S13 E4: The AI GTM Engine: Turning Intent Signals into Warm Outbound Pipeline with Connor Heggie, Co-Founder & CTO of Unify

Episode Summary

Executive Summary: Connor Hegge, co-founder/CTO of Unify, explains how the company evolved from a side project mining unstructured data into an AI-driven go-to-market platform for sales teams. The episode focuses on product thesis, MVP trade-offs, hiring philosophy, multiple infrastructure re-architectures, and why human+AI systems can supercharge—not replace—salespeople.

Main Topics: Origin story: from listicles to AI for sales (Priority: 5/5): Connor describes building AI-written product listicles before ChatGPT, which sparked his interest in extracting meaning from unstructured data. That insight, combined with his cofounder Austin’s go-to-market experience, led to Unify’s mission to help sellers find the right prospects at the right time. Product thesis: sales as a search problem (Priority: 5/5): The company frames sales not as persuasion, but as identifying companies and people that already have a problem the product solves. Unify’s core bet is that LLMs and software can make this discovery process repeatable, observable, and scalable. MVP build and early technical trade-offs (Priority: 4/5): The first version was built in about two and a half months by a small team, with a deliberate choice to build deeply scalable CRM and enrichment integrations rather than shipping a throwaway prototype. The team prioritized future durability over speed. Roadmap decisions driven by thesis and customer pain (Priority: 4/5): Connor emphasizes balancing founder conviction with customer feedback. Some features were obvious from demand, while others—like sequencing—were built from internal conviction despite investor skepticism because they completed the end-to-end workflow. Hiring philosophy and team culture (Priority: 4/5): Unify hires product-minded engineers and people who genuinely care about the mission and customers. Connor says the best teammates take problems personally, show ownership, and help the company maintain high integrity while moving fast. Scaling challenges and infrastructure evolution (Priority: 5/5): The team repeatedly reworked its data architecture as load grew: first Postgres, then Elasticsearch, then a move to a columnar store. Each shift reflected the difficulty of combining high-write CRM data with high-volume web traffic and flexible querying across custom schemas. Future direction: durable agentic workflows for GTM (Priority: 5/5): Unify is building its own agent harness for sales and go-to-market work, beyond code-generation-style agents. The long-term vision is proactive background agents that help businesses continuously discover opportunities and act on them at scale and reasonable cost.

Key Arguments: Unify’s product should be understood as a search and orchestration system, not a sales automation toy; its purpose is to find real buying opportunities and support the workflow end-to-end. Founder conviction matters when customers cannot yet articulate the solution, as seen with the sequencing product, which became a major differentiator despite initial resistance. Building for scale early can pay off if the system is likely to be core infrastructure; Unify’s early investment in integrations prevented later rewrites from becoming catastrophic. Sales teams are not replaced by AI; instead, AI makes each rep more productive, which can increase demand for skilled salespeople rather than reduce it. Hiring for care, ownership, and customer empathy is as important as hiring for technical brilliance because culture determines how teams respond when things break. Infrastructure choices must match the true load profile; high-write/high-update systems with flexible querying may require abandoning popular tools when they hit architectural limits. The future of go-to-market software is agentic: durable background systems that continuously surface, enrich, and act on signals rather than merely assisting with one-off tasks.

Data Points: Company founding date: January 17, 2023 - Connor says Unify started right after ChatGPT's launch period, though the idea predated it. Time to build MVP: 2.5 months - The initial CSV upload, enrichment, and Salesforce write-back workflow was built by the founding team in about two and a half months. Early side-project runtime: 15 minutes - The pre-ChatGPT AI listicle prototype took about 15 minutes to run. Early side-project cost: $50 per listicle - Connor recalls each generated listicle costing around $50 in compute/API usage. First customer CRM scale: 5 million records - The first customer integrated with Unify had 5 million Salesforce records, validating the decision to build for scale. CRM ingestion delay: Up to 15 minutes - Unify’s core data freshness constraint for customer CRM ingestion. Founding engineering team size: 2 engineers plus founders - Connor says two engineers joined about three weeks after the company started, alongside him and Austin. Postgres stability window: About 7 months - The first Postgres-based architecture worked for the first seven months of customers before degrading. Growth rate: 20% month over month - Connor mentions rapid growth while describing the Elasticsearch phase. Optimization effort: 6 weeks - Connor spent about six weeks rewriting/query-optimizing the Postgres query engine to keep the system alive. Data ingestion frequency: Every 15 minutes - Customer CRM records were updated at high frequency, contributing to the architecture challenges. Payout horizon for new architecture: Nearly 9 months - The columnar-store architecture has held for almost nine months at the time of the interview. Advisor cadence: 45 minutes to 1 hour weekly - Connor meets his advisor Jeff Quasey every week for a long advisory session.

Pivotal Quotes: "we were going to have to run this seed stage company's technology as our primary database, self-hosted on our Kubernetes cluster. Very scary proposition." — Connor Hegge: Describing the final breaking point when evaluating reverse-index technology for their database layer. "the job to be done of sales is to find people in companies that have a problem that my product uniquely solves." — Connor Hegge: His reframing of sales as a search/problem-matching system rather than persuasion. "we do these things not because they are easy, but because we thought they would be easy." — Sam (founding engineer): A mug inscription Connor cites while discussing the naivete of the team’s early scaling decisions.

Implications: The episode suggests the next wave of GTM software will be AI-native, data-rich, and agentic, but success will depend on deep infrastructure, strong product judgment, and teams that care enough to keep rebuilding when scale breaks the first design.