Over the past year, numerous companies, including Intuit Mailchimp, have been exploring the concept of vibe coding.

Intuit Mailchimp specializes in offering email marketing and automation services. It is part of the broader Intuit group, which has been steadily advancing its use of generative AI over recent years, introducing its own  GenOS and  agentic AI capabilities throughout its various business units.

Despite having its proprietary AI capabilities, Mailchimp has found situations where vibe coding tools are necessary. The impetus was a pressing need to meet an extremely tight deadline.

Mailchimp had to showcase a complex customer workflow to stakeholders without delay. Conventional design tools such as Figma were insufficient for producing the required working prototype. Some engineers at Mailchimp had already been quietly experimenting with AI coding tools. Under the deadline pressure, they decided to apply these tools to a real-world business challenge.

AI Scaling Hits Its Limits

Power caps, rising token costs, and inference delays are reshaping enterprise AI. Join our exclusive salon to discover how top teams are:

• Turning energy into a strategic advantage

• Architecting efficient inference for real throughput gains

• Unlocking competitive ROI with sustainable AI systems

Secure your spot to stay ahead: https://bit.ly/4mwGngO

“We encountered a fascinating scenario where we needed to swiftly prototype some features for our stakeholders, dealing with a fairly complex workflow,” said Shivang Shah, Chief Architect at Intuit Mailchimp, to VentureBeat.

The Mailchimp engineers utilized vibe coding tools and were impressed by the outcomes.

“Typically, something like this would take us days to accomplish,” Shah explained. “We managed to complete it in just a few hours, which was quite remarkable.”

This prototyping session catalyzed Mailchimp’s broader integration of AI coding tools. Now, by leveraging these tools, the company has accelerated development speeds by up to 40%, gaining critical insights into governance, tool selection, and the importance of human expertise, which other enterprises can adopt immediately.

The evolution from Q&A to ‘do it for me’

Mailchimp’s experience highlights a larger shift in how developers engage with AI. Initially, engineers used conversational AI tools for basic advice and algorithm recommendations.

“Even before vibe coding became popular, many engineers were already using conversational AI tools to determine if a particular algorithm was suitable for the problem they were addressing,” Shah observed.

With modern AI vibe coding tools, the paradigm has fundamentally shifted. It’s no longer just about answering questions but about actually performing some of the coding tasks.

This transition from consultation to delegation embodies the core value proposition that enterprises are navigating today.

Multi-Tool strategy beats single-vendor approach

Mailchimp opted for a diverse array of AI coding platforms rather than relying on a single solution. The company employs Cursor, Windsurf, Augment, Qodo, and GitHub Copilot, informed by an understanding of specialization.

“We realized that depending on the stage of software development, different tools offer distinct advantages or expertise, akin to having a specialized engineer working alongside you,” Shah stated.

This approach reflects how enterprises utilize various specialized tools for different development phases, avoiding a one-size-fits-all solution that may excel in some areas while lacking in others.

This strategy arose from practical experimentation rather than theoretical planning. Mailchimp discovered through use that different tools excelled at different tasks within their development workflow.

Governance frameworks prevent AI coding chaos

Mailchimp’s most significant lesson regarding vibe coding centers on governance. The company established both policy-based and process-integrated safeguards that other businesses can adopt.

The policy framework includes responsible AI reviews for any AI-based initiatives involving customer data. Process-integrated controls ensure that human oversight remains central. AI may conduct initial code reviews, but human approval is mandatory before deploying any code to production.

“There will always be a human involved,” Shah emphasized. “There will always be someone who needs to refine it, verify it, and ensure it truly addresses the right problem.”

This dual-layer approach addresses a common concern among enterprises. Companies want the productivity benefits of AI while maintaining code quality and security standards.

Context limitations require strategic prompting

Mailchimp discovered that AI coding tools have a notable limitation. These tools understand general programming patterns but lack specific business domain knowledge.

“AI has learned from industry standards as extensively as possible, yet it might not align with the existing user journeys we have as a product,” Shah noted.

This realization led to a crucial understanding. Successful AI coding requires engineers to provide increasingly specific context through well-crafted prompts, drawing on their technical and business expertise.

“Ultimately, you still need to grasp the technologies, the business, the domain, and the system architecture. AI amplifies what you know and what you can achieve with it,” Shah explained.

The practical implication for enterprises: teams need training on both the tools and how to effectively convey business context to AI systems.

Prototype-to-production gap remains significant

AI coding tools excel at rapid prototyping, but Mailchimp found that prototypes don’t automatically translate into production-ready code. Integration complexity, security requirements, and system architecture considerations still demand significant human expertise.

“Just because we have a prototype, we shouldn’t assume it can be completed in a certain timeframe,” Shah cautioned. “A prototype does not equate to being ready for production.”

This lesson helps enterprises set realistic expectations about the impact of AI coding tools on development timelines. The tools are invaluable for prototyping and initial development, but they are not a magic solution for the entire software development lifecycle.

Strategic focus shift toward higher-value work

The most transformative impact wasn’t merely speed. The tools enabled engineers to concentrate on higher-value activities. Mailchimp engineers now dedicate more time to system design, architecture, and customer workflow integration rather than repetitive coding tasks.

“It allows us to spend more time on system design and architecture,” Shah explained. “Moreover, it helps us integrate all the workflows for our customers, reducing time on mundane tasks.”

This shift indicates that enterprises should measure AI coding success beyond just productivity metrics. Companies should evaluate the strategic value of work that human developers can now prioritize.

The bottom line for enterprises

For companies aiming to lead in AI-enhanced development, Mailchimp’s experience underscores a critical principle. Success involves treating AI coding tools as advanced assistants that enhance human expertise rather than replace it.

Organizations that achieve this balance will secure sustainable competitive advantages. They’ll find the right mix of technical capability with human oversight, speed with governance, and productivity with quality.

For enterprises planning to implement AI coding tools later, Mailchimp’s journey from urgent experimentation to systematic deployment offers a proven roadmap. The fundamental insight remains unchanged: AI amplifies human developers, but human expertise and oversight are vital for production success.