The day Clay shipped an API, it broke its own pricing

For years, Clay sold you on the idea that it was not an API. It was the place your go-to-market work lived. The table, the workflow, the orchestration layer where you wired up 150 providers and let the waterfalls run. The whole pitch was that you did not want to stitch enrichment APIs together yourself. You wanted Clay to be the connective tissue, and you paid Clay handsomely to be it.

Then the coding agents got good. Everyone moved their orchestration into Claude and Cursor. And Clay, the company whose entire reason to exist was that it was not something you called from the outside, shipped the Clay MCP so you could call it from the outside.

I want to be careful here, because this is not a gotcha. Clay building an MCP is the right product move. Reps live in ChatGPT and Claude now, so meeting them there is smart. The problem is not the MCP. The problem is what the MCP quietly reveals about the pricing sitting underneath it. And almost nobody has done that math out loud.

I did. It starts with one number.

Clay runs two separate meters on every job. Data Credits buy the actual data from Clay's marketplace of providers, starting at $0.05 each on the Launch plan and getting cheaper only if you spend your way into bigger bundles. Actions are the second meter: every enrichment costs one Action on top of the credits, and you pay the Action even when you bring your own API key.

Clay is upfront about why. In their own documentation, the Action covers routing your request, calling the provider, running your workflow, and returning the result to your table. So even when Clay sources none of the data, it still charges for the act of moving that data into a Clay table.

On Clay, a profile is not one thing you buy. It is seven things you buy. Each field is its own enrichment, each runs its own waterfall across multiple providers, and each drags an Action behind it. Here are the real per-field credit costs, read straight from Clay's enrichment panel inside the product.

$1.30 against $0.02. That is 65 times more expensive on the data alone, and closer to 68 times once you count Clay's Actions. Call it 66x and you are being fair to both sides. It is not comparing a fat profile to a thin one. Both return the same core LinkedIn profile. The difference is purely architectural. One platform decided a profile is seven metered events plus seven Actions. The other decided a profile is one call. The 66x is not a discount you negotiate. It is the cost of a pricing philosophy.

A single profile is a debating point. Volume is where it becomes a line item your finance person circles in red.

There is exactly one place these two platforms come close, and it is worth being honest about it because it teaches you how to read everything else. Phone numbers. Clay's Mobile Phone enrichment is 13.4 credits, the single most expensive data point they sell, cascading across 13 providers. RichAPI's Phone Finder is 25 credits running the same kind of deep waterfall. On phone alone, the two are genuinely in the same neighborhood.

Clay MCP and RichAPI MCP look like the same kind of thing on the surface. Both let you drive enrichment from inside Claude, Cursor, or ChatGPT without touching a dashboard. Underneath, they are built on opposite economic assumptions, and that is the whole story.

When you run Clay through its MCP from inside Claude, the orchestration is no longer happening in Clay. Claude is deciding what to enrich, in what order, under what conditions. The thinking moved out of Clay and into the agent. But the meter did not move an inch. Clay still charges the full stack, credits and Actions, because the data still lands in a Clay table through Clay's marketplace.

RichAPI MCP starts from the other end. It assumes the orchestration is already yours, living in your agent or your code, because in 2026 it usually is. So it does not charge an orchestration meter, because there is no orchestration to sell you. It exposes the same endpoints an engineer would hit directly. Profile Enrichment for a credit, Email Finder for five, People Search for a tenth of a credit per result. One call, one price, charged only when you get a hit.

Step back from the two logos and this stops being a pricing spat and becomes a question about the shape of the entire GTM stack. For a decade the bet was that the platform was the valuable part, and the data was just fuel you poured in. Clay built a category and a multibillion-dollar valuation on that bet, and for the era of humans clicking around a UI, it was correct.

RichAPI is a straight bet on Era 2. Enrichment as an API, priced like an API, reachable from an MCP when you want the convenience and from raw HTTP when you do not. I am not going to tell you Clay is bad. It is a genuinely great product, and the numbers above do not measure everything. Data coverage and match rates are real, and you should test them on your own list before you move a dollar. For a big RevOps team with a dedicated Clay engineer, that platform fee still buys something. For a two-person team wiring enrichment into an agent, it buys almost nothing you are actually using.

So here is what I would do. Pull your own numbers. Enrich the same 100 profiles on Clay MCP and on RichAPI MCP. Put the two invoices side by side. Then decide which one is priced for the way you actually work now. The screenshots are real. The math is above. Sixty-six to one is where it starts.

Clay figures taken from Clay's live product and pricing pages, July 2026. Per-field credit costs read directly from the enrichment panel. RichAPI figures from the published endpoint catalog and credit bundles. Clay Launch credit rate $0.05; RichAPI entry bundle $0.020, falling to $0.015 at 100k. Actions priced at Clay's stated sub-cent rate.

RichAPI — the enrichment layer without the platform tax.