About CPA.com
CPA.com is a New York-headquartered company, founded in 2021, that operates at the intersection of accounting and technology. The company partners with accounting and auditing firms to help them adopt modern, AI-powered solutions — building and delivering SaaS products that bring innovation to a traditionally conservative, compliance-heavy industry.
Data privacy and security are central to how CPA.com operates. Working with audit and accounting partners means handling sensitive financial data, which places security, isolation, and compliance at the top of every infrastructure decision. “Data privacy is one huge thing to us,” says Airfield Deng, Manager of Data Engineering. “Making sure there is no leakage and that the isolation of the data is extremely important.”
What do they do?
Industry
Location
About CPA: Bringing AI and technology innovation to accounting professionals
CPA.com is a New York-headquartered company, founded in 2021, that operates at the intersection of accounting and technology. The company partners with accounting and auditing firms to help them adopt modern, AI-powered solutions — building and delivering SaaS products that bring innovation to a traditionally conservative, compliance-heavy industry.
Data privacy and security are central to how CPA.com operates. Working with audit and accounting partners means handling sensitive financial data, which places security, isolation, and compliance at the top of every infrastructure decision. “Data privacy is one huge thing to us,” says Airfield Deng, Manager of Data Engineering. “Making sure there is no leakage and that the isolation of the data is extremely important.”
A centralized warehouse feeding reporting, operations, and AI
Airfield Deng leads data engineering at CPA.com, acting as the architect of the data warehouse. His mandate: bring data in from a wide range of sources and put it in the right place — Snowflake — so that downstream consumers across the business can work with it reliably.
Data flows in from multiple directions: partner CRM systems, lakehouses, internal tools like Stripe and Zendesk, and MongoDB. Once in Snowflake, it powers:
Operations and customer support: Zendesk ticket data is loaded into Snowflake, where an AI semantic layer built on Snowflake Cortex helps the team categorize issues by product, identify recurring problems, and surface summaries for the support team — without requiring a data analyst in the loop.
Finance: Stripe transaction data flows into Snowflake, giving the finance team real-time visibility into billing and revenue.
Executive reporting: dashboards and automated reports reach as high as the COO and technology VPs, generated at frequencies ranging from twice daily (9am and 5pm) to weekly and monthly summaries.
Self-serve analytics: a text-to-SQL layer built by Snowflake Cortex lets non-technical teams – like the RevOps team – ask questions in plain language and get answers instantly, without needing to involve a data engineer or analyst.
Reports are delivered via Tableau, Power BI, and Python-based automations running on AWS, depending on team and use case.
The challenge: Dozens of hand-coded pipelines, constantly breaking
Before Hevo, every data pipeline at CPA.com was built manually by Airfield and his team. For each new project, each new partner connection, and each new reporting requirement, someone had to write code, connect to endpoints, manage authentication, and keep it running.
As the business grew and the number of partners, data sources, and use cases multiplied, the approach stopped scaling. Endpoints changed. APIs were updated. Pipelines broke. And the person responsible for fixing them was the same person responsible for building new ones.
"As the business grew, it became really hard for me to manage a dozen pipelines. And these pipelines break quite easily, they're very fragile because of all the moving parts."
Syndicating data out to partner CRM systems compounded the problem. Sending leads and opportunity data from the warehouse back to Salesforce required hand-coding against a notoriously unstable API — and the result was pipelines that worked until they didn’t, at which point someone had to diagnose and rebuild them.
Manual report compilation added another layer of cost. Data analysts spent 30 to 40 minutes pulling data from multiple spreadsheets to assemble a single report — time that multiplied quickly across a large and growing report catalogue, with human error baked into every step.
Hevo: Budget-friendly, easy to set up, and reliable from day one
CPA.com evaluated multiple ETL platforms before choosing Hevo, with data security at the top of the criteria list alongside ease of use, reliability, and cost. After testing Hevo on a proof-of-concept – connecting MongoDB pipelines, Salesforce integrations, and multiple other sources – the results were clear.
"After we tested a few demo projects : connecting to MongoDB, to Salesforce, multiple pipelines, it worked out pretty well. It turned out to be very stable and pretty reliable. That’s how we got into Hevo, and we’ve continued using it ever since."
Two things stood out during evaluation: Hevo’s pricing was significantly more budget-friendly than alternatives like Fivetran, and Hevo Activate: the reverse ETL capability for syndicating data out to CRM and marketing platforms was a differentiator at the time when comparable features were not yet available elsewhere
The impact: Automated reporting, eliminated errors, and engineering time back
Human error removed from reporting
With pipelines running automatically through Hevo into Snowflake, the manual work of assembling reports from spreadsheets is gone. Airfield estimates that a single report automation, which takes around five hours to build, once replaces one hour of analyst work every week, compounding into hundreds of hours saved over time. Across a large catalog of reports, the cumulative impact is significant.
"Before, the accounting team would have to compile reports from multiple spreadsheets, which could take 30 to 40 minutes. Now they just wait for the report to arrive — with almost guaranteed accuracy. And we’re only talking about one report. We generate a huge number of reports."
Report frequency increased without extra effort
Because data flows automatically and continuously — syncing bi-hourly from source systems — report frequency is no longer constrained by how long it takes to compile data manually. Teams that previously received weekly summaries now receive twice-daily updates, with end-of-day snapshots at 5pm giving every department a close-of-business view.
Data engineers focused on higher-value work
With Hevo handling the operational complexity of pipeline management, Airfield and his team can focus on the work that actually moves the business forward: designing AI-driven pipelines, building semantic layers on top of Snowflake, and enabling self-serve analytics across the organization.
Hevo-powered data as the foundation for AI-driven products
CPA.com is deeply invested in AI both as a product it sells to accounting firms and as a capability it runs internally. The reliable, centralized data that Hevo pipes into Snowflake is what makes both possible.
Text-to-SQL with Snowflake Cortex
A semantic layer built on Snowflake Cortex lets non-technical teams ask questions about business data in plain language. The system generates SQL, runs the query, and returns answers in seconds — without routing requests through a data analyst or engineer. Teams like RedOpt can now get the data slice they need immediately, rather than waiting for a report to be built.
AI-powered Zendesk ticket analysis
Zendesk support ticket data flows from the source into Snowflake via Hevo. An AI layer built on top categorizes tickets by product, surfaces recurring issues, and generates summaries for the customer support team. What would otherwise require manual review of hundreds of tickets is handled automatically, and emerging problems surface faster.
Vector database and beyond
CPA.com also uses Snowflake’s vector database capabilities as part of its AI stack, with multiple parallel AI-driven projects underway. The Hevo-managed data infrastructure serves as the foundation: without reliable, up-to-date data in Snowflake, none of the AI applications would have the inputs they need to function.
"Without Hevo, it’s not like it’s impossible — but it will become a very slow process. The whole AI layer on top of Snowflake depends on having clean, fresh data there in the first place."
What keeps CPA.com on Hevo
Stability and reliability: Hevo proved itself in the proof-of-concept phase and has remained the stable core of CPA.com’s ETL infrastructure. For mission-critical data serving finance, operations, and executive reporting, reliability is non-negotiable.
Ease of use: setup is straightforward, and the platform’s low-code environment means pipelines can be configured and launched without deep engineering overhead — important for a lean data team managing many sources.
Cost-effectiveness: compared to alternatives evaluated at the time, Hevo’s pricing fit CPA.com’s budget as a growth-stage SaaS company, without sacrificing the features needed for a complex, multi-source data stack.
Excited to see Hevo in action and understand how a modern data stack can help your business grow? Sign up for our 14-day free trial or register for a personalized demo with our product expert.
