--- title: "Claude for RevOps: Webinar with InTandem | Clustr" description: "Watch Clustr and InTandem's Claude and Claude Code 101 for RevOps: three live builds, a first Claude project, and the full transcript." url: https://getclustr.ai/webinars/claude-for-revops-webinar source: https://getclustr.ai/webinars/claude-for-revops-webinar --- # Claude and Claude Code for RevOps A live, hands-on hour Clustr ran with InTandem for revenue operations professionals. Lauren Goodell and Hugh Chin explain how AI is reshaping RevOps, when to use Claude chat, Projects and Claude Code, build three RevOps workflows live, and walk everyone through their first Claude project. Published August 12, 2026 · Speakers: [Lauren Goodell](https://www.linkedin.com/in/lauren-goodell/), [Hugh Chin](https://www.linkedin.com/in/hugh-chin-2020/) (Clustr) · Host: Michael Ingram (InTandem) ## How can RevOps teams use Claude? RevOps teams use Claude to clean CRM imports, brief sellers before meetings, review pipeline health, score calls, draft proposals and produce recurring reports. In this webinar with InTandem, Clustr built three live: a CRM janitor that cleans a messy CSV, a pre-meeting brief agent in n8n powered by Claude, and a weekly newsletter run as a Claude scheduled task. ### Key takeaways - Claude sits on top of your CRM, data and workflow tools. It is the brain, not the database. - Context is what separates chat from a Project: instructions, memory, company files and connected data. - Get good at Projects before jumping to Claude Code. - Write your RevOps rules into project instructions; that is where your expertise turns into reliable output. - Turn on only the connectors a task needs. Fewer tools means better decisions and fewer hallucinations. - Keep a person in the loop: every demo routes output to someone who reviews it before it is used. ## What did the three RevOps demos look like? Each build shows a different way to put Claude to work: a Project you run on demand, an agent inside an n8n workflow, and a Claude scheduled task. Every one keeps a person in the loop before anything is used. 1. TriggerMessy lead CSVConference leads, business cards and article sign-ups in one export. 2. InputProject instructionsYour RevOps rules: what good and bad data look like. 3. AI stepClaude cleans and mapsDuplicates, gaps, formats and fields mapped to the CRM's import format. 4. Human reviewReview the summaryA Markdown report of every rule applied and every change. 5. OutputClean CRM importA HubSpot-ready CSV in this demo. 1. TriggerEvery day at 7 a.m.Reads Google Calendar for the day's meetings and attendees. 2. InputHave we met before?Loops through attendees to separate new contacts. 3. AI stepClaude research agentWeb search, Apollo contact data, and the prospect's property details. 4. AI stepSecond model verifiesChecks Claude's output for hallucinations. 5. OutputHTML brief in the inboxBackground, headquarters and building ownership for a CRE broker. 1. TriggerWeekly scheduleA Claude scheduled task with Slack and Gmail connected. 2. AI stepClaude researchesWhat changed in AI that week, for the audience in its instructions. 3. Human reviewDraft lands in SlackClaude tags the team; nothing goes out unreviewed. 4. AI stepBrand formattingClaude applies typography, colors and logo, with sources tagged. 5. OutputHTML newsletterSent to subscribers after a final read. ## What does the webinar cover, minute by minute? Thirteen chapters across 49 minutes, from how RevOps has changed to building your first Claude project. Each timestamp opens the video at that point. 1. [0:00](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=0)Welcome from InTandem 2. [1:40](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=100)What the hour covers 3. [4:08](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=248)How AI has changed RevOps 4. [6:22](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=382)What Claude is, and chat vs Projects vs Claude Code 5. [11:48](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=708)Where AI shows up across the revenue process 6. [15:05](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=905)Demo 1: the CRM janitor 7. [20:11](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1211)Demo 2: a pre-meeting brief agent built in n8n with Claude 8. [25:07](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1507)Demo 3: a weekly newsletter with Claude scheduled tasks 9. [29:27](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1767)Q&A: n8n vs Zapier, and how many connectors to enable 10. [31:46](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1906)Build along: your first Claude project 11. [39:50](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=2390)Project instructions vs global instructions, and a privacy setting 12. [43:05](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=2585)Adding context, skills, connectors and choosing a model 13. [47:27](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=2847)Wrap-up ## Claude for RevOps, answered directly Drawn from the questions attendees asked live ### How can RevOps teams use Claude? ### What is the difference between Claude chat, Claude Projects and Claude Code? Chat is a one-off conversation, like hiring an outside expert with no context on your company. A Project adds instructions, memory and files, like a tenured VP who knows how you work. Claude Code can build and ship systems, like an ops team with all the keys. Clustr recommends getting good at Projects before moving to Claude Code. ### Can Claude connect directly to a CRM like HubSpot or Salesforce? Yes. Through MCP connectors, Claude can plug directly into HubSpot and Salesforce, both of which have native connections. Check the permissions when you connect, because different settings allow different functionality, and turn on only the connectors a task needs so Claude chooses tools more reliably. ### Should RevOps teams use n8n or Zapier with Claude? Both work. Zapier and Make are largely interchangeable for quick, simple automations. n8n offers more capability for agent workflows, has many AI models built in, and connects well with Claude, which can generate the JSON workflow file you import into n8n. Many of the same automations can also run natively as Claude scheduled tasks. ### What is the difference between Claude project instructions and global instructions? Global instructions, set in Claude's settings, apply to every conversation and project, for example never using em dashes and asking clarifying questions one at a time. Project instructions apply only inside one project and describe that project's specific job, rules and what good output looks like. Set global instructions first, then add project instructions. ### Which Claude model should RevOps teams use? It depends on the task. Sonnet handles most day-to-day work. Opus is the flagship reasoning model and a good default for data analysis and heavy spreadsheet work. Fable is the newest and most capable but uses more of your limits. Haiku is fast and inexpensive for high-volume tasks. Higher effort settings take longer and use limits faster. ## Full transcript Lightly edited from the recording for readability: caption errors corrected and filler words removed, meaning unchanged. ### [0:00](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=0)Welcome from InTandem **Michael Ingram:** Glad you all made the time today and are able to join us. I'm Michael Ingram, CEO of InTandem. I'll keep this brief; we want to hand this over to the Clustr crew as soon as possible. As most of you know, InTandem's business is built on great RevOps experts like many of you, connecting operators with companies that need that expertise. Many of you are also on the front lines of how RevOps is changing, and a big piece of that is how you use AI to supercharge the work you do every day. **Michael Ingram:** Today is really about sharing what other best-in-class operators are doing: real use cases and live demos. We want you digging in today, hopefully building a project, and asking questions in the chat. The Clustr team will answer as they can during the presentation. At a high level, they help train, enable and implement AI-forward solutions, with a heavy focus on go-to-market teams. Hugh and Lauren, over to you. ### [1:40](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=100)What the hour covers **Lauren Goodell:** Thank you, Michael, and thank you everyone for joining. Welcome to AI 101 for RevOps. This is not a deep dive. We're not going to get super technical, and there's no engineering background required. It's a practical introduction to Claude and Claude Code. Over the next hour you'll do four things: see how AI is actually reshaping RevOps, beyond how it's embedded in the tools you use; learn the difference between Claude and Claude Code and when to use each; see three real RevOps use cases demoed live; and leave having built your own first Claude project. **Lauren Goodell:** AI adoption is exactly what we do. Many of you took the SINA assessment beforehand. We start with a baseline literacy assessment of your organization, to see where everyone is and how well they know AI, then apply it to the tools you already have to build a training program that makes your whole organization AI fluent. Later you'll hear from my partner Hugh Chin, our head of AI delivery, who will walk you through the hands-on workshop. **Lauren Goodell:** Quick poll: how often do you use AI for work today? Never, a few times a month, weekly or daily? Every day, hourly; we love to see it. SINA is free for all of you. It's a baseline assessment of where you are, and it takes 20 to 30 minutes. It looks at five components: understanding AI, evaluating outputs, using AI responsibly, directing AI, and identifying the right use cases. ### [4:08](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=248)How AI has changed RevOps **Lauren Goodell:** When was the term RevOps coined? It was 2016, but as a function it has been around far longer. In 1990 Excel arrived, the first time we really organized data in a usable way. In 2005 Salesforce and CRM let us capture that organized data. In 2016 revenue intelligence meant we were starting to understand the data, which is why the term RevOps appeared. Now AI brings all of that together so we can act on the data. **Lauren Goodell:** That is where your role is shifting. RevOps used to mean reports, dashboards, processes and documentation. Today you're a strategic adviser, an architect for automation, designing AI workflows and building business systems. When Excel came out, people thought it would replace accountants. It made them more powerful, and that's what AI is for RevOps. Boston Consulting Group says about 70% of AI transformation success comes from people and process, which is exactly what you do. You're no longer doing all the work yourself; you're building your own team. Every agent you build isn't just a tool. It becomes a team member you have to manage and hone over time. ### [6:22](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=382)What Claude is, and chat vs Projects vs Claude Code **Lauren Goodell:** Who here is on Claude, Copilot or ChatGPT? I'm guilty of using all of them, but we'll focus on Claude today. First, definitions. Artificial intelligence is simply machines doing things that used to need human smarts; think of an old spam filter's if-this-then-that rules. Machine learning goes further and recognizes patterns in large amounts of data instead of following fixed rules. Generative AI, what we'll spend most of today on, generates new things: text, images, code and plans. That covers Copilot, ChatGPT, Gemini and Claude. **Lauren Goodell:** Claude is very good at reading, writing, thinking, analyzing, explaining and connecting ideas. It is not your CRM, not a database, not a dashboard and not your workflow engine. Claude sits on top of those things, brings the data together and becomes the brain. **Lauren Goodell:** Think of it like staffing a business. Chat is like hiring McKinsey: experts in the field who don't have context on your company, so the answer can be generic. A Claude Project is like hiring a tenured VP who knows what works, what doesn't and the language you use, and learns from how you've worked together. Claude Code is like your entire ops team with all the keys and passwords: it takes the context and the expertise and helps build a system and ship it. **Lauren Goodell:** Say you want a repeatable workflow that builds territory plans from your CRM and pipeline data. You could do it with chat, a project or code. A project makes it repeatable with historical data, and connecting Claude Code makes it automated. Underneath, these interfaces use the same models. Fable is the top tier, the most expensive and the one that thinks hardest. Opus is the flagship reasoning and judgment model. Sonnet is probably what you use day to day. Haiku is fast and cheap for volume at scale. **Lauren Goodell:** People often say they use chat every day and are ready for Claude Code. Our recommendation is to spend more time going from chat to Projects first. The difference is context. Chat is a conversation. A project adds history: company documents, CRM data, call recordings, playbooks and eventually your entire business. That context is what turns a chat into a project. ### [11:48](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=708)Where AI shows up across the revenue process **Lauren Goodell:** AI already touches every step of your sales process: prospecting and lead gen with tools like Clay, Apollo, ZoomInfo and LinkedIn Sales Navigator; outreach with Outreach, Salesloft and Lavender; qualification with predictive scoring and intent data; discovery and meetings with call recorders like Gong, Granola and Fireflies; proposals, negotiation, contracting and renewal tools. Your buyers are also using AI to look up what you do, so it matters to understand how it works. **Lauren Goodell:** Beyond the AI embedded in your tools, here are ten agents you can build for RevOps, which we cover in our one-day intensive: a CRM janitor that audits CRM data; a forecast coach that reviews pipeline health and surfaces risk before forecast calls; a call coach that scores calls; an enablement manager that builds training assets; an executive chief of staff for repetitive reporting; a dashboard analyst; a proposal and email assistant; an account researcher; a process auditor; and more. **Lauren Goodell:** Let's see it live. Hugh will walk through three builds from start to finish: a CRM janitor that takes a messy CRM export and flags duplicates, gaps and fixes; an agentic workflow in n8n that puts a pre-meeting brief in your inbox every day; and Claude scheduled tasks producing a weekly newsletter. ### [15:05](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=905)Demo 1: the CRM janitor **Hugh Chin:** First, a pretty messy CSV file. Your teams are always getting leads from conferences, news, articles and business cards, and it all gets dropped into a spreadsheet. What do you see that would cause errors if we uploaded this straight into a CRM? Missing data, gaps, duplicate emails, inconsistent zip codes, redundant entries. Exactly. We want Claude to apply the same extract, transform and load mapping rules you already use, so the file loads cleanly into the CRM with standardization intact. **Hugh Chin:** We set up a simple Claude project called CRM Janitor. The instructions are where your mastery comes in: you know what good RevOps looks like and what bad looks like, so put every rule you look for when cleaning these files into the instructions. Then I dropped in the messy CSV with a prompt that doesn't just say 'clean this up'. It spells out what good looks like and what the final report should contain. **Hugh Chin:** The first output is a transformation summary in Markdown: every rule it applied, every change and the output files. The second is a much cleaner CSV import, mapped to HubSpot in this case. Whatever CRM you use, HubSpot, Salesforce or GoHighLevel, map to its import format, and you can point Claude at that format. **Lauren Goodell:** A question from the chat: could you connect Claude directly to the CRM for this? **Hugh Chin:** Absolutely. MCP is essentially a connector that plugs Claude directly into a system. HubSpot and Salesforce both have native MCP connections, and their capabilities keep improving. Look at the permissions when you connect them, because different settings allow different functionality. ### [20:11](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1211)Demo 2: a pre-meeting brief agent built in n8n with Claude **Hugh Chin:** Next is a pre-meeting brief. Salespeople are constantly meeting new people without time to research who they're meeting and what matters about the meeting: did we go to the same school, where did they work before, what does their role do, and how do our services align with their organization? This build is in n8n, which also has a native connection with Claude, so you can describe what you want in natural language instead of learning to build everything by hand. Much of this can also be done in Claude itself, but n8n shows a nice visual workflow of each stage. **Hugh Chin:** Our client in this example was in commercial real estate in the DC area. For them it mattered to know not only who they were meeting, but the tenants and the location of the prospect's headquarters, and whether a lease was expiring soon. Every morning at 7 a.m. the workflow runs. It reads Google Calendar for the day's events and attendees, then loops to check whether we've met each person before. **Hugh Chin:** Then it hands off to an AI agent powered by Anthropic's Claude model. The agent has tools: it can search the web and read pages, it's connected to a data source, Apollo in this case, for email, contact information and other data, and it looks up specific information about the prospect's DC property. A second AI agent then verifies Claude's output, using another model to check for hallucinations. Finally it formats everything into an HTML email to the salesperson. **Hugh Chin:** Here's the output the salesperson receives before meeting Eric: where Eric has worked, his education and history, and because this is commercial real estate, where his headquarters is and who owns that building. That's what matters to a commercial real estate broker. Again, you can do much of this natively in Claude with its own automations and routines. **Hugh Chin:** Ashley asked whether we can share the JSON file. Yes. JSON is another way to work with n8n: you can have Claude generate a JSON file and import it directly into n8n to build the workflow. ### [25:07](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1507)Demo 3: a weekly newsletter with Claude scheduled tasks **Lauren Goodell:** Someone asked whether you'd do this in Cowork or Claude to run it on a schedule. **Hugh Chin:** You can run it on a schedule natively in Claude, and that's the next example. Just make sure all the connectors the workflow needs are enabled in the scheduled task, the same way the n8n build had connections to Apollo and other APIs. **Hugh Chin:** When Lauren joined the company, one of her ideas was keeping everyone up to date on AI: new model releases, Microsoft Copilot changes, token optimization. Salespeople often want a newsletter they can share with clients. So we created a scheduled task that runs every week, our Clustr Pulse Weekly. Claude goes out and does the research, but it needs the right instructions: who we sell to, what information we want, and how to report back to our team. It's connected to Slack and Gmail. **Hugh Chin:** We never send anything out without proofreading it ourselves. The draft lands in Slack, where Claude tags me to say the newsletter is ready. Lauren and I review and edit it, then bring it back into Claude, which knows our brand typography, colors and logo, and formats it as an HTML email for subscribers, with sources tagged. So if you're subscribed to our newsletter, I did not handwrite all of it. ### [29:27](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1767)Q&A: n8n vs Zapier, and how many connectors to enable **Lauren Goodell:** Jackson asked: do you have a preference between n8n and Zapier for projects like the meeting brief? **Hugh Chin:** I've used Make.com in the past. Make and Zapier are largely interchangeable for quick, easy integrations and automations, and Zapier has more market share. n8n has more capabilities, was earlier to agent features, and has many models built in: Anthropic's and OpenAI's, but also Gemini, DeepSeek and others. I also like its connection to Claude, because Claude can create the JSON file you import into n8n. **Lauren Goodell:** Elaine asked: is Claude looking at other applications connected to the CRM, or do you need to connect those separately? **Hugh Chin:** On my setup I have about 18 apps connected to Claude: our CRM, Google Drive, Gmail and our whole tech stack. If you're doing the same task repeatedly, toggle on only the connectors that task needs. Every time you give Claude a task it has to decide which tools to use, so with fewer tools it makes better judgments and hallucinates less. ### [31:46](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=1906)Build along: your first Claude project **Hugh Chin:** You can reach Claude through the web app at claude.ai or the desktop app on Mac or Windows. On the left, look for Projects. Projects keep you organized, like a file structure: for different clients I bundle each client's information into its own project. Chat is for one-offs and draws on the memory across your whole Claude account, while a project stays specific to one client, project or deliverable. I'll create a new project called My RevOps Teammates, to help my team with RevOps tasks. **Hugh Chin:** On the right are instructions, memory and context. Think of the project as a new hire: a new hire rarely gets great instructions or the access they need, so you have to provide them. We'll add instructions that show what good and bad look like. **Lauren Goodell:** For those following along, the instructions start with who the project is acting as: a RevOps AI teammate, and what the job is. Next come the core responsibilities, the things you do day to day; keep adding the weekly and monthly tasks over time so it keeps learning. Then how it should think, and how it should use project knowledge. **Hugh Chin:** Project instructions are one of the most important parts of any setup, because Claude reads them first and follows them for every request. Good examples: remove em dashes, ask us more questions, and flag risks, anomalies and anything it's assuming. Every organization has its own rules for its data, so add yours. **Hugh Chin:** Memory is the project's memory. If the new hire works a project with Nike and then moves to Walmart, it carries forward what it learned. It grows with each conversation, and you can edit out one-offs you don't want carried forward. Context is where you drop relevant files: onboarding documents, PDFs, proposals, brand guidelines and examples of your voice, so Claude understands your organization and role. You can upload from your device or connect other applications. ### [39:50](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=2390)Project instructions vs global instructions, and a privacy setting **Lauren Goodell:** A good question: what's the difference between project instructions and Claude's global instructions, and which comes first? **Hugh Chin:** Project instructions apply only to that project. Global instructions, in your settings, apply across everything. Mine tell Claude never to use em dashes, and when it's uncertain, to ask me questions one at a time and tell me how many it has. Better input gives better output, so I want Claude to interrogate me and learn as much as possible about the task. **Lauren Goodell:** Typically you set Claude's global instructions first, and they apply to every project. You don't need to repeat them in each project, but we do recommend project instructions for the specific skill or job that project does. **Hugh Chin:** Another tip: in Claude's settings under privacy, you can toggle off the option to help improve Anthropic's models. It's one more level of data privacy. ### [43:05](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=2585)Adding context, skills, connectors and choosing a model **Hugh Chin:** Back in the project, take the chance to show Claude what good looks like. We've all worked on presentations through many iterations until the master sales deck is final, or proposals that got approved. Drop those into context so Claude formats new work the same way. The plus sign in the chat adds files and photos, and you can also add earlier chats to the project. **Hugh Chin:** Skills are capabilities Claude has, like creating a document, a PowerPoint or an Excel file. In our longer day intensives we build skills to make tasks repeatable: if you're always aligning output to your voice and tone, create a skill for it. There's even a skill for creating skills. **Hugh Chin:** Connectors are what MCP means. Model Context Protocol was introduced by Anthropic, and it's a fancy way of saying connector: it lets Claude connect to platforms like Slack, HubSpot and Salesforce. For this RevOps teammate I'd toggle on Gmail, Calendar and Drive, and leave off the tools it doesn't need. **Hugh Chin:** Last, the model. Which model you use depends on the task. Fable 5 is the newest, but it uses more tokens and exhausts your usage faster. For data analysis or heavy Excel work we typically suggest Opus, though Sonnet often works too. You can also change the effort level, which changes how long it works; max effort on Opus takes longer and uses your limits faster. **Hugh Chin:** That wraps up setting up your first Claude project. The best way to learn is to find someone familiar with it, ask questions, and iterate: add more context, try different prompts and different models, and compare the outputs. ### [47:27](https://www.youtube.com/watch?v=VKaGKXzCZ9c&t=2847)Wrap-up **Lauren Goodell:** Today we covered how AI is reshaping RevOps, Claude versus Claude Code and the best ways to use each, three real RevOps use cases, and building your first Claude project. We'll follow up with the prompts and the JSON file. If you want to go deeper and build agents for your own work, we run a one-day RevOps intensive, and we bring this to organizations too, with a focus on go-to-market teams: sales reps, marketing and more. Connect with us on LinkedIn. AI is only going to make you more valuable. ## Want this for your RevOps or GTM team? We run hands-on Claude training for revenue, sales and go-to-market teams, built on your own data and workflows. We use cookies We use cookies and Google Analytics to understand how visitors use our site. [Privacy Policy](/privacy)