The AI hype cycle in CRM
Every CRM on the market is now "AI-powered." Visit any vendor's website and you'll see headlines about AI. Some claim they use AI to transform your sales process. Others promise that AI will write your emails, predict your deals, and essentially sell for you while you sit back and watch.
Most of it is hype.
The AI capabilities that actually exist in CRM systems today are relatively narrow. They're good at specific, well-defined tasks. They're terrible at anything that requires human judgment, industry expertise, or understanding the context of a real sales conversation. Yet CRM vendors market their AI features as if the technology has reached some level of general intelligence that it hasn't.
The result is predictable: sales teams adopt a CRM because of the "AI features," try them once or twice, find that they don't work the way they expected, and never use them again.
Gabriel takes a different approach. We've identified AI tasks that actually solve real problems in sales. We've built features that are genuinely useful, not just impressive. And we're honest about what the AI can and can't do. We don't promise that AI will close your deals. We promise that AI will save you time on repetitive tasks and help you make smarter decisions.
Practical AI feature #1: Email drafting that sounds like you
One of the most repetitive parts of sales is writing emails. Not the important emails — the carefully crafted, personalized messages to key prospects. Those you need to write yourself. But the follow-ups, the check-ins, the re-engagements, the status updates? Those are all variations on a theme.
Gabriel's email drafting feature does one thing: it helps you draft emails that sound like you, not like a robot or a marketing department.
Here's how it works: You tell Gabriel what you want to communicate. "Follow up on the proposal I sent last week. Ask if they have questions and see if they want a call." Gabriel drafts an email. But it doesn't write something that sounds like corporate speak. It writes something short, conversational, and in the voice of someone actually reaching out to another human.
You can then edit it, customize it, or send it as is. The key insight is that the AI handles the blank page problem. You're not starting from zero. You have a draft that you can edit in 30 seconds instead of writing from scratch.
The AI isn't smart enough to do your job. It is smart enough to save you time on the part of your job that's mostly just busywork.
Practical AI feature #2: Summaries of your notes
After a call or a meeting, your notes are usually a mess. Half sentences, a phone number, a reminder to send pricing. Gabriel can turn those notes into a clean summary with action items, so the next person who opens the record knows what happened and what comes next.
You still take the notes. The AI just tidies them up.
Practical AI feature #3: Build with AI
Setting up a cadence or an automation from a blank screen takes time. With Build with AI, you describe what you want ("a two-week cadence for new inbound leads: call, email, LinkedIn, call") and Gabriel drafts it for you to review. Nothing turns on until you've looked it over and saved it.
Build with AI works from your description. It doesn't send details about your individual leads, contacts or deals to the AI provider.
Practical AI feature #4: Less typing everywhere else
A few smaller jobs where AI saves real time:
Business-card scan. Snap a card at an event and Gabriel fills in the record.
Landing pages. In Pages, describe the offer and Gabriel drafts a full landing page in your brand colors and fonts. Each section has its own AI button to rewrite its copy, regenerate it or add a new section below.
Brand import. Upload a PDF with your branding and Gabriel picks up your colors.
Your key, your bill
Gabriel's AI runs on your own OpenAI or Anthropic key. You pay the provider for what you use, and Gabriel doesn't sell AI credits or mark them up. Admins can see usage in one place.
What ties it together
All of Gabriel's AI features share a common theme: they handle busywork. They're not replacing your judgment. They're not selling for you. They're not making decisions about your pipeline.
The rest of the work stays where it belongs. Lead scoring in Gabriel is a 0 to 100 score plus an ICP fit verdict against the Ideal Customer Profile you define, so you can see why a lead ranks where it does. Cadence steps are worked by a person: Gabriel puts the right email or script in front of your rep, and your rep decides when to send.
That's what practical AI in sales looks like. It's not flashy. But it saves your team time.
AI that doesn't make sense (and we don't do)
Here's what we don't do, even though other CRMs market these as AI features:
"AI-generated" personalization. Some CRMs claim their AI can personalize outreach to hundreds of prospects at once. What they mean is: the AI will fill in a person's name, company, and maybe industry into a template. That's not personalization, and it's not AI. That's mail merge. We don't do it because generic emails with a name fill-in aren't more effective than genuinely personalized outreach, and it trains your team to scale the wrong thing.
Automatic deal creation. Some CRMs can scan your emails and automatically create deals when they think something is an opportunity. This is a terrible idea. False positives are constant. Your pipeline gets flooded with deals that don't exist. Your forecasting becomes meaningless. We don't do this because garbage in means garbage out.
"Sentiment analysis" of emails. Some CRMs claim to read incoming emails and determine the sentiment — whether the prospect sounds happy, unhappy, interested, or disinterested. This rarely works. Language is complex. A prospect might write "thanks for your follow-up" and the AI reads it as positive interest when they're actually about to say no. We don't do this because decisions based on unreliable AI is worse than decisions based on your human judgment.
Automatic pricing recommendations. Some CRMs use AI to recommend pricing for deals. This is something you need to control directly. Your pricing is strategic. AI-generated pricing recommendations are useless unless they're based on information the AI doesn't have access to.
The bottom line
AI in CRM should do one of three things: save you time on busywork, help you see patterns you should notice, or keep your process consistent. If an AI feature doesn't accomplish one of those things, it's just hype.
Gabriel's AI features are built to save you time on busywork. They're not revolutionary. They're not going to replace your sales team or sell deals for you. But they will help you sell more efficiently, more consistently, and with fewer things falling through the cracks.
That's not hype. That's how AI can actually help you sell.