Podcast Summary: "I gave Claude Code our entire codebase. Our customers noticed."

How I AI | Host: Claire Vo | Guest: Al Chen (Galileo)
Date: April 6, 2026

Episode Overview

In this episode, host Claire Vo sits down with Al Chen, a field engineer at Galileo, to dive deep into how giving AI tools direct access to a company's codebase can radically improve customer support, streamline internal workflows, and elevate the overall customer experience. Al walks through his hands-on process of equipping Claude Code with the entire (multi-repo) Galileo codebase, leveraging Confluence, Notion, and Slack for additional context, and developing workflows that empower both customer-facing and technical teams. The conversation uncovers practical tactics, scripts, and mindsets that allow non-engineers to deliver highly technical answers—fast, accurate, and tailored to customer needs.

Key Discussion Points & Insights

The Challenge: Meeting Complex Customer Needs with Evolving Codebases

• Al Chen's pain point: Public documentation and generic AI answers were insufficient for nuanced, technical customer queries.
  • (“The minute I realized I couldn't really do my job was when I was trying to reference our public documentation … still wasn't coming up with [the] answer my customers were looking for.” - Al, 00:00)
• Customers want step-by-step explanations, not vague documentation.
  • (“They don't want the doc's answer. They want the step by step answer of how all these services cascade together.” - Claire, 00:11)

The Solution: Pulling Down the Entire Codebase & Using Claude Code

• Multi-repo environment: Galileo isn't in a monorepo structure—15 separate repos, which Al locally aggregates.
• Script creation: With Claude Code, Al automates pulling the latest main branches from all repos—no manual git pull for each!
  • (“I'm opening up the script right now… it’s like what, 16 lines? Didn't have to write this. I just said help me figure out a way to pull the latest main branches into my local repos.” - Al, 00:29, 09:37)
• Benefits:
  • Query the entire, up-to-date codebase in VS Code for accurate, context-rich answers
  • Reduces internal engineering team interruptions to near zero
  • Facilitates self-learning of the codebase

Memorable Moment

• Quote: "We can now all live in a little bit more chaos because the AI navigates all that information for us across systems... you can be in your code querying Confluence." (Claire, 00:38 / 15:00)

Best Practices: Making AI & Code Work Together

Aggregating Context Across Tools

• Code, Confluence, Notion, and Slack threads are all ingested for comprehensive, AI-driven answers.
• System can contextually tailor responses to individual customer quirks, based on micro-docs and customer-specific notes in Confluence.
  • (“I've seen from working with our DevOps team... when it's tailored to specific security requirements and deployment requirements, it's way more effective and just gives the customer more trust…” - Al, 13:00)

Custom Claude Code Commands

• Example: A DPL command for deploying into customer-specific environments, prioritizing Confluence then scanning code if needed.
• Al uses customer trait databases ("customer quirks pages") to further contextualize answers.

VS Code/IDE Setup

• Open the project at the collective repo/parent directory level to enable cross-repo queries with CLAUDE or other assistants.
  • “I don't think people use this trick enough... loading a project at the multi Repo level as opposed to at the individual repo level if you're trying to answer questions across the product is really important.” - Claire, 09:54

Beyond the Product: Supercharging Customer Experience

AI as a Competitive Edge in Customer Relationships

• Personalized, context-rich answers—faster and more accurate than conventional support.
• Notable Quote:
  “You can actually use AI to invest and compete on customer experience … when you don't just say here are our general docs ... instead you say, I heard you, I understand what your needs are and here are your custom docs on how you specifically need to deploy this.” - Claire, 17:04

Automation of Knowledge Base Creation

• Using Pylon, generate help articles directly from Slack support threads.
  • “When you're looking at a conversation like this in Pylon or in Slack … what Pylon allows us to do is … generate a help article. And right here I already have one...” - Al, 26:50
• This “living” knowledge base is more current and precise than official docs, and content is drawn from real customer interactions.

Virtuous Workflow Loops

• “And then” automation: Each customer interaction can trigger a cascade of actions—documentation, internal training updates, SEO, etc.
  • “If you had a perfectly staffed team, what would you do next? … then I would turn that into an article … then share with customer success ...” - Claire, 29:18

Human in the Loop: Where Do People Add Value?

Proofreading, Context, and Judgment

• Al doesn’t blindly copy Claude’s answers; he edits for tone and brevity, double-checks technical depth, and ensures “human-ness.”
  • “I don't just blindly copy and paste the answers I get from cloud code to my customers … I still try to proofread everything and make it sound more human… removing things that make it seem like it’s from a bot.” - Al, 20:30
• For deeply technical or sensitive cases, human review and engineering input are still essential.

Relationship and Empathy

• AI tools aren’t “fun to hang out with,” and customers want real relationships with vendor teams.
  • “You want to know that you have somebody to call...enjoy working with that person.” - Claire, 22:45

Tips for Spreading and Scaling the Practice

Al’s Internal Evangelism

• Al actively shares scripts, IDE setups, and workflow tricks throughout the Galileo team; uptake is organic.
  • “I'm just constantly sharing these tips and tricks to my teammates to make sure they're also functioning at their capacity.” - Al, 32:32

Democratizing Access—Not Just for Engineers

• Non-engineers should embrace loading up codebases, understanding git, and querying their repos; it's now a hard skill for everyone.
  • “No matter what role you're in… You gotta like learn a little bit how to code…” - Claire, 37:51

Notable Quotes & Memorable Moments

• "You got to split your quota with Claude code. That's really what we need to do … Coin operated Claude. That's going to be my new, my new skill." — Claire, 01:06 & 44:42
• "I treat it like my entry level analyst... I am relentless when it comes to asking AI to do things for me, especially when it comes to answering customer questions." — Al, 42:18

Timestamps for Important Segments

• 00:00–03:30 — The problem: why standard docs & even traditional AI assistants fail
• 03:33–06:45 — How Al leverages 15 repos and Claude Code for customer questions
• 08:32–10:54 — Automating code updates, creating a "pull all" script with Claude
• 11:47–15:00 — Custom commands and contextual cross-referencing with Confluence & code
• 16:26–20:06 — Aggregating knowledge sources and tailorable customer support
• 24:10–29:18 — Reactive support in Slack: from threads → AI-generated help docs → knowledge base
• 32:32–34:00 — Scaling best practices & team education/internal adoption
• 35:29–39:38 — Persuading engineering to share code access; upskilling non-engineers
• 42:18–44:42 — Prompting for better answers; 'think hard/thinking harder' prompt strategy
• 44:51–45:14 — Closing remarks, hiring pitch

Takeaways & Practical Tips

• Aggregate your full codebase (multi-repo or monorepo) in your IDE to maximize AI’s effectiveness.
• Automate code pulls with AI-generated scripts to save hours of toil.
• Combine AI with all knowledge sources—code, docs, Confluence, Slack—for truly in-depth, tailored support.
• Use AI to transform support threads into living documentation instantly and scalably.
• Proofread AI responses for accuracy and humanity—customers appreciate the personal touch.
• Evangelize these practices within your organization: demo, document, and coach.
• Non-engineers: the era of code-illiteracy is over—you don’t need to be a dev, but you need to be technical.
• Prompt engineering matters: “think harder,” “explain your reasoning,” “show sources,” etc., to guide AI to better output.

Final Thoughts

This episode provides actionable, real-world insights for anyone in customer-facing or technical enablement roles looking to leverage AI to bridge the gap between evolving codebases and ever-more-demanding customers. The methods Al shares can be adopted today for radically improved efficiency, learning, and customer satisfaction.

"There's no better time to learn how to code. Truly no better time to learn how to actually code... You have this like magic, super patient, infinitely wise teacher in your computer that you can use to learn to code."
— Claire Vo, 37:51

Connect with Al Chen on LinkedIn, and check out open roles at Galileo if you’re interested in these kinds of field engineering challenges.

[End of summary]