Topline · 5 Apr 2026 · From the week of 30 March
What World-Class AI in GTM Looks Like | Kyle Norton, CRO @ Owner.com
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Kyle Norton, CRO of Owner.com, joins hosts Sam Jacobs, AJ Bruno and Asad Zaman to describe how his revenue team uses AI. The conversation covers data foundations for sales AI, why decentralized rep-built AI fails compared with centrally built intelligence, the build-versus-buy split, a five-P framework for choosing AI projects, pipeline as the first AI priority, and the budget and org choices needed to move. The central argument is that AI pays off when the core intelligence is built centrally, deployed into existing tools, and aimed first at pipeline, rather than handed to reps as subscriptions.
For founders
- Kyle Norton says to build the core intelligence layer centrally and buy the workflow and user-experience tools around it, rather than recreating sales engagement UX.
- Kyle Norton says to start AI work with first-party and third-party data foundations, beginning with a map of the total market and a narrower serviceable obtainable market of accounts you can win today.
- Kyle Norton says to choose AI projects by mapping possibilities, estimating payoff and probability to get expected value, and weighing the effort required, rather than by what looks interesting.
- Kyle Norton says that in the early innings roughly 80% of AI effort should go to pipeline generation.
- Kyle Norton says that if budget is tight, leaders should stack-rank the organization and move out lower performers to fund a top applied AI hire, starting with one person.
For revenue leaders
- Kyle Norton reports that Owner.com's original BDR call-to-decision-maker connect rate was three to four percent; BDRs now talk to about 20 decision makers a day and book them at a 14 to 16 percent rate, changing BDR economics.
- Kyle Norton says his team's AI implementations were built centrally and deployed into the tools reps already use, such as Salesforce and Salesloft, rather than as separate apps or custom GPTs.
- Kyle Norton says an AI model that predicts lead connect likelihood shows high-connect leads picking up at about 2.3 times the rate of a normal lead.
- Kyle Norton describes an AI workflow that does pre-call research and gives BDRs just two or three points for each cold call, so reps don't research before dialing.
- AJ Bruno says AI-native companies are changing commission plans every quarter, with usage, credit and outcome-based pricing changing most.
What was said 24, most useful first
Estimate a private company's order volume from its review count, dividing new reviews by an assumed review rate. Listen
Kyle Norton gives an example for finding order volume at a business whose data is not readily available, such as a restaurant on third-party delivery apps. He counts new reviews month over month and divides by an assumed share of customers who leave reviews, which he gives as 10%. He says each business's system for this has to be built by the team itself.
“So I'll take that review number divided by 0.1.”
Listen to the episode Pipeline & demand generation Link to this Report a problem
Owner.com's BDR economics changed when decision-maker connect rate rose from three to four percent to 20 contacts a day booked at 14 to 16 percent. Listen
Kyle Norton says the original BDR motion had a call-to-decision-maker connect rate of about three to four percent, so 100 calls reached only about four decision makers. He says that after AI tooling, a BDR talks to about 20 decision makers a day and books at 14 to 16 percent. He says closed ARR divided by BDR comp is now well over 10x.
“Our original call to decision maker connect rate was like three to four percent.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
Decentralized AI rollouts, where every employee gets AI accounts and builds their own tools, did not produce business results at Owner.com; centrally built systems did. Listen
Kyle Norton says the popular advice is to give everyone Claude accounts and have them build, and he pushed AI adoption heavily about 12 months ago. In practice, though, the implementations that worked were highly centralized. A small group (the VP of BizOps and data, the VP of RevOps and Kyle) drove the work, and the VP of BizOps and data did most of the building. The tools were then deployed into the tools reps already used, such as Salesforce and Salesloft, rather than as new apps.
“what we really found in practice is that all implementations were all highly centralized.”
In the early innings of AI, roughly 80% of AI effort should go to pipeline generation. Listen
Kyle Norton says most CROs and founders, outside the fastest-growing companies, say their problem is pipeline, and that AI is a good pipeline tool. He gives examples of picking the right accounts, timing outreach, and generating compelling artifacts. His team also built an AI website grader as a lead magnet.
“80% of your AI efforts in your early innings should be pipeline focused.”
Listen to the episode Pipeline & demand generation Link to this Report a problem
Fund a world-class applied AI hire by stack-ranking the team and reallocating headcount, rather than waiting for budget. Listen
Kyle Norton says leaders should stack-rank their organization and identify people whose departure would not hurt, then use those roles to fund an applied AI hire. He says Owner.com has three or four people in applied AI and many open roles, and that one person is enough to start. A host from bootstrapped STA adds that the first such investment is the most important and makes the next ones obvious.
“Go and do the stack rank of your organization and figure out who are the people that if they walked out the door today, you'd be like, oh, okay, we'll be fine.”
Listen to the episode Hiring & team building Link to this Report a problem
Founder mode is valuable in moderation but tends to be over-extended when applied to the extreme. Listen
A participant says founder mode is still important, with the right energy and proactiveness, but that anything in the extreme can be detrimental. He says the framework often causes over-extension when implemented, citing Ben Horowitz on Andreessen portfolio companies that stopped hiring the people they needed. He says founder CEOs are not better CEOs just because they are founders.
“I still think founder mode is like really important. Anything in the extreme can be detrimental.”
Listen to the episode Leadership & culture Link to this Report a problem
Use AI to scan every sales call and fill in CRM fields, then back-test prompts on past calls to check new answers and pricing. Listen
Kyle Norton says his team's AI scans every call and fills in all the fields. The team writes prompts and back-tests and backfills them on previous calls to see what happened when reps gave particular answers. He says they use prompts to tell whether a newly rolled-out pricing change is working.
“It scans every call and fills in all the fields and we can, you know, like write prompts and then back tests and backfill from previous calls”
Define the serviceable obtainable market as the accounts you can win today, not the aspirational TAM, and enrich each one. Listen
Kyle Norton says the third-party foundation starts with a map of the entire market, then the SOM within it, meaning the companies you think you can get on a call with and win today. He says each SOM account should be enriched with the information needed to decide who to contact and how to prioritize. Some of this is off-the-shelf data such as ZoomInfo or Datalane, and some is custom scrapers and enrichment flows built in-house.
“Not your like aspirational fundraising TAM, but the actual customers you want to work with today”
Listen to the episode Strategy & market Link to this Report a problem
Ask your best reps what signals tell them a demo will close, then turn those signals into account-targeting criteria. Listen
Kyle Norton says he gives this prompt to teams building AI infrastructure: ask the best reps when they know a demo will be a close. He gives an example where a rep watches for three specific website or LinkedIn signals, or for a particular open role. Those signals are then turned into digital-footprint criteria that AI can detect across the market.
“go sit with your very best reps and ask them, when you sit down for a demo, when do you know you're like, I'm gonna close this one?”
Listen to the episode Pipeline & demand generation Link to this Report a problem
A predictive lead score for likelihood to pick up shows high-connect leads answer at about 2.3 times a normal lead's rate. Listen
Kyle Norton says Owner.com's head of applied AI built a score that predicts which leads are likely to pick up. He says calling high-connect leads yields a pickup rate about 2.3 times that of a normal lead. He notes that a vendor (transcribed as "Tightnecks") sells this as a product.
“if you call high connect leads, they pick up at like a 2.3x rate, like a normal lead, which is insane.”
Own the core intelligence internally and buy workflow and user-experience tools, according to Kyle Norton. Listen
Kyle Norton says the core intelligence of the organization needs to be built and managed centrally, and he doesn't think it should be bought. He says the things to buy are workflows and user experiences. He says build-versus-buy was a big topic of conversation at the Clay CRO summit he attended.
“I think you have to own the intelligence. You have to build the intelligence internally and buy things that are like workflows and user experience.”
Kyle Norton has not yet seen sales engagement vendors build viable AI intelligence, so he injects internal intelligence outputs into those platforms rather than rebuilding them. Listen
Kyle Norton says he has not seen sales engagement providers build an AI product that does real intelligence work such as next-best-action or deal recommendations, and asks whether the hosts have seen otherwise. He says to output internally built signals such as a deal health score or estimated win rate onto the lead record, using the engagement platforms as the place to display them. He says he would not vibe-code the daily rep tool because it needs rock-solid stability and a support ecosystem.
“I haven't seen any of the sales engagement providers build an AI product that does any of the intelligence stuff.”
AI does the pre-call research so a BDR gets only two or three points for each cold call instead of about 20. Listen
Kyle Norton says that to speed up the funnel, his team uses AI to do the research and fill in the information so the rep does not need to research before dialing. The BDR gets exactly what they need, which he describes as two or three things to say rather than 20. He says that from there the team can test different pattern interrupts and value offers.
“make sure that the rep doesn't have to do any research before picking up the phone and calling.”
Listen to the episode Sales process & deals Link to this Report a problem
Owner.com runs experiments on call timing and meeting placement to find when connect rates are highest. Listen
Kyle Norton says his AI lead ran tests to find which hours of the day the team should be calling. The tests control for dial volume to find the best time windows. He says meetings are placed in low-connect parts of the day so the team is as active as possible during high-connect periods.
“figure out exactly what times of day should we be calling”
Listen to the episode Sales process & deals Link to this Report a problem
Choose AI projects with a five-P framework: map the possibilities, estimate payoff and probability for expected value, then weigh effort. Listen
Kyle Norton says the framework starts from the most important business problems and then lists the ways to solve each one. Each option is scored on payoff times probability to get expected value, then compared with the effort required, which he calls perspiration. He says it is a riff on decision-making ideas from Annie Duke, and it works as a heuristic as well as a formal framework.
“So first you want to map all the possibilities, like what are the problems you want to solve, and what are the possible ways to solve that problem?”
Listen to the episode Strategy & market Link to this Report a problem
Product AI and go-to-market AI should be centralized separately, with everything from marketing through customer support as one group. Listen
Kyle Norton says the AI applied to how the product team works, and the AI built into the product, is a separate effort from go-to-market. He says that everything market-facing, from marketing through to customer support, should be treated as one basket that needs central data management and intelligence building. A host adds that in services businesses, where product and go-to-market are intertwined, you centralize even further.
“from marketing all the way through to customer support is a separate basket.”
Listen to the episode Strategy & market Link to this Report a problem
Engineers who still write 20% or more of their code by hand are very behind the market. Listen
A speaker says that if 20% or more of your code is still handwritten, you are very behind the market, and that the top 10% of engineers are no longer writing code manually. Another participant replies that he doesn't know the specific benchmark but agrees that more than 20% handwritten is a problem.
“20% or more of your code is handwritten still, you are very behind the market.”
Listen to the episode Hiring & team building Link to this Report a problem
AI-native companies are changing commission plans every quarter, with usage, credit and outcome-based pricing changing most. Listen
AJ Bruno says he plans to write about how AI-native companies change their commission plans each quarter. He says the variables changing most are usage, credits and outcome-based pricing, and that these companies are still working out what makes their margins viable.
“What we're seeing with, and I'm going to write the top line newsletter on this, AI native companies and commissions are changing their comp every quarter.”
Listen to the episode Pricing & packaging Link to this Report a problem
71% of SaaS companies enter the year without quotas, and some AI-native companies don't set quotas at all. Listen
AJ Bruno describes a shift where some AI-native companies have sales cycles shrinking from about 12 months to three because buying and budget open immediately, and they set rates and percentages rather than quotas. Another participant counters that whether a company pays commission depends on the value of its stock options. When Monday.com is cited as not paying commissions even as a public company, a participant says it didn't work and that only the fastest-growing companies in history can do it.
“71% of companies enter the year without quotas.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
Kyle Norton has moved from bearish to very bullish on Salesforce because it has become the hub for integrations, partners, governance and AI. Listen
Kyle Norton says Salesforce's acquisitions, including Bluebirds, Momentum and Informatica, show it is all in on AI. He says the ecosystem gives Salesforce an advantage as the pivot point where integrations, partners, roles and permissions sit. Another participant adds that growing a $40 billion business may be hard to show in public markets, but that this doesn't stop Salesforce from delivering value to CROs and sales teams.
“I went from being a bear on Salesforce 24 months ago to being like very bullish on Salesforce now”
Listen to the episode Strategy & market Link to this Report a problem
Within six to twelve months, Kyle Norton expects the average employee to work much like Steve, who runs an OpenClaw instance with context files and skills. Listen
Kyle Norton says Steve has an OpenClaw instance with context files, skills and APIs connected to everything. He expects tools like Claude's co-work and dispatch to bring this level of AI use to many more people. He predicts this will show up in GDP numbers over the next 12 to 24 months. Another participant adds that it will expose people who don't adopt it.
“I think the average employee in the next six to 12 months will be much more like a Steve.”
Start AI-driven sales work by building first-party and third-party data foundations before anything else. Listen
Kyle Norton recommends that every company start with good first-party and third-party data. He says without it the models cannot be given what they need to make smart decisions or produce good outputs. He describes this as the foundation that the rest of his AI work at Owner.com is built on.
“The recommendation I give to everybody is start with data. You have to start with good first party and third party data.”
Personal AI use gives faster emails, notes and decks, but not business use cases that move the needle. Listen
A host says decentralized AI gives the consumer benefits that individuals feel, such as emails written faster, meeting notes and decks built faster. He says it does not produce use cases that can be shown to move the business in a substantial way and are ready for production. He describes the output quality as depending on each person's taste.
“But you don't get like real business use cases, like the ones that you can say, ah, this move the needle for the business in a really substantial manner”
The predictive models Owner.com built for customer size and lead scoring are rebuilt and improved continuously, not set once. Listen
Kyle Norton says his team does not just update enrichment every day but also improves the scaffolding. He describes overhauling the first model that predicts customer size, with a new senior data scientist on her first major project and new data brought in. He says the team is also building new lead-scoring models.
“and not just every day enriching and updating the information, but actually every day making that scaffolding better.”