Ramp Adds Developer Tools to Platform With AI Coding Assistant

Spend management platform provider Ramp has launched a new AI coding agent, Ramp Inspect, alongside its accounting platform expansion.

Ramp Inspect joins a growing number of domain-specific coding assistants that companies are building to make developer experience a key differentiator in the enterprise fintech world.

For instance, “Stripe won by making developers love them. That playbook works in fintech,” Zach Bruggeman, a staff software engineer at Ramp, told The New Stack. “The CFO might sign the contract, but the engineering team decides if it actually gets implemented.”

For instance, last year financial services platform provider Block launched “Goose,” its AI coding agent, designed to help developers at all levels of sophistication — from non-developers to pros — easily build applications.

Block has made Goose available to its entire workforce of more than 12,000 people. Bruggeman said Ramp’s goal is to do the very same thing with Inspect and its employees.

Inspect writes code like any other coding agent, but closes the loop on verifying its work by having all the context and tools needed to prove it, as a Ramp engineer would, wrote Bruggeman, Jason Quense, principal engineer at Ramp, and Rahul Sengottuvelu, head of Applied AI at Ramp, in a blog post.

Regarding Inspect verifying its work: “For backend work, it can run tests, review telemetry, and query feature flags,” the Ramp post reads. “For frontend, it visually verifies its work and gives users screenshots and live previews. Agents should have agency, and so we made sure Inspect is never limited by missing context or tools, but only by model intelligence itself.”

The Ramp Inspect Difference

Bruggeman said that as a cloud-hosted tool, Inspect is different from local coding agents because of its accessibility.

“It’s fully hosted in the cloud and requires no local developer setup,” he told The New Stack. The tool spins up a complete virtual machine with the development environment ready, removing barriers for non-engineers,” he noted.

Inspect is powered by OpenCode, Modal and Cloudflare. It runs fully in the cloud, and starts in seconds, letting every builder work at the speed of thought, no setup required.

Moreover, Ramp has open sourced the full blueprint for Inspect so anybody can build their own Inspect. “Just give our spec to your current coding agent, and let it build your new favorite,” Bruggeman wrote in a LinkedIn post.

Bruggeman said, “The legacy players are trying to bolt AI onto 20-year-old architectures. We’re building it in from day one.” He acknowledged that QuickBooks and NetSuite have developer tools but suggested they’re not AI-native.

Bruggeman said Ramp designed Inspect for product managers, designers and other non-engineering builders who create specifications but previously couldn’t implement them.

“The only thing stopping them is this 12 pixel tall wall of code in the editor,” he said.

Use Cases

However, regarding use cases, Bruggeman said product managers now use Inspect during QA sessions to make changes in real time rather than writing tickets. And designers build working prototypes by collaborating with engineers through the tool’s multiplayer features, he said.

The approach reflects a broader trend in enterprise software. “There’s a lot of movement in this space right now,” Bruggeman acknowledged. But Ramp wanted end-to-end control over its tooling.

“The only way you’re going to ensure that it is the best, most productive, most efficient for your organization is to build it yourself,” he said.

Real-World Impact

Inspect has already made a real-world impact at Ramp.

In the past week, 30% of merged pull requests at Ramp were written by the AI agent. Bruggeman framed this as productivity gains rather than replacement. “These are all changes that usually would end up in a backlog,” he said.

The company maintains guardrails around code quality. “We’re not naive to the fact that AI is still not perfect at writing code,” Bruggeman said. All code is reviewed by humans before merging.

“Look, everyone’s building a coding agent right now. We get that. But generic tools don’t understand the difference between a debit and a credit,” Bruggeman said. “You can’t just throw GPT-4 at financial code and hope it works.”

Broader Strategy

In addition, the Inspect coding agent launch fits into Ramp’s broader AI strategy, said company spokesman Alex Rafter. Over the last year, the company has rolled out AI agents for expense policy enforcement, accounts payable automation and invoice processing. Ramp also released Ramp Sheets, an AI-powered spreadsheet tool, through its Ramp Labs innovation initiative.

“This is very much a theme for Ramp as a company, where we really want to push on the forefront of AI,” Bruggeman said. The focus extends beyond customer-facing products to internal tooling as well. “We want to give our builders the best AI tooling possible.”