
keep a short “about me” blurb you drop in at the top of any new conversation. Ten seconds, dramatically better outputs.
Use skills
skills store your voice, preferences, and context so Claude applies them automatically without re-briefing.
Reference documents
upload a brief, brand guide, or strategy doc at the start of a session so Claude has your full context to work from.
The more upfront context Claude has, the less you repeat yourself. Need help setting Claude up? Reach out to Workiflow.
Claude Models
Four models, one decision: how much horsepower does the task actually need? Speed and cost trade off against depth of reasoning.
Fastest, lowest cost. Best for high-volume, repetitive, or simple tasks where speed matters more than depth — quick summaries, form responses, lightweight automation. Smaller context window (200K tokens vs. 1M on the others).
Best balance of speed and intelligence — the right default for most work: drafting, analysis, research, content, Q&A. Fast enough to feel snappy, smart enough to handle nuance. On introductory pricing through August 2026.
Built for complex, multi-step work — agentic coding, enterprise workflows, sustained reasoning across a lot of context. Slower and pricier than Sonnet, but significantly more capable. Anthropic's recommended default if you're unsure.
Anthropic's most capable widely available model. Built for long-running agents, the hardest reasoning tasks, and highest-stakes outputs. Slowest and most expensive — use it when Opus isn't enough.
A simple way to think about it
| If you need… | Use… |
|---|---|
| Speed + volume | Haiku |
| Everyday tasks | Sonnet |
| Complex reasoning | Opus |
| Maximum capability | Fable 5 |
For most users in a tool like Cowork, Sonnet is where you'll live 90% of the time.
Sources · Models overview, Claude Platform Docs · Claude Fable 5 and Mythos 5, Anthropic.
Not sure which model or plan fits your team?
Pricing, Tokens & Usage
Not sure which plan or package is right for your team? Reach out to Workiflow — we help teams get set up right from day one.
What is a token?
A token is the unit Claude uses to measure everything it reads and writes. One token is roughly 4 characters of text — about 0.75 words. A short email is a few hundred tokens; a 10-page document is closer to 5,000–7,000.
Every time you use Claude it counts tokens in two directions: input (your prompt, any files, conversation history) and output (what Claude writes back). Both count toward your usage.
Why it matters: your plan determines how many tokens you can use before hitting a limit. Larger files, longer conversations, and bigger outputs consume tokens faster.
What are Claude's plans?
Individual plans
| Plan | Price | Best for |
|---|---|---|
| Free | $0 | Occasional use |
| Pro | $20/mo or $200/yr | Regular use |
| Max 5× | $100/mo | Frequent users across varied tasks |
| Max 20× | $200/mo | Daily power users |
Team & org plans
| Plan | Price | Best for |
|---|---|---|
| Team (Standard) | $25/seat/mo or $20 annually | Teams of 5–150 |
| Team (Premium) | $125/seat/mo or $100 annually | Power users within a team |
| Enterprise | Custom (usage-based) | Orgs 20+ needing security, compliance & scale |
Want help picking the right plan for your team?
What's the difference between Pro and Team?
Pro is for individuals. Team adds centralized billing, admin controls, SSO, spend caps, tool connectors (Google Drive, Slack, Microsoft 365, GitHub), and per-member usage limits rather than shared. Team Standard gives 1.25× more usage per session than Pro; Team Premium gives 6.25× more.
What's Enterprise, and when does it make sense?
Enterprise is for organizations that need security, compliance, and unlimited usage at scale. The seat fee covers access only — all usage is billed separately at standard API rates, so there's no cap to worry about.
It includes audit logs, SCIM, custom data retention, customer-managed encryption keys, and HIPAA-readiness for eligible orgs. Self-serve starts at 20 seats; sales-assisted starts at 50.
What happens when I hit my usage limit?
On paid plans, you hit a limit, not a wall. Team plans can purchase usage credits so members keep working after their included limit is reached. Enterprise plans on usage-based billing have no limits — you pay for what you use.
| If you are… | Use… |
|---|---|
| Testing Claude out | Free |
| A solo user doing regular work | Pro |
| A heavy individual user | Max 5× or 20× |
| A team with shared tooling needs | Team |
| An org needing compliance + scale | Enterprise |
Sources · Choose a Claude plan · What is the Team plan? · What is the Enterprise plan? — support.claude.com
Claude × monday.com
The monday.com MCP connector turns Claude from a conversation into an operator inside your account. Here's what teams ask most.
MCP stands for Model Context Protocol — the bridge that lets Claude connect to external tools like monday.com and take actions inside them, not just talk about them. Without MCP, Claude is a conversation. With it, Claude becomes an operator inside your tech stack.
More than most people expect. Claude can read boards, create and update items, change column values, build automations, create boards from scratch, query data across the account, write docs, and spin up entire workspaces from a plain-language description.
Claude executes what you ask, so specificity matters. It won't act on its own — it responds to your prompts. Best practice is to test in a sandbox board before running Claude on production data, especially for bulk updates.
It depends on how the connection is configured. By default, Claude has the access level of the authenticated user — if you're an admin, Claude can see what you can see. Scope can be restricted at the org level.
No. The monday.com MCP connector is installable through Cowork without writing any code. Configuration is handled at the admin level.
Yes — unless you pair it with scheduled tasks or build it into an automation workflow. On its own, Claude is prompt-driven, not always-on.
Monday's native AI is scoped to specific in-product features — summaries, formula suggestions, item generation. Claude via MCP has broader reasoning, works across the full account, and takes multi-step actions from complex instructions. They complement each other rather than compete.
It does what you said, not what you meant — the biggest learning curve. Clear, specific prompts reduce this significantly. For high-stakes actions, ask Claude to describe what it's about to do before doing it. It will.
A monday.com account (any tier), the MCP connector installed in Cowork, and admin permissions to authorize the connection. That's it.
What the integration can do
Read & query
Pull data from any board, group, or item; search across boards at once; summarize status, workload, or pipeline health; answer questions like “what's overdue in my marketing board?”
Create & build
Create boards, groups, items, and columns; build entire workspaces from a plain-language description; generate pre-populated boards, docs, and dashboards.
Update & manage
Change statuses, values, assignees, and dates; move items between groups or boards; bulk-update in one prompt; add updates and comments.
Automate
Build and configure automations; set up triggers and actions without the UI; chain multi-step workflows from a single instruction.
Analyze & report
Cross-reference data across boards; identify patterns, blockers, and bottlenecks; generate standup, status, and leadership summaries; surface pipeline insights.
CRM-specific
Create and update contacts, leads, and deals; log activity and move deals through stages; draft follow-up emails or updates based on CRM data.
The real differentiator
Multi-step workflows
Claude chains actions together. One prompt — “find all overdue items assigned to the design team, move them to the At Risk group, and draft a Slack message summarizing the blockers” — executes as a single workflow, not three separate tasks.
Want Claude connected to your monday.com account?
Security & Governance
No. By default, Anthropic does not use data submitted through the API or claude.ai paid plans to train its models. On Enterprise plans this is backed by a Data Processing Agreement. On a free plan, review Anthropic's current terms — data handling differs. Pressure-test: confirm your plan tier before making compliance claims internally.
On Team and Enterprise plans, admins manage this through Organization Settings — user roles, access permissions, per-user spend limits, and which connectors and tools are available. Free and Pro plans are individual, with no org-level controls.
Yes. On Team and Enterprise plans, admins control which MCP connectors are enabled for the organization — a connector that isn't installed can't be used. Network access settings can also be managed by organization Owners under Capabilities.
On Enterprise plans, yes. The Compliance API gives programmatic access to activity logs, chat histories, and file content, filterable by user and time range; the Analytics API provides aggregated engagement metrics. On Team plans, admins have billing visibility and usage summaries but not full chat-level audit logs.
Enterprise plans support HIPAA-ready configurations for eligible organizations, including a Business Associate Agreement (BAA) through Organization Settings. SOC 2 Type II and other compliance documentation is available via Anthropic's Trust Portal (trust.anthropic.com). GDPR data-processing terms are in the commercial terms. For US data-residency requirements, Enterprise plans offer US-only inference to keep processing within the United States.
On Enterprise plans, system prompts can be configured at the org level to set Claude's behavior, tone, scope, and restrictions for all users — enforcing consistent outputs, restricting topics, and ensuring brand or compliance alignment across the team.
On Enterprise plans, admins set spend limits at both the org and individual-user level, with usage billed at API rates and tracked in real time. On Team plans, usage credits can be purchased and capped. Free and Pro plans operate within fixed session limits — no additional billing exposure.
Is our data used to train Claude?
No. By default, Anthropic does not use data submitted through the API or claude.ai paid plans to train its models. On Enterprise plans this is backed by a Data Processing Agreement. On a free plan, review Anthropic's current terms — data handling differs.
Pressure-test: confirm your plan tier before making compliance claims internally.
Governance checklist — before you roll out org-wide
- / 01Assign an internal Claude owner — one admin who controls authentication, permissions, and policy.
- / 02Define approved use cases and document them for your team.
- / 03Set up spend limits at the user level.
- / 04Configure system prompts for consistent behavior where needed.
- / 05Enable audit logging on Enterprise, or establish a review process on Team.
- / 06Brief all users on data handling — what not to paste in (passwords, PII, confidential client data).
- / 07Schedule a 30-day review to assess usage patterns and adjust access.
What should employees never put into Claude?
Set a clear internal policy before rollout. At minimum, employees should not input passwords or credentials, personally identifiable information (PII) beyond what your DPA covers, confidential client data without explicit authorization, or information subject to regulatory restrictions outside your compliance configuration.
Skills, Hooks & Artifacts
Three building blocks turn Claude from a chat window into a governed, repeatable part of your stack. Here's what each one is.
What is a Skill?
A skill is a pre-built module that gives Claude specialized knowledge, instructions, and context for a specific domain or task. Think of it as a briefing document Claude reads before it starts work.
Instead of re-explaining your brand voice, your role, your audience, or your preferred output format every time you open a new chat, a skill stores all of that and applies it automatically. The kiki-ruff-marketing skill, for example, tells Claude who Kiki is, how she writes, who she's writing for, and what to never do — without her typing a word of it.
Skills can be built for almost anything — a company's brand voice, a workflow like building monday.com demos, a document format like a Word doc or PPTX, or an industry use case like immigration law. Once installed, Claude uses the skill every time that task comes up.
Skills are installed through Cowork or Claude Code. Your org can use pre-built skills from Anthropic, build custom ones, or work with a partner like Workiflow to create skills tailored to your workflows.
What is a Hook?
A hook is a piece of logic that runs automatically at a specific point in Claude's workflow, before or after Claude takes an action.
Where skills shape what Claude knows, hooks shape what Claude does at the edges of its actions. A pre-tool hook runs before Claude uses a tool — logging every action before it executes, or requiring approval before a write operation fires. A post-tool hook runs after — formatting the output, sending a notification, or triggering a downstream step.
Hooks are primarily a developer and admin-level feature used in Claude Code and API deployments. For most business users, they operate invisibly in the background as part of a governed deployment. For technical teams, they're how you enforce policy, add safety checks, and integrate Claude cleanly into existing systems.
Practical example
A hook fires before Claude writes to a monday.com board, checks whether the target board is on the approved list, and blocks the action if it isn't.
What is an Artifact?
An artifact is a persistent, self-contained output Claude creates that lives beyond the conversation.
A standard Claude response disappears when you close the chat. An artifact is different — it's a file or interactive page that saves, can be re-opened, and in many cases pulls fresh data each time it loads. Artifacts can be HTML pages, documents, code files, interactive dashboards, spreadsheets, or visual reports.
The practical value: instead of asking Claude to summarize your monday.com pipeline every Monday morning, you build an artifact once. It becomes a live page that re-fetches the data each time you open it — the answer is always current, and you've stopped doing the manual work entirely.
Artifacts are created through Cowork and can connect to any tool Claude has access to — monday.com, Slack, Google Drive, QuickBooks, and more. They're the difference between a one-time answer and something you keep using.
Claude Terminology
A plain-English glossary of the terms you'll run into working with Claude.
Agent / Agentic AI
Claude operating autonomously across multiple steps to complete a goal, rather than answering a single question. An agent plans, executes, adjusts, and finishes without a prompt at every step. This is the basis of tools like Cowork.
Artifact
A persistent output Claude creates — a document, dashboard, interactive page, or file — that saves and can be reopened or refreshed. See full definition in Section 06.
Claude Code
A terminal-based tool that lets developers delegate coding tasks to Claude from the command line. Separate from the chat interface; built for technical workflows.
Connector
A specific integration that links Claude to an external tool — monday.com, Slack, Google Drive, QuickBooks. Connectors make Claude an operator inside your stack, not just a standalone assistant.
Context Window
How much information Claude can hold in working memory at once, measured in tokens. Most current models support a 1M-token window — roughly 750,000 words. Beyond it, Claude loses access to the earliest parts of the conversation.
Hallucination
When Claude produces information that sounds confident and plausible but is factually incorrect. Most common with specific data points, recent events, niche topics, and precise numbers. Always verify high-stakes outputs against primary sources.
Hook
Logic that runs automatically before or after Claude takes an action — used to enforce policy, log behavior, require approvals, or trigger downstream steps. See full definition in Section 06.
Inference
The process of Claude generating a response. When you send a prompt, Claude runs inference — it processes the input and produces output. Inference consumes tokens and drives usage-based billing on Enterprise plans.
Knowledge Base
A set of documents or data sources attached to a Project that Claude references when responding. Instead of re-uploading the same files, a knowledge base makes them permanently available in that Project's context.
MCP (Model Context Protocol)
The open standard that allows Claude to connect to external tools and take actions inside them. MCP is what makes the monday.com integration possible — the bridge between Claude's intelligence and your tech stack.
Plugin
An installable bundle that adds new capabilities to Claude — typically MCP connectors, skills, and tools packaged together. The monday.com Demo Builder is an example of a plugin.
Project
A saved workspace in Claude with persistent context, files, and instructions. Unlike a one-off chat, a Project remembers your setup and applies it every time — useful for a client account, a recurring campaign, or a product area.
Prompt
The instruction or question you give Claude. Prompt quality is the single biggest driver of output quality. See the prompting section in Section 01.
Skill
A pre-built module that gives Claude specialized knowledge and instructions for a specific domain or task. See full definition in Section 06.
System Prompt
Background instructions configured before a conversation starts that shape how Claude behaves throughout — tone, scope, restrictions, role. On Enterprise plans, system prompts can be set org-wide for consistent behavior.
Token
The unit Claude uses to measure text — both input and output. Roughly 4 characters or 0.75 words per token. Every prompt and every response consumes tokens, which determines usage and cost.
Tool Use
Claude's ability to call external functions, APIs, or connectors during a conversation to retrieve data or take actions. Looking something up in monday.com, reading a file, or running a search is all tool use.
Agent / Agentic AI
Claude operating autonomously across multiple steps to complete a goal, rather than answering a single question. An agent plans, executes, adjusts, and finishes without a prompt at every step. This is the basis of tools like Cowork.
/ Start
Ready to deploy Claude the right way?
Most organizations don't fail at AI adoption because the technology doesn't work. They fail because there's no structure behind it — no governance, no workflow design, no one accountable for making it stick.
Workiflow's certified architects have spent half a decade and 30,000+ hours building on monday.com for over 1,000 clients. We're a monday.com Platinum Partner, CRM Specialist, and Claude Certified Architects — we connect these tools, govern them properly, and embed them into workflows that actually get used.