LangChain drove a 6-week migration, increased adoption and onboarding speed

The challenge: Legacy BI tooling couldn’t keep pace with exponential growth

Initially, the LangChain team had data sources across Segment, BigQuery, dbt, and Salesforce, but no easy way to access or build ad hoc insights outside of a few dashboards on their Legacy BI tool. Creating insights outside of structured dashboards required their sole data person to build abstractions and YAML files before any analysis could even begin.

This meant that non-technical users had to wait in the data team queue and weren't able to pull the data they needed at all. Technical users, on the other hand, had to stitch together workarounds with Claude Code and Python programs—querying the data warehouse directly. Getting answers was time consuming and sharing trusted insights was even harder.

On top of all this, the data team had zero visibility to the questions team members were asking and therefore could not build infrastructure to support more complex ad hoc questions and deep dive analyses. And as the company grew, so did the volume and complexity of questions—making it clear that how they were working was unsustainable.

It was clear. LangChain needed a tool that made trusted data accessible for everyone. And so, they turned to Hex.

The solution: One platform that compounds context across the organization

At LangChain, Hex initially started as a tool for the SQL and python-fluent to have a more flexible environment for exploratory analysis. But over time, Hex adoption spread across the company and it became the platform that connected every data source and analytics workflow across technical and non-technical users.

With most users self-serving their data requests through Hex, the analytics engineering team shifted their focus to data ingestion, context curation, and data migrations. They used the Context Studio to gain visibility to the questions users asked, and updated workspace guides, warehouse descriptions, and semantic models based on Context Suggestions. Other cross-functional teams, like the forward deployed engineering team and customer engineering team, now have full visibility into their operations and product adoption without routing anything through the data team first. Today, 100% of the company has access to data via Hex.

"Context Studio changed how we manage agent performance. It surfaces gaps, proposes fixes, and suggests exactly where to add context. The agent improves without me auditing every conversation." — Emily Hawkins, Head of Analytics Engineering, LangChain

The legacy BI data migration was completed in 6 weeks

Over time, managing data across two platforms became cumbersome. LangChain ended their contract with their legacy BI platform and migrated everything to Hex. Emily Hawkins, Head of Analytics Engineering, was in charge of the deprecation process. Although she initially had concerns about the volume of migration work, it became one of the most productive exercises the data team ran all year.

The process started with an audit. Using metadata from the legacy platform, the team identified which dashboards were actively used and which had gone stale. Separate reporting instances that tracked the same underlying data for core company metrics, customer insights, and team operational dashboards were organized and consolidated into different data apps. Other dashboards that had not been touched in months were deprecated, cleaning up stale data and creating a trusted single source of truth.

Throughout the process, the Hex agent handled the heavy lifting. The team copied YAML files from the legacy platform, pasted them into the agent, and prompted it to rebuild dashboards in Hex. Within the first pass, the agent was able to recreate roughly 90% of the dashboards — including queries, charts, and structure — leaving only review and refinement to the team. They migrated and consolidated 60% of their dashboards and deprecated the remaining, completing the full migration in 6 weeks.

"Being able to copy YAML files and give them to the Hex Agent to fully rebuild the dashboards helped us complete 90% of the migration. It was the best migration I've ever done." — Logan Cochran, Analytics Engineer, LangChain

Increased org and customer health visibility drove double-digit gains in adoption and retention

Accessibility and accuracy are table stakes for the customer engineering team to drive adoption and retention. Without easy access to product telemetry and customer account information, the team was dependent on the data team and cycles of pulling data directly from BigQuery via Claude Code and Python scripts to get insights.

After getting access to Hex, Nithin Bose, VP of Customer Engineering, used Hex’s Notebook Agent to build the MVP of the team’s customer 360 dashboard over a weekend. Using natural language, he instructed the agent to join data from product telemetry, support cases, and account data into one single view. Iterating on the dashboards over time, Nithin used the agent to mature the insights from core adoption metrics to utilization insights for customer renewal conversations. With the Hex agent, Nithin’s team no longer needed to wait for the analytics engineering team to build abstractions and YAML files to build customer insights. Now team members can chat with the data app to dive deeper with exploratory analysis.

Nithin also used the Hex agent to build a customer maturity matrix, visualizing the strength and weaknesses of each customer account across key dimensions. This dashboard took only three hours to build — which was 88% faster than the estimated three days it would have taken to build with Python.

With more visibility and operational rigor, the customer engineering team saw double-digit percentage improvements for both customer adoption and retention in three months.

What used to be a combination of Claude Code and python pulling data directly from the warehouse now incorporates Hex as the single source of truth and distribution mechanism for all data. Nithin’s team builds 0-1 analyses in Claude Code and is able to ship it to Hex via CLI to maintain governance, shareability, and observability across all teams. The team even connected Hex to Notion via MCP to provide additional context for the agent. Hex serves as the data layer, while Claude Code allows the team to operate where they’re already working.

“Hex’s developer-first mentality has been awesome. Hex has great visual and design sensibility. The ability to use Hex in any way we choose is valuable because now we can connect into Hex via my favorite dev tools.” — Nithin Bose, VP of Customer Engineering, LangChain

Three business-critical dashboards built in one week with the Hex Agent

Before Hex, the deployed engineering team had fragmented visibility to how team members were distributed across regions and accounts. Neil Dahlke, Director of Deployed Engineers, knew this had to change, but he had no way to see which team members were assigned to accounts based on ARR, region, and churn risk. At the time, the data he needed for this lived across Salesforce, Segment, Metronome, dbt, and the legacy BI tool — none of it in one place or easy to join. When he cross-referenced the information in the legacy BI tool, he discovered that the data was not accurate.

To tackle this, Neil tried building what he needed in their legacy BI platform, but making a handful of changes took a week and produced insights that were not accurate. And fixing the data meant navigating the traditional drag-and-drop tool that wasn't built for the kind of modifications he needed.

After getting access to Hex, Neil imported his YAML files into the Notebook Agent and rebuilt his dashboards with additional insights. When stats looked off, he asked the agent why the numbers didn't match his systems of record. The agent then was able to find discrepancies and correct them. With Hex, Neil was able to build three dashboards — cumulative enterprise customer usage, an enterprise 360, and a deployed engineering team view — all within a week, using natural language prompts with the agent handling the SQL and python on the backend.

Now Neil opens Hex every Monday morning and is able to see how his team is performing, which accounts haven't had contact in months, and where new hires should be assigned. Hex unlocked insights that Neil didn’t have access to, which increased customer onboarding speed by ~30% and drove a ~50% increase in the team’s customer onboarding capacity.

"It took me a week to make several changes in our legacy BI tool that were nowhere near as impactful as everything I built with Hex's agents in a week." — Neil Dahlke, Director of Deployed Engineering, LangChain

The impact: from a data request queue to a data-fluent organization

What started as a more flexible analytics tool for SQL and python-native team members became the single analytics platform for every function at LangChain. The data team stopped fielding one-off requests and started building the context and infrastructure that let the whole company find trusted answers independently.

With Hex, LangChain achieved:

• 6 week legacy BI tool migration to Hex, migrating 60% of legacy dashboards and retiring the rest
• 90% of the migration completed by the Notebook Agent, converting YAML files from the legacy BI platform into production-ready Hex dashboards on first pass
• 100% of the company has access to data via Hex
• 88% reduction in build time for the customer maturity matrix, from an estimated 3 days in Python to 3 hours in Hex
• ~30% increased customer onboarding speed
• ~50% increase in the team’s customer onboarding capacity
• Three business-critical dashboards built in one week
• Double-digit percentage improvement in both customer adoption and retention in three months due increased visibility and operational rigor

LangChain's analytics engineering team is now focused on deepening the context layer that makes Hex's agent more accurate across the organization — adding semantic models, workspace guides, and dbt integrations that compound in value as more teams build on top of them. Next up: an automated insights agent using the Hex MCP alongside LangSmith's agent builder to proactively surface metric changes to relevant teams before they think to ask.

"Hex’s agent increased data access for everyone at LangChain, while giving us the confidence that users were getting accurate answers with trusted data. — Emily Hawkins, Head of Analytics Engineering, LangChain

Emily Hawkins, Analytics Engineering; Logan Cochran, Analytics Engineer; Neil Dahlke, Director of Deployed Engineering; and Nithin Bose, VP of Customer Engineering — LangChain