Modal

The challenge: Everyone could query. Few people did.

Modal's business and culture runs on data — from observability metrics, to container telemetry, to user activity. To capture all of it, the team built sophisticated infrastructure including two data warehouses with millions of events ingested per minute and real time dashboards tracking product performance. However, even with such comprehensive data, the team needed to lower the barriers for data accessibility to increase the demand for data.

Before Modal had a robust data stack, team members ran analyses directly on production databases and in local scripts. Most people at Modal could write SQL and/or Python, including business users, but knowing how to write queries did not lead to more people exploring questions. Spinning up a notebook, finding the right data, constructing the logic and validating the output could easily turn a five-minute question into a half-day project. For ad hoc and exploratory questions, the effort rarely felt worth it and many times people would not pursue the analysis at all.

As Modal's team grew, the team needed a way to make data exploration as effortless as asking a question.

The solution: Hex became Modal's analytics operating system

Modal wanted to give users a way to answer questions without creating a new notebook for every ad hoc question. Hex made that possible with a lightweight setup on top of existing data models, the ability to share reports, and a Slack integration that enabled every team to explore data on their own.

Joining two data warehouses in Hex, saved months of a data warehouse migration

Modal had a dual data warehouse setup with Snowflake and ClickHouse to avoid migrating large tables. Where Snowflake held the majority of Modal's data and powered most of its analytical work, ClickHouse managed observability, ingesting 1 to 2 million events per minute across 500 billion logs, and powering customer-facing dashboards that tracked function performance, call timelines, and latency percentiles.

Hex's native ClickHouse connector meant Modal could see both data sources in one place without migrating to a single warehouse or building a platform themselves.

“Having a native ClickHouse connection in Hex allows us to blend data with Snowflake without having to sync over a large amount of data” - Kenny Ning, Head of Data

Modal set up the Hex agent with less than 50 lines of context

Instead of using a semantic layer to get reliable analytics, Modal's team directs Hex toward endorsed dbt models alongside a workspace rules file and several guides. Hex's agent identified and analyzed accurate data with minimal context setup, supporting conversational analytics across the organization. It took Kenny less than a day to set up the Hex agent with <50 lines of context.

Today six functions across Modal trust Hex’s agents for daily reporting and complicated ad hoc analyses that can break revenue down to the line item. Modal continues to add context for Hex’s agents to reference by using projects as context. This simple context setup can be fully managed by a one person data team.

"The barrier to getting reliable insights is much lower than I thought. Endorsed dbt models and a workspace rules file were enough to support our conversational analytics instance.” - Kenny Ning, Head of Data

Hex's agent increased active users across six functions

When it came to wide-spread adoption across the company, Hex's Slack integration made the biggest impact. Team members could now ask @Hex a question directly in Slack and get a data-backed answer. Especially for non-data scientists, having a way to easily access and analyze data unlocked the organization’s potential. And the numbers back it up.

The breadth of adoption shifted how the whole company interacted with data. As Modal's go-to-market and sales teams grew, the Slack integration became their primary tool for business updates. Instead of submitting formal requests, teammates got immediate answers and insights from Hex when they needed them most. The Modal team could now make strategic, and data-backed decisions during meetings, between calls, and while building campaigns.

"Having such quick access to data about our customers and being able to share it quickly changes how we built products. Instead of long arguments, we just ask Hex, and make a decision and move on". - Can Duruk, Product Engineering at Modal

As the team continued to use Hex, Kenny referenced Hex’s Context Studio to monitor how users engaged with the agent and ensure the context powering those answers stayed accurate as the organization grew.

Today, as adoption continues to grow, the use of Hex at Modal spans 6 functions, including: data, leadership, growth, product engineering, business operations, and sales. Additionally, the VP of Marketing and Head of Sales have become daily active users.

Modal's team drove 1,300 AI messages — split evenly between the Notebook Agent and Threads — with 53% of all Threads conversations starting directly in Slack without a formal rollout or training.

“The Slack integration removed the lag between question and answer. Our teams now get real-time insights in meetings, which has dramatically reduced follow-ups and helped us move faster on critical decisions.” - Kenny Ning, Head of Data

Modal's VP of Marketing opened $100M in pipeline starting with a screenshot

When Modal's CEO, Erik Bernhardsson, shared a tweet listing the top 50 AI neolabs, David Dorman, VP of Marketing and Growth, saw an opportunity to explore the list in Hex. He wanted to know: how many of these companies were already on Modal? Which ones had meaningful spend? Which accounts should they focus their outreach on?

Before Hex, this question would have been a multi-day project and given to someone to manually research. Instead, Hex was able to extract data from an uploaded image, infer domains, cross-reference Modal's user base, and segment companies by spend level. Hex then mapped the list of companies against Modal’s customer database, prioritizing companies worth targeting and ones with expansion opportunities. The entire process took less than 24 hours, which was 75% faster than having the same analysis completed manually. This analysis identified 15 well funded neolabs that had Modal accounts but no usage, collectively representing $100M+ in spend potential.

"Hex is magical. It changed what our team thinks is possible with data. The bottleneck used to be getting to the data, now it's just deciding what to do with it." — David Dorman, VP of Marketing and Growth

The Impact: From reactive requests to proactive intelligence

Hex became more than a BI or notebook tool, it became a thought partner that evolved how the organization interacted with data, unlocking exploratory analyses that would have not come to fruition otherwise.

With Hex, Modal:

• Opened $100M+ in pipeline by analyzing an image and targeting outreach for the top 50 AI neolabs
• Connected 2 data warehouses on one analytics platform
• Enabled 6 cross-functional teams to work with data independently, including leadership, sales, marketing, engineering, design and people teams
• Increased conversational analytics with 1,300 agent uses with the Notebook Agent, Threads, and the Slack integration
• 53% of all Threads use originates from the Slack Integration
• Built less than 50 lines of context, eliminating the need to adopt or maintain a separate semantic layer
• Compressed a multi-step analysis by 75% from days to under 24 hours, enabling the CEO to reach out to high-value prospects on the same day
• Unlocked analytical capabilities and analyses that wouldn’t have been pursued due to complexity and prioritization

As Modal grows, Kenny’s goal is to help Hex’s agent get smarter and unlock new analytic capabilities without making the context more difficult to manage. Modal will continue to increase reliability and self-service with Hex’s agents by leveraging projects as context to scale knowledge across the business.

“What’s most powerful is that the impact scales. When everyone can access and act on data instantly, the whole business moves faster — and that’s where the real value comes from.” - Kenny Ning, Head of Data

Kenny Ning, Head of Data and David Dorman, VP of Marketing and Growth, Modal