What does our pipeline look like? Update the Torchlight opportunity based on notes from Gong. What if a CRM agent could take care of all of this for you? Let’s build that CRM agent.

Let’s go ahead and create an agent in Gumloop and let’s give it a name, something appropriate like CRM agent, and then let’s simply connect Salesforce. Now, you might use HubSpot, Airtable, Pipedrive, Attio, Affinity. I’m not here to tell you what to use. Connect whatever your team uses.

Now, we can already ask this agent questions. “What does our pipeline look like?” This agent is going to explore your CRM and come back with an answer. It can also update information. “Upgrade jeff@example.com from a contact to a lead,” or “Move Torchlight to closed lost.” Sad, never a message you want to send, but at least now an agent will update it for you. Just with that, we’ve built a CRM agent.

Let’s keep going though. Now, the model underneath our agent, Claude from Anthropic, GPT from OpenAI, understands broadly how your CRM is structured. They’ve been trained on a kajillion Salesforce queries and HubSpot code from ages ago. But your instance is different. What that means is that although the agent will eventually get to an answer, it’s wasting a lot of time figuring out what your setup is, and wasting tokens in the process looking at your whole CRM setup every time.

To speed things up, we can ask the agent to create a skill. This is a set of instructions that it can read to understand how you structure your CRM. Something like, “Explore our Salesforce and write yourself a skill.” It’ll look at all of the fields and tables, and write itself a little guidebook so it can understand how you structure your Salesforce, or Attio, or Airtable. You get the point. Now, “What does our pipeline look like?” goes from an arduous process to something quick, fast, and easy.

Now, let’s keep building. Agents can use more than one tool. Don’t be shy, add everything your team uses. Gmail, Apollo for enrichment, Firecrawl for scraping websites, and Gong for meeting notes.

Now, “What does our pipeline look like?” can become, “Email me a pipeline report highlighting deals that need next steps.” The agent looks at Salesforce, gets all the right data, and then uses Gmail to send us that report.

Love the report? We can get it every Friday at 9 a.m. with something like, “Send me this report at 9 a.m. every Friday.”

We’ve built something great here. Remember when we used to click into our CRM three minutes ago? No more of that. Let’s give the same superpower to the whole team by adding the agent to Slack. Simply select where you want this agent to live, and now anyone on the team can ask and update information just like you are, in one place.

CRM agents let you connect the dots across systems, keep a cleaner CRM by making it easy to update, and what you do with this CRM agent from here is fully up to you.