Built This Week/AI by Industry

AI by Industry

AI in Sales

The sales AI that sticks does the work reps skip: filling the CRM, answering leads instantly and personalizing outbound at scale. Coaching alone, one founder told us, is a vitamin, not a painkiller.

By Jordan Metzner & Sam Nadler · From our episodes · Updated October 2026

Watch: Episode 45, Your Pipeline Is Lying to You. This AI Fixes It After Every Sales Call from 10:00

On Built This Week, AI in sales works best on the chores reps skip. Anis Bennaceur of Attention built a company on auto-filling the CRM after real-time coaching failed to stick. Anthony Monteiro of Auto Acquire AI argues that car dealers lose sellers because nobody answers them fast enough. Our own sales builds do the same kind of work: scoring leads, researching prospects and drafting outreach so a person only reviews and sends.

Coaching is a vitamin. People wanna buy painkillers, and the biggest painkiller was actually filling the CRM.

— Anis Bennaceur, Episode 45

How companies we talked to use AI in sales

Attention: filling every CRM field after every call

Attention started as a real-time coach during calls, but within a few weeks users didn't care about it. Anis flipped the product to auto-fill Salesforce and HubSpot from calls, then emails, chat and Slack. It fills every field, not just a summary; his example is an SDR hearing that a prospect is locked into a competitor's contract for six months. Today Attention also suggests next best actions, such as drafting a message to an unreached decision maker on an at-risk deal. (Episode 45, from 08:29)

Auto Acquire AI: instant offers instead of callbacks

Anthony's view is that a dealer's website should price a car on the spot: "It's math. It's algorithmic." Auto Acquire gives dealers instant offers, a phone-guided inspection checked by computer vision, and outreach to existing customers. A dealer who needs five F-150s can message the owners in its database with a price attached. Sam lived the problem that week: his dealer promised a price and still hadn't replied 24 hours later. (Episode 25, from 07:39)

How we run outbound

In Episode 7, Sam explained our move from "spray and pray" outbound to Clay. Clay enriches leads, uses OpenAI to personalize each message with a calculation tied to the prospect's open roles, and pushes the result into Instantly. He said it was early, but getting a little more traction than the old method.

What we built

  • Monte Carlo pipeline forecast dashboard: Sam built on Attention's open-source coaching stack and ran 10,000 trials on fake data. The AI forecast came in at $1.17M against a $1.44M CRM commit. Anis said Attention had never tried Monte Carlo forecasting and called the result close to what most companies see.
  • Dealer lead intent-to-action workflow: a web-form webhook goes to AI intent scoring. High-confidence leads get an automated SMS ("Would you like to schedule a test drive?") and low-confidence leads go to an email drip. Anthony suggested flipping it to buy cars from owners.
  • n8n sales outreach workflow: researches Russell 2000 companies with Perplexity, finds the CEO's email, has ChatGPT write a structured custom email, logs to Google Sheets and drafts it in Gmail, on a daily schedule.
  • Claude-generated n8n outreach workflow: a prompt in Claude produced an importable n8n workflow in seconds, though the last 10% of connecting it is still work.

What's still hard

  • Coaching doesn't stick. Anis found buyers wouldn't pay for real-time coaching after the zero-interest-rate era ended. Forecasting has the same problem: leaders want actions that improve the number, not just the number.
  • AI-native CRMs. Anis said every buyer he met who tried one hated it and switched back to Salesforce or HubSpot, so Attention won't build a CRM.
  • Agents vs workflows. Anis said deterministic workflows often beat agents. Jordan agreed; he hasn't found an agent SDK use case that beat step-by-step LLM calls.
  • Personalization at scale. Sam wants outreach to look like real research on each prospect, but said that is hard to do at volume.
  • AI-sounding copy. Anis flagged that the coaching feedback in Sam's demo read as AI-written and pointed to a "humanizer" he released.

Episodes on AI in sales

FAQ

How is AI used in sales?

The guests we hosted use it to auto-fill CRM fields from calls and emails (Attention), score inbound leads and text instant offers (Auto Acquire AI), and suggest next best actions on at-risk deals. We use it for personalized outbound with Clay and an n8n workflow.

Is AI sales coaching worth it?

Attention CEO Anis Bennaceur started with real-time call coaching and found users stopped caring about it within weeks. He pivoted to auto-filling the CRM, which he calls the painkiller, and kept coaching as a secondary feature.

Should sales teams use AI agents or workflows?

Anis said deterministic workflows often work as well as or better than agents and avoid heavy inference cost. Jordan agreed: when he builds, he usually ends up with step-by-step LLM calls rather than the agent SDK.

How do you personalize cold outreach with AI?

Sam uses Clay to enrich leads and add a calculation tied to each prospect's open roles, then pushes the messages to Instantly. Jordan's n8n workflow researches Russell 2000 companies with Perplexity and drafts a custom email to each CEO.