Ramp

Sierra

Ramp

Ramp makes money from interchange fees — the 1.5-2.5% that merchants pay on every credit card transaction. Unlike consumer cards that share interchange with users through rewards, Ramp gives a flat 1.5% cashback and keeps the rest.

The real business model is becoming the financial operating system for companies: once a company uses Ramp's card, they also use Ramp for expense management, bill pay, accounting automation, and procurement — all of which increase switching costs and customer lifetime value.

Sierra

Companies pay Sierra to run their customer service with AI agents instead of hiring rooms full of humans. The agent talks to your customers over chat.

It answers questions. It processes returns.

It updates subscriptions. Here is the clever part.

Sierra charges based on outcomes, not seats. Brands pay when the AI actually resolves a problem, not just for having the software installed.

That pricing is unusual. Most software charges a flat monthly fee no matter what.

Sierra only gets paid when the bot does the job.

Ramp

Eric Glyman and Karim Atiyeh had previously co-founded Paribus, a tool that automatically got refunds when prices dropped on things you'd already bought. Capital One acquired Paribus in 2016.

The experience taught them something: businesses were terrible at managing their spending, and the tools they used — corporate credit cards from Amex and Chase — were designed to encourage spending, not control it.

In 2019, they launched Ramp with a contrarian premise. Every other corporate card company made money by getting businesses to spend more (higher spend = more interchange revenue).

Ramp would make money from interchange too, but would actively help businesses spend less through automated expense management, duplicate subscription detection, and price negotiation.

The pitch to CFOs was irresistible: get a corporate card with 1.5% cashback, and we'll also find you an average of 5% savings on your total spending through our software. The card was the wedge.

The expense management platform was the real product.

Sierra

Bret Taylor had already done the impossible twice. He co-created Google Maps.

He later became co-CEO of Salesforce, chaired Twitter's board during the messy Elon Musk takeover, and became chairman of OpenAI in late 2023. Clay Bavor spent 18 years at Google, most recently running its Labs group.

The two friends left their big jobs in 2023 to build something together. They picked customer service on purpose.

It is the least glamorous corner of software. Their logic was simple.

Companies hate their support costs. Customers hate waiting on hold.

AI that actually resolves problems could fix both sides at once. They launched Sierra in February 2024 with $110 million from Sequoia and Benchmark.

Ramp

Ramp grew by selling savings, not credit. The pitch to finance teams was: "We'll save you more money than we cost you." In an era when every company was looking to cut costs, Ramp offered a corporate card that came with a free expense management platform that actively found savings.

CFOs couldn't say no.

The product-led approach bypassed traditional enterprise sales cycles. A finance manager could sign up for Ramp, issue cards, and start seeing savings within a week — no six-month procurement process, no IT integration project.

The free expense management tools were so good that companies switched from Concur, Expensify, and Brex just for the software, with the card as a bonus.

Speed of execution was the differentiator. Ramp shipped features faster than any competitor.

They went from a corporate card to a full financial operations platform in three years. Every quarter, Ramp launched features that competitors took a year to build.

By 2024, over 25,000 businesses were using Ramp and the company was processing tens of billions in annualized spend.

Sierra

Sierra went straight for big brands, not tiny startups. SoFi, Ramp, ADT, Sonos and Brex signed up early.

By 2026 it said more than 40% of the Fortune 50 were customers. The trick was trust.

Enterprises are terrified of an AI bot going rogue and insulting a customer. So Sierra built guardrails and let each company control exactly how the agent behaves.

Bret Taylor's contacts did the rest. When the chairman of OpenAI calls a Fortune 500 CEO, they pick up.

Sierra hit $100 million in annual recurring revenue in November 2025, seven quarters after launch, and $150 million by February 2026. In April 2026 it launched Ghostwriter, which builds and deploys a new agent from a plain-English description.

Ramp

Brex is the obvious competitor. Brex launched two years before Ramp with a similar corporate card concept and had the first-mover advantage.

But Brex pivoted away from small businesses to focus on enterprise in 2022 — angering thousands of existing customers — while Ramp doubled down on serving companies of all sizes. The competition has become a case study in strategic focus versus strategic pivots.

The "spend less" positioning has a mathematical ceiling. If Ramp's AI genuinely helps companies spend less, the interchange revenue from those companies also decreases.

There's an inherent tension between the mission (reduce spending) and the revenue model (earn a percentage of spending). Ramp has managed this by growing the customer base faster than individual customer spending declines.

Enterprise sales is the next frontier and it's expensive. Moving upmarket from startups and mid-market companies to Fortune 500 enterprises requires a sales team, implementation support, and enterprise features that cost real money to build and sell.

Ramp has been investing heavily in enterprise capabilities, but competing with Amex and JP Morgan for large corporate accounts is a different game than winning startups.

Sierra

Everyone is building AI customer service agents right now. Decagon has raised at a multi-billion valuation.

Salesforce, ServiceNow and Intercom all shipped their own agents. Sierra is not the cheap option.

It sells premium AI to giant companies that move slowly. The bigger risk sits underneath the product.

Sierra does not make its own AI models. It rents them from providers like OpenAI and Anthropic.

If those model makers decide to compete directly, Sierra's edge gets thin fast, and its co-founder chairing one of them only makes that question louder. A $15.8 billion valuation on roughly $150 million in revenue leaves almost no room for a stumble.

Ramp

Ramp Corporate Card is the core — unlimited physical and virtual cards with 1.5% cashback and built-in spend controls. Ramp Expense Management automates receipt matching, policy enforcement, and reimbursements.

Ramp Bill Pay handles vendor payments and AP automation. Ramp Procurement manages vendor contracts and purchase approvals.

Ramp Intelligence uses AI to identify duplicate subscriptions, negotiate better rates, and flag wasteful spending. Ramp Flex offers flexible payment terms for businesses that need to extend their payables cycle.

Ramp Accounting automates close processes and syncs with QuickBooks, Xero, NetSuite, and Sage.

Sierra

The core product is the Sierra AI agent. It is a branded customer service bot that companies deploy on their website, app, and phone lines.

It handles the boring stuff. Order tracking.

Refunds. Password resets.

Account changes. Sierra also built its own tools to test and monitor these agents so they do not make things up.

In 2024 the company added a voice version that talks to customers on the phone. In April 2026 it added Ghostwriter, a tool that builds and deploys a new agent from a plain-English description.

The whole pitch is one AI agent that carries your brand's name and never sleeps.

Ramp

Founders Fund, D1 Capital, Stripe, Goldman Sachs, Thrive Capital, General Catalyst, Khosla Ventures

Sierra

Sequoia Capital and Benchmark wrote the first checks at launch in February 2024, $110 million between them. Greenoaks Capital then led both the $175 million round in 2024 and the $350 million round in September 2025.

In May 2026 Tiger Global and Google's GV led a $950 million round at a $15.8 billion valuation, with Benchmark, Sequoia and Greenoaks back in. That takes total funding to about $1.6 billion.

That is partly the product and partly Bret Taylor. Investors bet on founders who have already built billion-dollar things.