From Rolodex to Revenue Engine: How We're Scaling Sales Across an Entire Portfolio of AI Companies

For the first several years of building Soar, almost every customer we signed came through my personal network or a direct reach-out on LinkedIn. That’s not an accident. I’ve been doing tech startups since 1990, and I’ve spent a lifetime cultivating relationships. Today I have about 7,000 contacts in my phone and Google Contacts, and 74,000 followers and connections on LinkedIn. That’s a lot of doors to knock on.

The problem with building a company on personal relationships: it scales with one person. One person isn’t enough anymore.

Soar was always designed to be more than a single product or a single market. We’re an AI Studio: a parent company building a portfolio of AI-powered subsidiaries, each one deploying fleets of AI agents into a specific vertical. As those companies grow, so does the sales challenge. Ten salespeople can’t all lean on my Rolodex. They need their own pipeline, and each subsidiary needs its own repeatable motion that turns marketing spend into scheduled calls with organizations ready to buy.

The reason we structured Soar as an AI Studio is that the same core technology (fleets of AI agents that help every person on a team do their job better) works across completely different buyers. Rather than try to sell everyone with a single message, we built focused companies.

Workplace AI is our enterprise play. We help mid-size to large companies deploy AI agent fleets across their workforce: agents that handle the workload, execute on decisions, and give every employee an AI co-pilot calibrated to their role. The average client isn’t buying a tool. They’re changing how their team works.

Founder Portal brings the same model to startups. Founders get AI feedback on their pitch deck, guidance on which angels and VCs are most likely to fund them, and real-time insights after every investor meeting so they can get sharper with each one. An AI co-pilot walks them through every step from idea to exit. Starting a company is hard enough. We want to tilt the odds.

Faith Portal serves churches and nonprofits. These organizations run on mission and community, but they’re often under-resourced for the operational and communications support AI agents handle well. FaithPortal gives faith communities tools that large enterprises take for granted, built around the values and needs that matter most to them.

Education Portal takes the fleet into higher education, private labeled and customized for each university or college, serving students and alumni organizations. Every institution has its own identity and its own goals. EducationPortal is built around each school’s brand and outcomes, not dropped in as a generic product.

Four markets, four buyer profiles, one platform underneath all of them.

Right now, there are hundreds of thousands of businesses, nonprofits, universities, and startups in the U.S. that are going to buy or build AI agent functionality. The market isn’t arriving. It’s here. The question is who builds a go-to-market motion that scales.

Our head of marketing cut his teeth buying media at Google and Facebook. He knows how to optimize spend. Last month, our cost per sales meeting was $75. Last week, it dropped to $37. We expect to push it below $25 soon, and AI agents are a big reason why: the same agents we’re selling are helping us sell them.

Run the math. If our close rate is 10% and the average sale is $12,000 per year for a team of 10 users, then every dollar we spend driving meetings has a clear and measurable return. Lower the cost per meeting, hold the close rate, and you’ve got a lever you can pull as hard as your capital allows. Multiply that across four subsidiaries in four verticals.

If these metrics hold (and I believe they will), we might hire far more than 10 salespeople this year.

I watched this dynamic play out up close at Ancestry in the early years. We scaled our call center to 800 people. Why? Because the math made it obvious. The average rep was generating $120 per hour by upgrading customers from trials to paid, or from monthly to annual subscriptions.

We loved annual subscriptions. When a customer pays for the year upfront, your cash flow transforms overnight. Monthly subscriptions were a constant headache. Credit cards expire. Cards max out. Dunning logic is painful to maintain. We spent enormous energy just trying to figure out why payments weren’t going through.

Annual upfront is better for the customer, better for the company, and frankly better for your sanity.

I remember a conversation with Josh James back in his Omniture days, when the company was still called Superstats. He made a decision that looked reckless from the outside: he stopped chasing volume in small-business web analytics and went all-in on enterprise. At the time, he had exactly two enterprise clients: eBay and CNET.

He bet the company on it.

Years later, Adobe paid $1.8 billion for Omniture.

The lesson I took from Josh is that a focused motion built on repeatable sales processes and long-term contracts creates a different company than a high-volume, low-touch churn model. The unit economics are different. So are the relationships. The compounding is too.

But Josh’s situation isn’t ours. Omniture sold one product. We built Soar as a portfolio specifically so that different verticals get purpose-built solutions. A 40-person church staff and a 4,000-person enterprise have different needs, different budgets, and different timelines. They shouldn’t be shoved into the same product and told to figure it out. FaithPortal exists because faith organizations deserve a product built for how they actually work. Same for education. Same for founders. The focus isn’t about who’s big enough. It’s about who we can serve deeply.

This year we’re carefully analyzing every GTM lever across the entire portfolio:

Paid media is the most measurable. Our marketing team is optimizing ad spend daily, and AI agents are helping us do it faster and cheaper than any human team could alone. We expect cost per meeting to keep falling across all four subsidiaries.

In-person events still matter more than people admit. There’s something about a handshake and a shared meal that accelerates trust in ways that a sequence of LinkedIn messages can’t. Whether that’s a founders conference for FounderPortal, a higher education summit for EducationPortal, or an enterprise leadership event for Workplace AI, we’re being intentional about which events we attend and why.

LinkedIn and personal reach will always be in the mix for me. But the goal now is to systematize what I do naturally: identify the right organizations, personalize the outreach, follow up with discipline. And deploy that across a full team operating in four different markets simultaneously.

We expect real growth this year. The market for AI agent fleets is here, the demand is building, and we have a system that’s working across the whole portfolio, not just one company.

The businesses, universities, churches, and startups that figure out the sales engine first will define this market. I’ve been on both sides of that lesson. I’m not interested in learning it again the hard way.

If your organization is ready to explore what a fleet of AI agents could do for your team, we’d love to start the conversation. Visit soar.com to find the right subsidiary for your needs.

Paul Allen is the co-founder of Ancestry.com and CEO of Soar, an AI Studio building a portfolio of AI companies — including Workplace AI, FounderPortal, FaithPortal, and EducationPortal — to uplift humanity. He has been building technology companies since 1990.