Reynforce and MCP gateways: keep your gateway, add Reynforce
An MCP gateway decides which tools your AI agents can use. Reynforce decides what those agents may promise your customers, and keeps a record of every promise they make.
The problem a gateway can't see
Nothing went wrong with access. The agent was allowed to send email. What went wrong is what it said.
That is the gap. A gateway looks at who is calling and which tool. It does not read the sentence going to your customer, and it has nothing to check that sentence against.
What Reynforce does
Reynforce gives your agents three things.
- A list of what's true about your company. Prices, discounts, what each plan includes, what you're certified for. Every line has a named owner and the date someone last checked it.
- Rules for each agent. What it may promise on its own, what needs a person first, and what it may never say.
- A record of every promise. What was said, to whom, what it relied on, and whether it was allowed. Kept so nobody can edit it.
Before your agent promises a price, a date or a feature, it asks Reynforce. Reynforce answers yes, no, or "ask a person first".
How the two work together
Reynforce is an MCP server. Your gateway serves it like any other.
- Add your Reynforce workspace to the gateway as one more MCP server.
- Give each agent its own key.
- The gateway keeps deciding what each agent can reach. Reynforce checks what each agent says to buyers.
Reynforce never sits between your agent and your other tools, and it never runs your agent. Each product does one job and keeps its own log.
| MCP gateway | Reynforce | |
|---|---|---|
| The question | Can this agent use this tool? | Can this agent promise this to a buyer? |
| What it checks | Who is calling, which tool, which login | What your company says is true, and who owns it |
| Who usually buys it | Security, IT and platform teams | The head of sales or RevOps |
| What you have afterwards | A log of tool calls | A record of every promise made in your name |
Our view
We think every company running agents at scale will want both. A gateway is the right way to control access. But most of the risk in sales and support is not access. It's an agent saying yes to something nobody agreed.
Runlayer is a good example of a gateway done well. It calls itself an "AI control plane and enablement layer for AI agents", and lets you set what agents can access "under which identity, budget, OAuth grant, and runtime conditions". It's built for "platform, security, IT, and AI enablement teams". That's the right job, for the right people. It just isn't the job of deciding what an agent tells a buyer.
The short version: the gateway asks "may it call this?" Reynforce asks "may it say this?" You need an answer to both.
What it costs
- Do it yourself: £99 a month per agent. You set it up; a second agent is a second seat.
- Enterprise: ten or more agents. We agree a price with you.
- The 7-Day Agent Approval: done for you, one-off — Complete £7,000 · Standard £4,500 · Starter £2,500. Approved in 30 days, or you don't pay — as long as you complete your part of the setup.
14-day free trial. No card is needed to start, and nothing is charged unless you add one.
Sources
- Runlayer home page, read 20 October 2026
- Runlayer, written for AI agents, read 20 October 2026
- Runlayer catalog documentation, read 20 October 2026