AI is changing how software gets built—but the real story is what happens when a PLM company uses AI every day to ship faster, catch issues earlier, and then turns those lessons into better AI for customers.
In Episode 3 of Inside the Solution, Propel’s insider series on how to solve today’s most pressing manufacturing challenges, Propel CMO Dario Ambrosini sits down with CTO Kishore Subramanian to unpack how Propel uses AI across the entire software development lifecycle—from requirements and “stories” to coding, testing, release notes, and even support.
Then Kishore connects the dots to Propel One: how internal AI practices (like AI-assisted reviews) inspire agentic workflows in PLM/QMS, so product teams can spot risk earlier, reduce downstream chaos, and move faster with guardrails.
Key Takeaways
AI didn’t just speed up coding, it rewired the whole development process.
Kishore explains the shift of the last 6–8 months: code generation has matured fast, so the bottleneck moves upstream. The focus “shifts left” toward defining the what (requirements, intent, guardrails) because the how (implementation) is dramatically easier.
Vibe coding builds prototypes. AI-assisted coding ships enterprise software.
A clear distinction: vibe coding is fast, prompt-driven, and often accepted blindly—great for experimentation. Propel’s approach is AI-assisted coding, where engineers act like an orchestra conductor: setting direction, enforcing guardrails, reviewing outputs, and ensuring production-grade quality and security.
The Propel team use AI from requirements to deployment, even for testing and release notes.
Propel applies AI throughout the SDLC: turning concepts into requirements and stories, accelerating coding, improving unit/integration test coverage, generating and expanding test cases (including corner cases), and supporting release readiness work like risk assessment and release notes.
Context is everything: generic AI doesn’t know your product.
Kishore highlights a core lesson: you can’t get production value from a generic tool without context. Propel invests in training/internal knowledge + guardrails so AI understands Propel’s product reality—rather than guessing from public data.
Propel's internal ‘product expert’ AI assistant that scaled customer support and onboarding.
Propel built an internal AI agent, “Propel Sage," to answer “how-to” questions across support, engineering, QA, and implementation teams—speeding onboarding, improving customer responses, and deflecting questions away from engineering.
Efficiency on top of efficiency: fewer interruptions, more building.
AI helps developers build faster, and also reduces the constant drain of internal Q&A. The compounding effect: more engineering time goes into product improvements instead of repeating explanations.
The big "aha" moment: code review maps to ECO review.
A standout moment: Propel uses AI agents for code review to catch issues early—cheaper than fixing bugs later. Kishore draws a direct parallel to manufacturing: apply the same idea to ECO/change reviews by using agents to flag risk, highlight downstream impact (supply chain, production, compliance), and recommend actions before problems turn into scrap, rework, delays, or audit pain.
Learn More
Want to see how these internal learnings show up in the product?
Explore Propel One to learn how Propel’s agentic AI brings the same “catch it early, move faster with guardrails” mindset to PLM and QMS, so your teams spend less time on tedious work and more time making the right decisions.
About Inside the Solution
Inside the Solution brings you behind-the-scenes conversations with Propel's product experts on solving modern manufacturing challenges. Each episode breaks down product development problems across roles, teams, and industries — and the innovative solutions transforming how companies bring products to market.
.png)
