RampandCoinbaseAreBuildingWithAI.YouCanToo—WithoutHiring6Engineers.

Two of the most operationally aggressive companies in fintech just showed everyone what AI adoption actually looks like at scale. The results are staggering — and the implications for every mid-market company are impossible to ignore.

ramp: 6,300% increase in internal AI usage

Ramp, the $32B corporate card and spend management platform, didn't just “adopt AI.” They rewired how the entire company works.

Under CPO Geoff Charles, Ramp went from barely anyone using AI to non-engineers shipping production code. Charles laid out the playbook in a detailed interview — and the highlights are worth paying attention to:

• 50% of Ramp's code is now written by AI, with expectations to reach 80%.
• Non-engineers are shipping production features using AI coding agents — product managers, analysts, and operations staff building real tools.
• They built internal AI agents for customer research that compress 8 days of work into 8 minutes — scanning support tickets, call transcripts, competitor data, and their own codebase simultaneously.
• An internal Claude Code PM skill turns product managers into spec-writing machines: it frames the problem, launches parallel research agents, and outputs clean, evidence-grounded specifications.
• They developed an L0–L3 framework to systematically move every employee from basic AI usage to shipping production-quality work.

The results speak for themselves: Ramp's revenue growth has accelerated four quarters in a row. CEO Eric Glyman told the company they'd become “the most productive company in the world.” By the numbers, they're making a serious case.

coinbase: AI writes the code, AI gets the wallet

Coinbase is taking an equally aggressive stance, but from a different angle. CEO Brian Armstrong set a public target: more than 50% of daily code written at Coinbase should be AI-generated. They've hit that number, and they're pushing further.

But Coinbase isn't stopping at code generation. They're building infrastructure for AI agents to operate autonomously:

• Agentic Wallets — purpose-built infrastructure that lets AI agents hold, spend, earn, and trade money autonomously, with programmable spending limits and compliance-ready screening.
• AgentKit— a developer toolkit for creating “human-backed agents” with identity verification and trust layers.
• x402 Protocol — a machine-to-machine payments protocol (built with Cloudflare) enabling agents to pay for compute, API access, and data without human intervention.

Armstrong's message to the team was unambiguous: adopt AI or move on. Coinbase let go of engineers who refused to integrate AI into their workflows. That's not a suggestion — it's a strategic mandate.

the part nobody talks about

Here's what gets lost in the excitement: Ramp and Coinbase are massive engineering organizations. When Ramp builds an internal AI agent for customer research, or a Claude Code PM skill, or an L0–L3 adoption framework — that's a team of engineers dedicated to building internal tooling. When Coinbase ships Agentic Wallets and a machine-to-machine payments protocol, that's an infrastructure team with deep resources.

Ramp's own article credits multiple engineers directly involved in building these systems. That's not a knock — it's the reality of what it takes to do this well.

Most companies don't have 6 engineers to dedicate to internal AI tooling. They don't have the bandwidth to build an L0–L3 framework, custom coding agents, or autonomous financial infrastructure. But they still need the outcome.

you don't need an internal AI team. you need a build partner.

The gap between “we should be using AI” and “AI is integrated into how we operate” isn't a strategy gap — it's an execution gap. And it's exactly what Rockwell AI exists to close.

We build the same kinds of systems that Ramp and Coinbase are building internally — but for companies that don't have (and don't need) a dedicated AI engineering team:

• Custom AI agents for customer support, operations, compliance, data analysis — scoped to your workflows, connected to your systems.
• Process automation that eliminates the manual work your team does every day — intake, routing, document processing, reporting.
• Cross-platform data layers that let you query your CRM, accounting, project management, and communication tools with natural language instead of tab-switching.
• AI integration strategy — not a slide deck, but an actual implementation plan with a working prototype, built by operators who understand your business constraints.

The difference is the model. Ramp hired internally. Coinbase hired internally. You work with us — a fractional AI team with the same technical depth, but without the headcount, ramp-up time, or overhead of building an internal AI department from scratch.

the window is now

Geoff Charles at Ramp said it plainly: “If you're not using Claude Code, you're probably underperforming.” Brian Armstrong at Coinbase backed it up with action — restructuring teams around AI output.

These aren't early adopters experimenting. These are billion-dollar companies restructuring their operations around AI because the productivity gains are too large to leave on the table.

The question isn't whether your company should be doing this. The question is whether you're going to build the team to do it yourself — or work with one that already exists.

Scope your first AI integration with Rockwell, or book a call to talk through what makes sense for your operation.