The Day I Weaponized Our Data Warehouse

One Tuesday morning, I found myself staring at three years’ worth of sales conversations in Gong. I knew there had to be a way to combine them with AI for real sales coaching, not as a gimmick or just another summary tool. Four hours later, SE Buddy was born.

What started as a quick side project for me to improve my own performance is changing how our SEs grow. Juniors are ramping faster, mid-levels are finding ways to elevate their game, and our black belts are sharpening their edge. In the first month alone, we saw deals progress from intro call to a meaningful next step 15% faster—that’s roughly $500,000 in pipeline per SE—from something I built over a lunch break. Here’s how, and why, I did it. 

The problem every Solution Engineer faces

There are a few challenges practically every SE team faces. First, there’s a significant skills gap between the team’s strongest performers and everyone else—and because an SE’s performance is tied to deal success, this gap directly influences the outcome of a deal. Best practices also typically stay locked in top performers’ heads because there isn’t usually a system in place for them to share it with others. 

At Sigma, we sit on a mountain of perfectly organized, carefully modeled, data. I knew there had to be a way to use it to boost the performance of our Sales team and get them to 100%.

Coaching to level the playing field is mostly lacking, and feedback gets softened or skipped because emotions and social dynamics get in the way. So, companies hire pricey external sales coaches once a year who don’t know your business in depth, and provide generic training that maybe gets SEs to 60% of where they need to be. But what SE teams really need is continuous improvement, and coaching that’s raw, honest, and unbiased. 

All that got me thinking. At Sigma, we sit on a mountain of perfectly organized, carefully modeled, data. I knew there had to be a way to use it to boost the performance of our Sales team and get them to 100%. We just needed to ask the right questions.

The discovery that changed everything

As a principal SE, I’m always looking for ways to get better and often block off time for call coaching. The day I built SE Buddy, I realized that it just isn’t scalable for us to keep grinding through transcripts one by one. So, I opened Sigma. 

I’d been toying with the new AI() function, which lets you run an LLM against any row of data. So instead of copying one transcript into ChatGPT and manually prompting it, I could now run that same workflow across each one automatically. Imagine having a list of thousands of calls—AI() lets you score and analyze all of them at once with your own prompt. That’s when it clicked: I didn’t need to coach calls one at a time anymore. I could coach the whole backlog at enterprise scale.

Under the hood: The secret sauce of SE Buddy's AI-powered coaching

To test my theory, I pulled up one of the worst calls I could find and piped it through Sigma’s AI Query function. I was amazed—it didn’t just summarize what happened, it understood it. It made me think that if I could do this with one transcript, I could do it with all of them. I could bring in sensitive, confidential context safely, and finally get feedback that was truly usable. 

Think of it like giving a coach unlimited access to your company's brain, so it can deliver feedback that’s spot-on for your world.

It all starts with Sigma's AI Query function. Think of it like giving a coach unlimited access to your company's brain, so it can deliver feedback that’s spot-on for your world. At its core, AI Query lets me run powerful LLMs, specifically Claude 4 Sonnet, directly against our cloud data warehouse (CDW). The beauty here is security and scale. My data never leaves our environment and I have full access to all our internal treasures—customer calls, sales enablement decks, bootcamp materials, customer stories, and transcripts from our top performers—as well as tribal knowledge like how our best SEs crush discovery, or how they run flawless proof-of-concepts (POCs).

To make it work, I built a series of "knowledge bases" that operate as curated collections of content that I feed into the LLM. For each area, I combined external best practices like widely published MEDDPICC frameworks with our internal trade secrets. For example, I pulled together all our discovery materials, high-performer transcripts, and tribal tips on scoping POCs, then fine-tuned it with public sales methodologies. The result was a "Discovery Knowledge Base" that gives the AI hyper-specific context. I did the same for others: Business Value Assessment, Competitive, Demo Best Practices, Objection Handling, POV Best Practices, Sigma Product, and Tribal Wisdom. The screenshot below shows all these knowledge bases ready to plug in to the AI query.

‍

What I built: SE Buddy (the enterprise AI weapon)

Looking at all that potential, I just had to build this into something the whole team could use.

Here’s how it works: AI ingests any call transcript, analyzes it, and scores it instantly across six Sigma-specific dimensions—discovery, demo flow, objection handling, technical depth, business value alignment, and call close. Then it gives the SE coaching they can use right away, like how each call went, what they did well, what they need to improve, and the specific actions to improve.

The magic happens when SEs start seeing this feedback every day, and begin to understand what good looks like. They make the move from unconscious incompetence to being consciously competent, and become aware of every small detail like how they introduce themselves, answer questions, or recap value. This helps them improve in a focused way.

When my manager saw how well SE Buddy worked, the first thing he said was, “Call coaching is just the start. This approach can scale across AE's, CS, and BDR's.” Then we showed it to our CRO. He reviewed the output from a couple of calls, and said, “This is really good coaching. We need to deploy this to the whole team.” 

Quickly, other department heads started coming to me and asking “Can you build one for customer success? Can you make one for onboarding?” That’s when it hit leadership: every process can be AI-powered. 

Why this changes everything

SE Buddy changed the trajectory of our entire team, closing that skills gap and levelling up our entire organization. Competitors simply won’t be able to catch up, and those who don’t operationalize AI for their GTM teams won’t be around for long. This is a race to create, deploy, and weaponize AI applications. Sigma is the only way to do it with speed, security, and scale.

The only question is—will you weaponize your data warehouse, or let it sit there like an expensive filing cabinet?

The only question is—will you weaponize your data warehouse, or let it sit there like an expensive filing cabinet? For anyone considering building something similar, I’d say start with one repeatable process, plug it into Sigma + AI, and see where it takes you. Every day you wait is a day someone else gets stronger.

‍