Modal’s growing product usage required reliable account resolution
As Modal’s AI infrastructure platform scaled, so did the volume of signals its GTM team needed to interpret. To understand the companies behind growing product activity, the team needed account-level data. Its GTM model depended on keeping account teams informed on high-value accounts, identifying self-serve customers ready for a sales handoff, and automating work across its broader customer base. Without that data, expansion signals could go unnoticed.
When Chris Prinz joined Modal as its first GTM engineer, he began building the data system and workflows that could turn product usage into action. “We needed to figure out which customers to prioritize next,” he recalls. “That meant connecting our signups back to companies.”
Product signals were highly fragmented and difficult to access
Modal’s source data was organized around individual users and product workspaces rather than company-level accounts. To link product usage to the right users and companies, Chris connected signups, work and personal emails, and event activity. He enriched those records with attributes for segmentation and scoring, and sent the results back to the systems where sales and marketing could use them.
Sales and marketing also relied on contact and account data to plan events, build lists, and conduct account research. Chris fielded ad hoc requests through his coding agent, so each request depended on his availability. The data also needed to be precise once it reached other teams. “We can’t waste reps’ time or leave money on the table because the data isn’t accurate,” he explains.
Enrichment had to fit a code-managed GTM pipeline
Account data depended on enrichment, but that step was the least visible part of the system. Before Chris joined Modal, parts of the process ran through Clay tables and an agency-led setup. As he built a broader GTM system in code, he couldn’t easily inspect and version the table-based workflow alongside the rest of Modal’s data logic. The workflow also sat outside that code-managed system, preventing a coding agent from managing the pipeline end to end.
As GTM requirements and workflows changed, failures became harder to trace. When records didn’t reach Attio as expected, he was left to handle troubleshooting and reruns manually. As a result, it was becoming clear to Chris that Modal needed more composable GTM infrastructure.
Retaining workflow logic within Modal’s codebase
Chris could build and own the workflow logic himself, but every new enrichment source risked adding another integration to build, evaluate, and maintain. He needed a way for agents to test and use external data without tying the GTM system to a separate integration for every provider.
Last year, Deepline’s team had reached out to Chris about test driving the early product. When Deepline publicly launched in March of 2026, it offered one API to reach many enrichment providers, which fit the exact system that he wanted to design at Modal. Chris tested it the week it launched, then quickly onboarded Modal and began ramping up.
“The other data brokers I considered were moving into orchestration, and I wanted data access separate from that. We wanted to own our code and orchestration stack.”
Deepline transforms product activity into actionable account context
Chris built Modal’s GTM system in code, using Deepline to access external enrichment and research providers. With one key and one bill, Modal could now evaluate and call multiple data sources through the same code-owned workflow. Deepline handled provider access, while Chris retained control of the selection logic, orchestration, and downstream data models.
1. Turn a first enrichment test into a production pipeline
Chris started by defining the fields required for segmentation, including company funding and contact job title, along with the criteria for judging the output: coverage, accuracy, and cost. Using Deepline’s documentation and an API key, he asked Codex to build and test the first workflow against those criteria. He used the test to estimate how spend and coverage would scale. “A day later,” he recalls, “I had a fully operational production pipeline in our codebase.”
2. Connect fragmented activity to account-level context
With the initial setup complete, Chris manages enrichment through a SQL table of candidate records. His script retrieves contact and company attributes through Deepline, skipping records enriched within the previous month.
In Snowflake, Modal’s dbt project matches those records to product users, event registrants, companies, and workspaces. The pipeline then sends scores, segments, and usage data to the CRM and lifecycle marketing tools. That gives Modal’s GTM teams the context to identify who is using the product, understand the signals around each account, and decide where a sales or lifecycle action is warranted.
3. Choose the right data source for each enrichment job
To improve account hierarchy data, Chris selected 200+ customers across different company stages and verticals, then used Codex and Deepline to evaluate dozens of providers against that benchmark. In less than 30 minutes, he compared parent-company and subsidiary relationships across data sources and analyzed conflicting results. This evaluation surfaced an optimal solution that used Parallel Web Systems with structured web research to outperform data providers that claimed to specialize in that problem.
This combination improved data accuracy by 19% for 98% lower cost per result than alternative providers. Without Deepline, Chris would’ve had to contact providers, request samples, define evaluation criteria, and run the evals himself. Instead, he could now leverage a coding agent to execute the full workflow against Modal’s exact needs.
Chris applies this same framework when adding new enrichment fields. He compares coverage, precision, recall, and cost per accurate result across providers and waterfalls, then investigates records where sources disagree. In turn, Modal can now select the source or combination that performs best for each job instead of defaulting to a single provider.
4. Give sales and marketing self-serve access to account research
Once Deepline was connected to Modal’s GTM systems, Chris extended the same data access to an internal Slack assistant. Built with Hone and connected to Modal’s CRM, email campaign tools, and Snowflake, the assistant handles ad hoc requests, so teams no longer wait for Chris to complete each task.
Sales and field marketing requests go straight to the assistant, which uses Deepline to query enrichment providers for contacts or fill data gaps, such as missing emails and phone numbers. Account executives use it the same way to enrich their own lists. These teams gain faster access to the account context they need to act on product, event, and engagement signals.
“Field marketers can go into Slack and describe the event they want to run. The agent uses Deepline to find contacts and works with the marketer to refine the list.”
15% more contact coverage than the closest evaluated provider
With Deepline, Modal’s GTM system connects product activity to the people and companies behind it, enabling GTM teams to prioritize outreach, support expansion, and act on emerging signals. Modal retains ownership of the workflows and business logic behind that system, while Deepline provides flexible access to the external data its pipelines and agents need.
40% decrease in cost per qualified enriched contact
Evaluating providers through Deepline helps Modal easily find the optimal setup for accuracy, coverage and cost. Now, Chris can seamlessly run evals across many permutations of enrichment providers and effectively pick the best combination.
15% more contact coverage than the closest evaluated provider
More usable account and contact information with valid data attached can reach Modal’s GTM workflows.
140+ tasks handled by Modal’s internal agent in the last 30 days
Self-service research moves routine list-building and enrichment requests out of GTM engineering’s queue.
70-100 hours saved per month with Deepline’s agent & enrichment platform
Without it, Chris would’ve had to upskill his team into other GTME tools or increase headcount.
19% accuracy lift for account hierarchy data at 98% lower cost per result
Combining Parallel Web Systems with structured web research outperformed other providers.
Building Modal’s next workflow without rebuilding the stack
Looking ahead to the coming year, Chris plans to extend Deepline to additional enrichment pipelines while continuing to build Modal’s workflow logic in-house. The internal assistant will also keep handling routine research and list-building requests in Slack, reducing the number of tasks that reach GTM engineering.
As Modal’s needs change, Chris can define a new outcome, evaluate the relevant data sources, and add the resulting automation to Modal’s codebase. Deepline gives his team a repeatable path from new GTM data needs to code-owned automations, without building and maintaining a new provider integration each time.
“With Deepline, we can build the workflows our business needs next, without waiting for a vendor’s roadmap to catch up.”
