# Artifacts Source: https://docs.canvas.inc/artifacts Tables and documents your agent creates Artifacts are the outputs your agent produces. They appear as tabs in the right panel and persist after the conversation ends. ## Tables When a workflow finds companies, people, or contacts, results go into a table. You can search, sort, filter, and export to CSV. ## Documents When a workflow involves research or content generation, results go into a document rendered as rich markdown. ## Usage Artifacts are created automatically. You can reference them in follow-up messages, export tables as CSV, or delete ones you no longer need. # Automations Source: https://docs.canvas.inc/automations Run agent workflows on a recurring schedule Automations run a prompt on a recurring schedule without manual intervention. Each run produces a full conversation you can review in the agent's history. ## Creating an automation Ask your agent to set one up: > "Schedule a daily automation at 7am to pull my Granola meetings and draft follow-up emails." The agent synthesizes a prompt, confirms the details, and starts running it on schedule. ## Run history Every automation logs its runs with status, duration, and a summary. Click any run to see the full conversation. ## Managing automations You can pause, edit, or delete automations from the detail page. Pausing stops runs temporarily without losing your setup. # Context Source: https://docs.canvas.inc/business-context Tell your agent who you are and who you sell to Context tells your agent about your company, customers, and outreach style. It's loaded into every session automatically, so you never have to repeat yourself. It typically covers your ICP, target personas, industries, value props, and messaging tone. ## Setup Go to **Context** in the sidebar. Click **Set up with AI** to have the agent walk you through it, or click **Or write it yourself** to enter it manually. You can edit or refine your context at any time. # Canvas Data MCP Source: https://docs.canvas.inc/canvas-data Built-in tools for company search, people discovery, contact enrichment, and social intelligence Canvas Data is your built-in data toolkit. When you ask Canvas to find companies, look up contacts, or research a prospect, it calls these tools automatically — no setup required. Understanding what's available helps you write better prompts. ## Modules * **[Companies](/data/companies)** — search B2B companies by industry, size, funding, location * **[People](/data/people)** — find people at companies by title, seniority, department * **[Contacts](/data/contacts)** — get verified email addresses and phone numbers * **[Places](/data/places)** — find local businesses via Google Maps * **[LinkedIn](/data/linkedin)** — profiles, posts, reactions, and URL lookup * **[Instagram](/data/instagram)** — profiles, follower counts, and recent posts * **[AI Research](/data/ai-research)** — AI-powered scoring, categorization, and research * **[Integrations](/integrations)** — connected apps like Slack, Salesforce, and HubSpot ## How it works Just describe what you need in natural language. Canvas picks the right tool, calls it, and presents the results. > "Find 10 AI startups in Austin, find their CTOs, and get their email addresses" Canvas will chain three tools automatically: company search → people search → contact enrichment. ## Credits Each tool has a per-use credit cost. See individual module pages for pricing. Credits are tracked per message and visible on the billing page. ## Full reference For parameters, returned fields, and example prompts, see each module page linked above. # AI Research Source: https://docs.canvas.inc/data/ai-research AI-powered scoring, categorization, and content generation Run AI-powered research on any data — score leads, categorize companies, generate personalized content, or extract structured insights. **Credit cost:** 2 credits per call (standard), 20 credits per call (deep research). Don't use AI Research for factual data like emails or phone numbers — use [Contacts](/data/contacts) for that. Don't use it for current news — Canvas has web search built in. *** ## Result types Choose the output format that fits your use case: ### Text Free-form text response. Use for summaries, research briefs, personalized content. **Example prompts:** * "Summarize what this company does in one sentence" * "Write a personalized email opener for this prospect" * "Research recent news about Stripe — funding, hiring, product launches" ### Number Returns a single number. Use for scoring and ranking. **Example prompts:** * "Score this company 1-100 for ICP fit based on: B2B SaaS, 50-500 employees, has a sales team" * "Rate this lead's seniority 1-10" Be specific about scoring criteria. "Score 1-100 based on company size, industry fit, and title seniority" works much better than just "score this lead". ### Option Pick from predefined choices. Use for categorization and qualification. **Example prompts:** * "Qualify this lead as HOT, WARM, or COLD based on title seniority and company size" * "Categorize this company as COMPETITOR, PARTNER, or PROSPECT" ### Structured Output Extract multiple typed fields at once. Use for complex analysis. **Example prompts:** * "Analyze this lead for outreach — return qualified (yes/no), score (1-100), pain points, and a personalized opener" * "Extract company size, industry, and main product from this data" *** ## Common workflows **Lead scoring:** > "Find 20 SaaS companies, then score each one 1-100 for ICP fit" **Personalized outreach:** > "For each contact, write a personalized one-sentence email opener" **Categorization:** > "Categorize these companies as COMPETITOR, PARTNER, or PROSPECT" # Companies Source: https://docs.canvas.inc/data/companies Search and discover B2B companies by industry, size, funding, and location Find B2B companies using structured filters or natural language queries. Canvas has two company search tools that it picks between automatically based on your prompt. **Credit cost:** 1 credit per company returned. *** ## Structured search Best when you know specific criteria — industry, employee count, funding stage, location. ### Example prompts * "Find 20 SaaS companies in San Francisco with 50-500 employees" * "Find Series A fintech startups in the US" * "Find AI companies founded after 2020 with at least \$5M funding" ### Available filters | Filter | What it does | Example | | -------------- | ------------------------------------- | ----------------------------------- | | Keywords | Match company descriptions (OR logic) | "developer tools", "fintech" | | Industries | Strict industry category | "Software", "Healthcare", "Finance" | | Location | City, state, or country | "Austin, TX", "California" | | Countries | Country filter | "United States", "Canada" | | Employee count | Min/max headcount | 50–500 employees | | Funding stage | Stage filter | seed, series\_a, series\_b, etc. | | Funding amount | Min/max total raised | $1M–$50M | | Founded year | Min/max founding year | 2018–2023 | | Accelerators | Y Combinator, Techstars, etc. | Specific batches supported | *** ## Natural language search Best when your criteria are too nuanced for filters, or when structured search returns no results. ### Example prompts * "Find AI companies building developer tools in San Francisco" * "Find Series C healthcare SaaS companies" * "Find recently funded climate tech startups" *** ## What you get back | Field | Description | | --------------- | --------------------- | | name | Company name | | website | Company website | | industry | Industry category | | description | Company description | | employee\_count | Headcount | | location | Headquarters | | linkedin\_url | LinkedIn company page | | logo\_url | Company logo | | founded\_year | Year founded | | funding\_stage | Latest funding stage | | total\_funding | Total funding raised | *** ## Tips * **Less is more** — don't over-filter. Start broad, narrow if needed. * **Keywords are OR logic** — "fintech" or "payments" matches either. * **Natural language fallback** — if structured search returns 0 results, Canvas automatically retries with natural language. # Contacts Source: https://docs.canvas.inc/data/contacts Get verified email addresses and phone numbers Enrich people with verified email addresses and phone numbers. This is the standard next step after finding people — the people → contacts pipeline. **Credit cost:** 4 credits per email, 40 credits per phone number. *** ## How it works Canvas sends each person through a waterfall of data providers to find verified contact information. The more context you provide (name, company, LinkedIn URL), the higher the match rate. ### Example prompts * "Get the email for Jaspar Carmichael-Jack, CEO at Artisan" * "Find emails for these people" * "Get email and phone for the VP of Sales at Stripe" *** ## Email types You can ask for specific email types: | Type | When to use | | ----------------- | ----------------------------------------------- | | **any** (default) | Returns whatever's available — work or personal | | **work** | Prioritizes work/corporate email | | **personal** | Prioritizes personal email (Gmail, etc.) | ### Example prompts * "Get their personal email addresses" * "Find work emails for these contacts" *** ## What you get back | Field | Description | | --------------- | ---------------------------------------------- | | name | Full name | | email | Best email based on your preference | | personal\_email | Personal email when available | | work\_email | Work email when available | | email\_type | Classification: personal, work, other, unknown | | email\_status | Verification: valid, risky, invalid | | phone | Phone number (if requested) | | phone\_status | Phone verification status | | title | Job title | | company | Company name | | linkedin\_url | LinkedIn URL | *** ## Tips for better match rates 1. **Include LinkedIn URL** — dramatically improves accuracy. If you have people without LinkedIn URLs, ask Canvas to find them first. 2. **Use domain, not company name** — "acme.com" works better than "Acme Inc". 3. **Include job title** — helps disambiguate common names. 4. **Check email\_status** — filter to "valid" for deliverable addresses. *** ## Typical pipeline Most contact enrichment follows this pattern: > "Find 5 AI startups in SF, find their CEOs, and get their email addresses" Canvas chains: **Companies → People → Contacts** automatically. # Examples Source: https://docs.canvas.inc/data/examples End-to-end workflow examples using Canvas Data Real workflows showing how Canvas chains tools together from a single prompt. *** ## Outbound Prospecting > "Find 50 SaaS companies in the US with 50-500 employees, find their VP of Sales, get their emails, and push them to Instantly as a new campaign" **What Canvas does:** 1. **Companies** — finds 50 companies matching the criteria 2. **People** — searches for VP of Sales at each company 3. **Contacts** — enriches each person with verified email 4. **Integration (Instantly)** — creates a campaign and adds leads with personalized sequences 5. **Table** — presents the results with name, title, company, email, LinkedIn *** ## Inbound Lead Qualification > "Research this person's company, score them 1-100 for ICP fit, and if they're a hot lead, create a Salesforce lead and ping #sales in Slack" **What Canvas does:** 1. **AI Research** — researches the company (what they do, size, industry) 2. **AI Research** — scores ICP fit on your criteria 3. **Integration (Salesforce)** — creates a lead record for hot prospects 4. **Integration (Slack)** — alerts the team in #sales with the score and context 5. **Document** — presents a qualification brief *** ## LinkedIn Engagement Lead Gen > "Get reactions from this LinkedIn post, filter to directors and VPs, get their emails, and add them to my HubSpot contacts" **What Canvas does:** 1. **LinkedIn Reactions** — pulls everyone who reacted to the post 2. **Filter** — keeps only directors, VPs, and C-suite titles 3. **Contacts** — enriches with verified email addresses 4. **Integration (HubSpot)** — creates contacts with engagement source tagged 5. **Table** — presents qualified engaged prospects *** ## Account Research > "Research these 5 target accounts — what they do, recent news, and potential pain points — and save the briefs to Notion" **What Canvas does:** 1. **AI Research** (per company) — company overview, recent news, pain points 2. **Integration (Notion)** — creates a page per account with the research brief 3. **Document** — presents a summary *** ## Local Business Lead Gen > "Find 30 auto repair shops in Denver with at least 4 stars and a website, and export to Google Sheets" **What Canvas does:** 1. **Places** — searches Google Maps with rating and website filters 2. **Integration (Google Sheets)** — exports results to a spreadsheet 3. **Table** — presents results with name, address, phone, website, rating, reviews *** ## Full Pipeline with CRM Sync > "Find healthcare IT companies with 100-1000 employees, find their CRO or VP of Partnerships, get emails, score each lead, push to Salesforce, and send hot leads to Slack" **What Canvas does:** 1. **Companies** — finds healthcare IT companies in the size range 2. **People** — finds CRO / VP of Partnerships at each 3. **Contacts** — gets verified email addresses 4. **AI Research** — scores each lead 1-100 5. **Integration (Salesforce)** — creates lead records with scores 6. **Integration (Slack)** — posts hot leads (score 80+) to #sales-alerts 7. **Table** — presents the full enriched list with scores # Instagram Source: https://docs.canvas.inc/data/instagram Profile data and recent posts from public accounts Get Instagram profile data, follower metrics, and recent posts. Only works on public profiles. **Credit cost:** 5 credits per profile. *** ## Example prompts * "Get the Instagram profile for @nike" * "Pull Instagram data for these influencer accounts" * "Show me follower counts and recent posts for these brands" *** ## What you get back ### Profile | Field | Description | | -------------- | --------------------- | | username | Instagram username | | fullName | Display name | | bio | Biography text | | followersCount | Follower count | | followingCount | Following count | | postsCount | Total post count | | profilePicUrl | Profile picture URL | | isVerified | Verified badge status | ### Posts (included per profile) | Field | Description | | ------------- | ------------------------- | | caption | Post caption | | timestamp | When posted | | likesCount | Like count | | commentsCount | Comment count | | mediaType | IMAGE, VIDEO, or CAROUSEL | | permalink | Direct link to post | *** ## Use cases * **Influencer research** — compare follower counts, engagement rates, and content themes * **Brand monitoring** — track competitor Instagram presence * **Social proof** — pull engagement data to qualify influencer partnerships # LinkedIn Source: https://docs.canvas.inc/data/linkedin Profiles, posts, reactions, and URL lookup Access LinkedIn data for social selling, research, and engagement tracking. Four tools, each serving a different use case. **Credit cost:** 4 credits per lookup (all tools). *** ## Profiles Get detailed profile data from LinkedIn URLs. ### Example prompts * "Get the LinkedIn profile for linkedin.com/in/jaspar-carmichael-jack" * "Pull LinkedIn profiles for these people" ### What you get back | Field | Description | | ----------- | --------------------- | | fullName | Full name | | headline | Profile headline | | about | About/summary section | | jobTitle | Current job title | | companyName | Current company | | connections | Connection count | | followers | Follower count | | email | Email (if public) | | topSkills | Listed skills | | country | Country | | education | Education history | *** ## Posts Get recent posts from LinkedIn profiles, including engagement metrics. ### Example prompts * "Get the last 10 posts from this LinkedIn profile" * "Show me recent posts from these industry leaders" ### What you get back | Field | Description | | ----------- | --------------------- | | postText | Post content | | postLink | Direct link to post | | postedAgo | When posted | | numLikes | Like count | | numComments | Comment count | | numShares | Share count | | isRepost | Whether it's a repost | *** ## Find URLs Find LinkedIn profile URLs from a person's name and company. Useful before contact enrichment — having a LinkedIn URL significantly improves email match rates. ### Example prompts * "Find the LinkedIn URL for Jaspar Carmichael-Jack at Artisan" * "Find LinkedIn URLs for these people" ### What you get back | Field | Description | | ----------- | ---------------------- | | linkedinUrl | Found profile URL | | confidence | Match confidence (0–1) | This tool uses first name and last name separately. Canvas splits the name automatically. *** ## Reactions Get people who reacted to a LinkedIn post. Great for finding engaged prospects around competitor or industry content. ### Example prompts * "Who liked this LinkedIn post? \[post URL]" * "Get reactions from this competitor's announcement post" * "Find people who celebrated this post" ### What you get back | Field | Description | | ------------ | ---------------------------------------------------- | | fullName | Reactor's name | | profileUrl | LinkedIn profile URL | | headline | Profile headline | | reactionType | LIKE, CELEBRATE, SUPPORT, LOVE, INSIGHTFUL, or FUNNY | | jobTitle | Current job title | | companyName | Current company | *** ## Common workflow Find engaged prospects and reach out: > "Get reactions from this LinkedIn post, filter to directors and VPs, then get their email addresses" Canvas chains: **Reactions → filter by title → Contact enrichment** # People Source: https://docs.canvas.inc/data/people Find people at companies by role, seniority, and department Find people at known companies by job title, seniority level, and department. Also supports natural language people search when you don't have a specific company list. **Credit cost:** 1 credit per person returned. *** ## Search at companies Best when you already have companies and want to find specific roles at them. ### Example prompts * "Find the CEO at Artisan and Spendflo" * "Find VP of Sales at these companies" * "Find engineering leaders at these 10 startups" ### How titles work Canvas translates your intent into specific LinkedIn-style job titles: | You say | Canvas searches for | | --------------------- | ---------------------------------------------------------- | | "founders" | CEO, Founder, Co-Founder | | "sales leaders" | CRO, VP of Sales, Head of Sales, Sales Director | | "engineering leaders" | CTO, VP of Engineering, Director of Engineering | | "product leaders" | CPO, VP of Product, Head of Product | | "ML engineers" | ML Engineer, Machine Learning Engineer, Senior ML Engineer | ### Seniority levels Every people search requires a seniority level to avoid overly broad queries: | Level | Matches titles like | | -------------------------- | ---------------------------------- | | c-suite | CEO, CTO, CFO, COO, CMO | | vp / svp | VP of X, SVP of X | | director / head | Director of X, Head of X | | manager | Engineering Manager, Sales Manager | | lead | Tech Lead, Team Lead | | senior / staff / principal | Senior Engineer, Staff Engineer | *** ## Natural language search Best when you want to describe who you're looking for without having specific companies. ### Example prompts * "Find senior ML engineers at Series B fintech companies in SF" * "Find product managers who previously worked at Google" * "Find CTOs at healthcare startups with 50+ employees" *** ## List all people Returns everyone at a company without role filters. Use when you want a full org chart. ### Example prompts * "List all people at Artisan" * "Show me everyone at this company" *** ## What you get back | Field | Description | | ------------- | --------------------------------- | | name | Full name | | first\_name | First name | | last\_name | Last name | | title | Job title | | company | Company name | | linkedin\_url | LinkedIn profile URL | | email | Email (sometimes, not guaranteed) | | phone | Phone (sometimes, not guaranteed) | People search doesn't always return emails. Chain to contact enrichment for verified email addresses. # Places Source: https://docs.canvas.inc/data/places Find local businesses via Google Maps Find local businesses like restaurants, dentists, gyms, and retail stores using Google Maps data. This is for physical businesses — use [Companies](/data/companies) for B2B/tech company search. **Credit cost:** 6 credits per place returned. *** ## Example prompts * "Find 10 top-rated sushi restaurants in Austin TX" * "Find dentists near downtown Brooklyn with at least 4.5 stars" * "Find yoga studios within 2 miles of my office" * "Find coffee shops in Seattle that are open now" *** ## Available filters | Filter | What it does | Example | | ----------- | ------------------------------ | ------------------------------- | | Query | Business type to search | "sushi restaurants", "dentists" | | Location | City or area (required) | "Austin, TX", "Manhattan" | | Radius | Search radius in meters | 5000 (default) | | Min rating | Minimum Google rating | 4.0 | | Min reviews | Minimum review count | 50 | | Has website | Only businesses with a website | true (default) | *** ## What you get back | Field | Description | | ------------- | -------------------------------------- | | name | Business name | | address | Full street address | | phone | Phone number | | website | Website URL | | rating | Google rating (1–5) | | review\_count | Number of reviews | | place\_id | Google Place ID | | types | Business type categories | | price\_level | Price level (1–4) | | hours | Operating hours and open/closed status | | location | Latitude and longitude | *** ## Use cases * **Local lead generation** — find businesses that might need your services * **Competitive analysis** — map competitors in a region * **Territory mapping** — find all businesses of a type across cities # File System & Sandbox Source: https://docs.canvas.inc/file-system Your agent runs in a cloud sandbox with its own persistent file workspace Your agent doesn't run on your machine. It runs in an isolated cloud sandbox, so it can work in the background without tying up your browser or laptop. Each agent gets its own persistent file workspace. Files it creates (CSVs, reports, images, exports) are saved automatically and stick around across conversations. ## Sandbox Every agent session runs in a secure, isolated sandbox in the cloud. This means: * **Background execution** - start a workflow and close your laptop. Your agent keeps running. * **No local setup** - nothing to install, no dependencies to manage, no ports to open. * **Isolation** - each agent runs in its own environment. One agent's work never interferes with another's. ## Files Agents create files as part of their work, like exporting a lead list to CSV, generating a research report, or saving intermediate data. These files live in the agent's workspace and persist across conversations. You can: * **Preview files** directly in the app - spreadsheet view for CSVs, rendered images, and more. * **Download** any file to your machine. * **Pick up where you left off** - files from a previous conversation are still there when you come back. ## Sharing files in chat Agents can share files directly in the conversation. When an agent creates something worth reviewing, it surfaces a clickable file card in the chat. Click to preview or download. # Welcome Source: https://docs.canvas.inc/index Coding agents that run your entire GTM. Find companies, enrich contacts, sync your CRM, and more. Just describe what you need. Canvas Data MCP gives you one interface to companies, people, contacts, LinkedIn, Instagram, AI research, and integrations with Salesforce, HubSpot, Slack, and more. Just describe what you need and Canvas does the rest. Canvas agent interface showing a prospecting workflow with results table ## Get started Run your first workflow in under a minute. ## Talk to us Bugs, ideas, and feedback: [founders@canvas.inc](mailto:founders@canvas.inc) # Integrations Source: https://docs.canvas.inc/integrations Connect external apps and use them from prompts Connect apps like Slack, Salesforce, and HubSpot via OAuth. Once connected, Canvas can use them automatically from any prompt. > "Send a Slack message to #sales about this lead" ## Connecting an app Go to **Connectors** in the sidebar, find the app, click **Connect**, and complete the OAuth flow. That's it. ## Available apps | App | Categories | | --------------- | -------------------------- | | Salesforce | CRM, Sales | | HubSpot | CRM, Marketing | | Attio | CRM, Sales | | Apollo | CRM, Sales | | Instantly | Sales, Email | | HeyReach | Marketing, Sales | | Slack | Communication | | Gmail | Communication, Email | | Notion | Productivity | | Google Calendar | Productivity, Calendar | | Google Sheets | Productivity, Data | | Granola | Productivity, Meetings | | Gong | Analytics, Sales | | Calendly | Productivity, Scheduling | | Cal.com | Productivity, Scheduling | | Supabase | Developer Tools, Databases | # Playbooks Source: https://docs.canvas.inc/playbooks End-to-end prompt recipes for common GTM workflows Step-by-step workflows you can run by typing prompts into Canvas. *** ## Outbound pipeline Build a targeted prospect list from scratch. **Step 1:** Find companies > Find SaaS companies in the US with 50-500 employees that have raised a Series A or B. **Step 2:** Find decision makers > Find the VP of Sales and Head of Marketing at each company. **Step 3:** Get contact info > Get their emails and phone numbers. **Step 4:** Personalize > Write a one-sentence personalized opener for each person based on their company and role. **Step 5:** Push to CRM > Create HubSpot contacts for everyone with a score above 60. *** ## Inbound demo qualification Qualify and route someone who just booked a demo. **Step 1:** Research > [jane@acme.com](mailto:jane@acme.com) just booked a demo. Research Acme Corp: what they do, how big they are, and any recent news. **Step 2:** Enrich > Enrich Jane's contact info and find other decision makers at Acme. **Step 3:** Score > Score this lead 1-100 based on title seniority, company size, and industry fit. **Step 4:** Route > If the score is above 70, send a Slack message to #sales-alerts with the details. *** ## Event follow-up Process a list of event attendees and prioritize outreach. **Step 1:** Upload > Here's a CSV of attendees from last week's conference. **Step 2:** Enrich > Fill in missing emails, phone numbers, and titles. **Step 3:** Score and segment > Score everyone 1-100 and split into hot, warm, and nurture segments. **Step 4:** Export > Export the hot leads as CSV. *** ## LinkedIn engagement prospecting Find and enrich people who engaged with a specific LinkedIn post. **Step 1:** Get reactions > Find people who liked or commented on this LinkedIn post: \[paste URL] **Step 2:** Filter > Filter to people with Director, VP, or C-suite titles. **Step 3:** Enrich > Get their emails and phone numbers. **Step 4:** Push to CRM > Push the qualified leads to Salesforce. *** ## Account-based meeting prep Research a target account before a meeting. **Step 1:** Company research > Do a deep research on Ramp: what they do, recent news, funding, and competitive landscape. **Step 2:** Find key people > Find the leadership team at Ramp and get their LinkedIn profiles. **Step 3:** Recent activity > Get the last 5 LinkedIn posts from the person I'm meeting with: \[paste LinkedIn URL] **Step 4:** Brief > Summarize everything into a one-page meeting prep brief. *** ## Weekly CRM cleanup Clean and enrich stale CRM data on a recurring schedule. **Step 1:** Set up the automation > Schedule a weekly automation every Monday at 9am: pull all Salesforce leads created this week, enrich missing fields, score them, and flag anyone below 30. Canvas runs this automatically every week. Review results in the automation's run history. # Quickstart Source: https://docs.canvas.inc/quickstart ## Your first workflow Type a prompt like this into the Canvas agent: > Find fintech companies with 50-200 employees, get the VPs of Engineering, and pull their emails. Canvas takes it from there. Typing a prospecting prompt into the Canvas agent ### What happens Canvas searches for companies matching your criteria: industry, size, location, funding stage, and more. It identifies the right people at each company based on the titles and seniority you asked for. It enriches each person with verified emails and phone numbers using multiple data sources. Your results appear as a table on the right: names, titles, companies, emails, and phone numbers. You can search, sort, filter, or export as CSV. Table artifact showing enriched leads with names, titles, emails ### Push to your CRM If you've connected a CRM, just ask: > Push these leads to HubSpot. Canvas creates contacts in your CRM automatically. See [Integrations](/integrations) to connect your apps. ## What's next See everything Canvas can do under the hood. Explore real GTM workflows like outbound, inbound, and account research. # Skills Source: https://docs.canvas.inc/skills Reusable playbooks that customize how your agent behaves Skills are reusable playbooks that tell your agent what to do in specific scenarios. Each skill has a name, a trigger description, and a set of instructions the agent follows when activated. ## How they work All enabled skills are loaded into every agent session. The agent picks the right one based on what you ask. For example, a skill called "My Outbound Pipeline" with the trigger "when I say 'run outbound'" activates whenever you type "run outbound." ## Creating a skill From the **Skills** page in the sidebar, click **New Skill** and fill in the name, description, and instructions. You can also ask the agent directly: > "Create a skill called 'Weekly Report' that compiles my pipeline metrics every time I say 'weekly report'." ## Examples * **Outbound pipeline**: "When I say 'run outbound', find companies matching my ICP, identify decision-makers, and enrich their contact info." * **Meeting prep**: "When I say 'prep for meeting with \[company]', research the company, find recent news, and identify key people." * **Lead scoring**: "When I ask to score leads, evaluate each lead against my ICP criteria and rank them by fit." # Account Research Source: https://docs.canvas.inc/use/account-research Deep dives on target accounts, competitive intel, and meeting prep Canvas can research companies in depth: recent news, funding, product launches, tech stack, competitive landscape, and key people. Use it for meeting prep, account planning, or competitive intel. ## How it works Ask Canvas to research any company: > Research Stripe: recent funding, product launches, and who the key decision makers are in sales. Canvas will search the web, synthesize findings, and return a structured brief. For deeper research, it uses multi-step AI research that checks multiple sources. ## Example prompts > Write a one-page brief on Ramp before my meeting tomorrow. Include recent news, company size, and key people. > What tech stack does Figma use? Do they use Kubernetes? > Research these 10 target accounts. For each one, summarize what they do, their funding stage, and potential pain points. > Compare these three competitors: what do they each do differently? ## What Canvas uses under the hood * [AI Research](/data/ai-research): AI-powered research with `sonar-pro` or `sonar-deep-research` * [Companies](/data/companies): company data lookup * [LinkedIn](/data/linkedin): profile data and recent post activity # CRM Hygiene Source: https://docs.canvas.inc/use/crm-hygiene Clean, enrich, and dedupe your contact data Canvas can take a messy CRM export or lead list and clean it up: filling in missing fields, deduplicating records, and enriching stale data. ## How it works Upload a CSV or paste a list, then tell Canvas what to fix: > Enrich this list: fill in missing emails, phone numbers, and titles. Canvas reads your data, identifies gaps, enriches each contact, and returns a cleaned-up table. ## Example prompts > Here's a CRM export. Fill in missing emails and phone numbers for everyone. > Dedupe this list by email and keep the most complete record for each person. > Score these 200 contacts 1-100 based on title seniority and company size. Flag anyone below 30. > Update job titles for everyone in this list. Some are outdated. ## What Canvas uses under the hood * [Contacts](/data/contacts): fill in missing emails and phones * [AI Research](/data/ai-research): scoring, categorization, and data validation * [LinkedIn](/data/linkedin): find LinkedIn URLs for contacts * [Integrations](/integrations): push cleaned data back to your CRM # Inbound Qualification Source: https://docs.canvas.inc/use/inbound-qualification Enrich, score, and route inbound leads automatically When a lead comes in, whether from a form fill, demo booking, or event, Canvas can enrich them, score them against your ICP, and tell you whether they're worth pursuing. ## How it works Give Canvas a name, email, or company and ask it to qualify: > Someone from Acme Corp just booked a demo. Research them, find the right decision makers, and score them against my ICP. Canvas will: 1. **Research the company**: what they do, size, industry, funding, recent news 2. **Enrich the contact**: fill in title, phone, LinkedIn, and any missing fields 3. **Score and qualify**: rate the lead against your ICP criteria (set up in [Business Context](/business-context)) ## Example prompts > This person just signed up: [jane@acme.com](mailto:jane@acme.com). Research their company and tell me if they're a fit. > Enrich these 20 demo signups from this week and score them 1-100 based on company size and title seniority. > Someone from a Series B fintech just booked a call. Find the other decision makers at their company and get contact info. ## What Canvas uses under the hood * [Contacts](/data/contacts): email and phone enrichment * [AI Research](/data/ai-research): company research and lead scoring * [People](/data/people): find additional decision makers * [Integrations](/integrations): CRM and Slack integration # Outbound Prospecting Source: https://docs.canvas.inc/use/outbound-prospecting Build targeted prospect lists with verified contact info Build a list of target companies, find the right people, and get their verified contact info, all from a single prompt. ## How it works Describe who you're looking for and Canvas handles the rest: > Find SaaS companies with 50-500 employees, get their VPs of Engineering, and pull emails and phone numbers. Canvas runs through three steps automatically: 1. **Finds companies** matching your criteria (industry, size, location, funding, tech stack) 2. **Finds people** at those companies by title, seniority, or department 3. **Enriches contacts** with verified emails and phone numbers Results appear as a table you can search, sort, filter, and export as CSV. ## Example prompts > Find Series A fintech startups in the US, get the Head of Sales at each, and pull their emails. > Build a list of companies using Snowflake with 100+ employees. Find the data engineering leads and get contact info. > Find venture-backed AI companies founded after 2020, get C-suite contacts, and push them to Instantly to start a drip campaign. ## What Canvas uses under the hood * [Companies](/data/companies): company discovery by firmographic criteria * [Places](/data/places): local business discovery via Google Maps * [People](/data/people): find people by title, seniority, department * [Contacts](/data/contacts): waterfall email and phone enrichment # Pipeline Automation Source: https://docs.canvas.inc/use/pipeline-automation Schedule recurring workflows and trigger actions automatically Canvas can run workflows on a schedule, from daily prospecting to weekly enrichment to recurring research, without you being there. ## How it works Ask Canvas to set up an automation: > Schedule a daily automation at 8am to find 10 new AI startups, get their CTOs' emails, and push them to HubSpot. Canvas creates a recurring job that runs on your schedule. Each run produces a full conversation you can review afterward. ## Example prompts > Every Monday, enrich any new leads added to Salesforce this week and score them. > Schedule a daily automation at 7am to pull my Granola meetings and draft follow-up emails. > Run outbound prospecting every weekday: find 20 companies matching my ICP and enrich the decision makers. See [Automations](/automations) for setup details. ## What Canvas uses under the hood Any SDK method can be used in an automation. Common patterns: * [Companies](/data/companies) + [People](/data/people) + [Contacts](/data/contacts): recurring prospecting * [AI Research](/data/ai-research): scheduled research and scoring * [Integrations](/integrations): CRM syncs and Slack alerts