Dr. Helen Gu on Building InsightFinder, AIOps, and the “Last Mile” of Enterprise AI

In this episode of Triangle Tweener Talks, we unpack what it really takes to go from professor to CEO, how InsightFinder built trust in a skeptical enterprise market, and where LLMs help (and don’t) when you’re dealing with machine telemetry data. They also explore multi-agent workflows, “composite AI,” practical enterprise adoption hurdles, and Helen’s advice for students navigating an AI-shaped future.Highlights coveredHelen’s origin story: NASA Pathfinder work → distributed systems reliability → ML-based predictionThe Google chapter: being invited to evaluate anomaly-detection algorithms with SRE teamsBootstrapping InsightFinder via NSF/SBIR funding + early angels, before raising traditional VCThe professor-to-CEO transition: prioritization over “balance,” and learning to adapt dailyWhy founders should lead early sales (especially when the product is new-to-the-world)How InsightFinder runs enterprise PoCs using a “replay mechanism” on historical incidents“Composite AI” + using LLMs to translate technical insights into understandable narrativesIf you’ve ever wondered what “AI that actually works” looks like in the enterprise, and how a research-driven founder earns trust at Fortune scale, this one’s a must-listen.Timestamps00:02:12 — Intro to Helen + what InsightFinder does00:04:32 — Helen’s background at NC State00:05:49 — Google discovers the research00:06:24 — NSF/SBIR bootstrap + company start00:07:10 — Early ML roots (since 2000)00:08:54 — NASA Pathfinder origin story00:12:03 — Teaching + student questions evolving00:13:28 — Student → PhD → InsightFinder spark00:14:36 — Professor + CEO time management00:17:39 — Learning sales as a founder00:21:24 — Funding path: SBIR + angels + first VC00:22:44 — IDEA Fund connection story00:24:19 — LLM era impact + “composite AI”00:26:45 — LLMs as the interface layer00:28:20 — Plain-English explanation of InsightFinder00:31:04 — Agent workflows (Jira, probing, reports)00:32:31 — Multi-agent + SLM orchestration00:35:32 — PoCs: dogfood + replay mechanism00:37:41 — How early detection works (hours ahead)00:39:00 — Series B + scaling go-to-market00:43:00 — LLMs: maturity + “last mile” problem00:45:30 — Fine-tuning + trust risks00:47:14 — Advice for students + fundamentals#TriangleTweenerTalks #TriangleStartups #NCState #AIOps #Observability #SiteReliability #SRE #DistributedSystems #MachineLearning #EnterpriseAI #LLMs #AgenticAI #MCP #StartupJourney #FounderStories #B2BSoftware #DeepTech #RaleighDurham #NorthCarolinaTech--- This episode of Triangle Tweener Talks is hosted by Scot Wingo, presented and produced by Triangle Tweener Fund, with creative assets and design support from Walk West. We couldn’t share posts like this without our amazing sponsors:  Gold Sponsors: - Balentine: https://www.balentine.com/triangle-entrepreneurs - EisnerAmpner: https://www.eisneramper.com - Robinson Bradshaw: https://www.robinsonbradshaw.com  Silver Sponsors: - Automated Consulting Group: https://automated.co - Bank of America: https://business.bofa.com/en-us/content/technology-industry-group.html  ------Triangle Tweener Talks is sponsored by:Atomic Object: https://atomicobject.com/