Conversation
Highlights

How Buildxact’s CRO Rebuilt Their Sales Motion for the AI Era

The unit economics of SMB sales have fundamentally broken. Liam Fraser, Chief Revenue Officer at Buildxact, wasn’t dealing with an optimization problem—he was staring at a structural collapse in how construction management software gets sold.

“An outbound for SMB used to be hire an army of SDRs, buy tons of lists and just have them dialing 100 numbers a day and hope for the best,” Liam explains in a recent episode of BUILDERS. “No business today could start with that in mind. The economies of that just do not meet the needs of where we’re at today.”

Buildxact serves small and medium-sized general contractors building custom homes. Their software ingests construction plans, automates takeoffs (measuring plans to calculate materials), generates high-fidelity estimates, and manages the project lifecycle through scheduling, subcontractor coordination, and back-office integrations. With an inbound-heavy motion historically driven by SEO, they were one of the market leaders in their segment.

Then the traffic patterns started shifting. Not gradually—materially. Qualified prospects were bypassing Google entirely, going straight to ChatGPT, Claude, and Gemini to research construction management software.

The question wasn’t whether to implement AI. It was how to systematically rebuild their entire sales motion without chasing hype.

Optimizing for LLM Citation Architecture

Buildxact’s response wasn’t to tweak meta descriptions or add FAQ schema. They retooled their entire content corpus for what Liam calls AEO and GEO—Answer Engine Optimization and Generative Engine Optimization.

This required rethinking information architecture for citation patterns. How do LLMs parse, weight, and surface information? What content structures increase probability of citation? How do you position against competitors when the “results page” is a synthesized paragraph?

“We’ve basically retooled all of our content and sort of hacking that experience, and we’re seeing really fantastic results,” Liam shares. “A lot of our leads now come in going, oh, you guys were the first to come up in my search on ChatGPT and gave you a really strong profile and recommendation.”

The validation came directly from pipeline. Prospects weren’t just finding Buildxact through LLMs—they were explicitly mentioning it as their discovery channel, often noting the quality of the recommendation context provided. This suggests Buildxact isn’t just appearing in LLM responses; they’re being cited with strong contextual positioning.

For revenue leaders watching search behavior migrate, this represents an inflection point. The playbook isn’t “SEO plus AI”—it’s a fundamental replatforming of how you structure, present, and distribute information for citation engines.

Decomposing Workflows to Atomic Automation Units

With outbound economics broken, Liam’s team needed a new architecture. His framework starts with radical decomposition: break every workflow to its smallest repeatable component, then systematically automate each unit.

“I like to think of what should we think about with AI,” Liam explains. “And it’s looking at tasks and taking them down almost to the atomic unit as close as you can, and then think about how they can get automated. And the more those tasks are being done by a human being, the better, and I think the greater leverage and the return of putting in an agent or an automation to take advantage of that.”

For outbound, atomic tasks mapped to: list acquisition → data enrichment → ICP matching → message personalization. They deployed Clay for the enrichment and matching layers, creating a system that processes leads without human touch until the message is ready to send.

The differentiation came in personalization depth. Instead of surface signals—hiring patterns, funding announcements, LinkedIn activity—Buildxact pulls from their own product data. Not generic vertical insights, but specific operational context about the prospect’s business model, typical project profile, and likely pain points based on market segment.

“We’re able to take that and personalize that message, even bringing some of our own data out of our own product,” Liam notes. “So that when we put that message out, that first icebreaker message out as a cold outbound email, it’s arriving to Brett and you’re feeling like we know a lot about you and your business and we see incredible open rate and the ability to turn that into an open conversation with our sales reps has improved out of sight.”

This creates asymmetric advantage. Competitors can replicate the automation stack—Clay, enrichment APIs, LLM-powered personalization. They can’t replicate first-party product intelligence deployed programmatically at the point of outreach.

Building Methodology-Specific Coaching Infrastructure

Beyond pipeline generation, Liam’s team deployed AI across the sales execution layer. Automated call summaries and follow-up emails recover hours of rep time weekly—table stakes at this point.

The sophisticated application is in coaching infrastructure. They built an AI agent with a custom scoring rubric that evaluates calls against Buildxact’s specific sales methodology—not generic conversation intelligence metrics, but their defined gold standard for qualification, discovery, and demo execution.

“We built a rubric and we built an agent that you can profile a call and get a score against that so an individual rep can go in at the end of their call and assess themselves,” Liam explains. “How did that go against what we see as our gold standard? And then obviously the sales leaders are able to do that at some scale.”

Reps self-assess immediately post-call. Managers coach at scale across the entire team without manually reviewing recordings. The system provides both efficiency (automated actions) and quality improvement (consistent methodology application).

This requires upfront investment in codifying your sales methodology into evaluable criteria. Most conversation intelligence tools surface generic metrics—talk ratio, question frequency, sentiment. Buildxact’s approach embeds their specific go-to-market playbook into the evaluation framework itself.

Disciplined Experimentation Protocol

Not every implementation worked immediately. Buildxact deployed Sales Ape, an SMS-based AI agent, to re-engage prospects stuck in what Liam calls the “dead zone”—leads who won’t schedule qualification or pre-demo calls despite multiple touchpoints.

The thesis was sound: their ICP (contractors) are field-heavy, phone-averse during working hours, but responsive to text. An agent trained on company data could conduct asynchronous qualification via SMS for opted-in prospects.

Initial attempts failed. Rather than kill the experiment or persist indefinitely, they applied a structured protocol.

“Have a real experimentation mindset, be very clear about the goals that you want to see on running an experiment and a timeframe against that, and be prepared to move on good, bad, indifferent based on the outcome you get from it,” Liam advises.

They iterated on training data, refined prompt engineering, adjusted conversation flows. It’s now showing promise. The broader pattern: define success criteria and time bounds upfront. Measure ruthlessly. Decide explicitly to iterate or terminate based on data, not momentum or sunk cost.

Implementation Insights for Revenue Leaders

Liam’s career path—marketing background, product marketing, strategy, then sales leadership over the past 10-15 years—informs his approach to AI implementation. He’s never carried a traditional sales bag, instead focusing on team enablement and strategy.

That perspective shapes how he thinks about AI deployment across his 30-person revenue organization. It’s not about adopting every tool that enters the market. It’s about systematic identification of leverage points, disciplined experimentation, and infrastructure investment that compounds over time.

“Best thing you can do every day, and it’s something I challenge myself with is what’s something new I can learn today,” Liam shares. “I don’t care where that learning’s coming from. It could be talking to one of my SDRs and as much as it could be talking to a mentor.”

The result is a revenue organization that’s fundamentally transformed its motion while maintaining performance. Inbound pipeline flows through LLM discovery channels. Outbound runs at scale without traditional headcount. Coaching happens systematically with methodology-specific evaluation.

For B2B founders navigating AI implementation, Buildxact’s approach offers a tactical framework: decompose to atomic tasks, experiment with defined parameters, build infrastructure that codifies your specific methodology. The unit economics of traditional sales have shifted. The question is whether your go-to-market architecture will shift with them.