Build a Self-Repairing Context Loop for AI Data Agents

Part of Agentic Analytics: Validation and Context

Hosted by Shane Butler, Sravya Madipalli, and Hai Guan

Wed, Sep 9, 2026

5:00 PM UTC (45 minutes)

Virtual (Zoom)

Free to join

Invite your network

Go deeper with a course

Automate AI Evals with Claude Code
Shane Butler
View syllabus

What you'll learn

Detect when an answer is drifting

Spot the signal that an agent's number is off: the same question returning different results across runs.

Have the agent fix its own definitions

Let the agent trace the bad answer back to a stale metric definition and rewrite it in the repo.

Re-run and watch the variance collapse

Run the same question again against the repaired context and see the answers converge on one number.

Why this only works with repo-based context

Definitions live in files the agent can read and edit, so a fix persists instead of vanishing when the chat ends.

Why this topic matters

Most AI analysis quietly rots: a metric gets redefined, the agent keeps answering from stale context, and nobody notices until the number is wrong. The fix is not more prompting. It is a loop where the agent detects the drift, repairs its own definitions in the repo, and re-runs until variance collapses. We build that loop live, so you watch the agent catch itself and land on one honest answer.

You'll learn from

Shane Butler

Co-founder, AI Analyst Lab

Shane Butler is a Co-Founder of the AI Analyst Lab. Previously he led evaluation strategy for AI product development in the legal tech domain. He has more than ten years of experience in product data science and causal inference, with prior roles at Stripe, Nextdoor, and PwC. Shane is also the co-host of the AI podcast Data Neighbor, where he interviews product, data, and engineering leaders who are pioneering the next generation of data science and analytics in an AI-driven landscape.

Sravya Madipalli

Senior DS Leader (Ex-Microsoft)

Sravya Madipalli is a Senior Manager of Data Science with 14+ years of experience helping teams make better decisions with data. She has built and led data science and product analytics teams at Microsoft, eBay, Nextdoor, and Superhuman (prev. Grammarly), working closely with product, engineering, marketing, and leadership. Her expertise spans experimentation, metrics design, modeling, analytics, and translating complex user behavior into clear, actionable insights.

Hai Guan

Head of Data at Ontra, Ex-LinkedIn

Hai Guan leads the data organization at Ontra, the leading legal tech AI solutions for private markets. He previously led Data Science & Analytics at LinkedIn, Nextdoor, Pinterest, and Meta. He's spent a decade teaching product development teams how to ask questions that actually drive decisions—and now teaches how to combine that judgment with AI to move 10x faster.

See all products from AI

Sign up to join this lesson

By continuing, you agree to Maven's Terms and Privacy Policy.