(1) Logan Junger: The Startup Mindset That Makes AI Actually Work in CS

From Spreadsheet Dump to Scrappy Builder

Logan's first experiment with AI in customer success was, by his own admission, a failure. It was early days after OpenAI made ChatGPT widely accessible, and like many CS leaders riding the hype, he went big. He dumped a spreadsheet of client data into the tool and asked it to find the signals for customers at risk. No context, no framework, no structure. It didn't work.

He abandoned the approach and decided to make more incremental bets. That false start taught him something important: AI isn't a big brain you can just point at a problem. It's a small brain that needs skills, context, and guardrails before it's useful.

But the seed of Logan's AI approach was actually planted years earlier, in the startup world. Working at early-stage companies with ambitious goals and limited resources, he developed a builder's instinct — a bias toward hacking things together with whatever was available rather than waiting for the perfect tool.

That instinct proved to be the real unlock when AI arrived. Where others were trying to find the one killer AI use case, Logan was already wired to think in workarounds and iterations. AI just made his existing approach faster and higher quality. Email drip campaigns that used to take days and read poorly now came together in a fraction of the time with better copy. Internal tools that would have required an engineering team could be spun up in an evening.

What's Actually Working

One story captures Logan's approach perfectly. He was working with a CS team of five covering about a thousand customers and needed to notify all of them about account redistributions — personalized emails from their new CSMs. There was no marketing platform, no bulk email tool, nothing. So in one evening, he used Replit to build a mini email server. It connected to a private Google Cloud API for security, had a terminal-style interface for uploading CSVs, supported customizable templates with merge tags, and could send emails on behalf of individual reps. It wasn't pretty, but it worked — and it's still functional today.

"A lot of times I'm going in to empty space, to places where things should be happening, customers should be kept up to date, they should have interactions — and there's just nothing." — Logan Junger, Founder at Jungwell

His proudest current project takes this even further. He's building out a CS program from scratch at an organization that had a single support-oriented rep. What previously would have required hiring a team of five, Logan has architected using AI across both the customer-facing and operational sides. The sole CSM now has clear priorities, automated record-keeping, and the infrastructure to know what work needs to be done and what the results are — without drowning in spreadsheets.

The obstacles along the way have been just as instructive. At Gusto, Logan saw the potential in mining customer communications across channels for risk signals and expansion opportunities. He never got the cross-functional buy-in to centralize everything into one LLM, but he did hack together his own projects to start gleaning insights. CS teams, he notes, are often treated as an afterthought when it comes to resourcing and rev ops support — and that gap creates both the frustration and the necessity that drives creative AI adoption.

He also learned to flip the standard AI adoption playbook. While most companies were telling employees to find one way to introduce AI into their workflow, Logan told his team to start everything in AI. Learn where the capacity constraints are, understand the limitations, and then you'll naturally find where the real value lives.

"I kind of flipped it and said: start everything in AI. Start to understand its capacities and limitations, and then you'll get a clear idea of where the value opportunity really is for your workflows." — Logan Junger, Founder at Jungwell

The Hard-Won Lessons

The first lesson is about expectations. Over-ambition is itself an obstacle. Logan watched a CSM accidentally paste the ChatGPT dialogue prompt — the "let me know if you'd like any changes" line — directly into a client email. People stumble when they treat AI as a finished product rather than a tool that needs human judgment layered on top.

The second is about staying human at scale. Logan thinks a lot about what happens when every vendor has access to the same AI signals. If five different companies detect that a customer's usage dropped because they went on vacation, and all five send identical "we noticed you've been less active" messages the same week — that's not customer success. That's noise. He's experimenting with the oblique strategy method to introduce pattern disruption into his prompts — injecting unpredictability back into AI-driven interactions.

What's Next

Logan is cutting social media and doubling down on two things: learning to build with AI (he's deep into Claude Code right now, building tools despite having no technical background) and investing in face-to-face human connections. He sees a coming surge in physical-world interactions as a counterbalance to the AI-everything trend — more coffee meetings, more co-working spaces, more in-person events.

Looking ahead, he's watching the customer feedback loop closely. Founders who claim they've replaced their CS team with AI agents are, in his view, in for a reckoning. But he's intellectually honest enough to admit that what humans do in that feedback loop — listening, pattern-matching, applying past experience — isn't insurmountable for AI to learn. That tension is where the interesting work lives.

"You should always be looking to replace yourself. There's a big difference between you controlling it and having it thrown upon you. If you're out in front, constantly looking for ways to leverage AI to take off the workload, you also have control over the next thing you're going to take on." — Logan Junger, Founder at Jungwell

The choice, as Logan sees it, is simple: be the one in the driver's seat charting your own future, or wait around for someone to tell you what's next.

The Sonora Project captures how CS leaders are navigating AI — the real experiments, practical wins, and hard-won lessons shaping the future of customer success.