Connect your business to Claude and ChatGPT
Also called: Custom MCP connector
Service 5 · A2A readiness
The next step after connectors is AI agents talking to each other: a buyer's agent asking a supplier's agent for stock, prices and delivery dates, and placing an order within limits. It's early, and we'll say so plainly. But the groundwork is the same work that makes your business easier to deal with today, so we do it in stages, and each stage is useful on its own.
Person: Can you supply 40 units of the 2-litre jug by 14 March, at trade prices?
Assistant: Yes, 40 are available at the trade price list. The order comes to £1,280 and can ship on 10 March.
Person: Please place it.
Assistant: Orders over £1,000 from a new buyer need a person's approval, so I've sent it to the sales team. You'll hear back today.
Made-up businesses. Your rules decide what an agent may agree to alone.

Agent2Agent (A2A), an open standard for agents to find each other and work together, is now run by the Linux Foundation, with many large software companies behind it.
Clear, machine-readable facts and well-defined actions help search engines, assistants and your own team today, whether or not agents arrive at your door this year.
Before any agent acts for you, you need to decide what it may agree to, what needs a person and how you'll know who's on the other side. That thinking takes time.
You can stop after any stage. Each one leaves you better placed than before.
Products, prices, stock, terms and delivery in a structured form, from one source of truth, kept up to date.
A short, clear list: check stock, get a quote, place an order, book a slot. Each one defined, with limits.
Spending limits, which customers get which terms, and what always goes to a person. Written down and built into the system.
A way to check the other agent really acts for the business it claims to, and a record of what was agreed.
One real partner, one real flow, watched closely. Only when the first four stages are in place.
Who you buy from and sell to, and where agents could save time on either side.
Where you stand on each stage, what's worth doing now and what can wait.
We build the stages you choose, each tested and useful on its own.
When you and a partner are both ready, we connect one flow and watch it closely.
An hour with whoever knows your trade customers and terms best, and access to the system that holds your products and prices.
Agent2Agent (A2A) is an open standard for AI agents to find each other and work together, now run by the Linux Foundation. In practice it means a buyer's agent could ask a supplier's agent for stock, prices and delivery dates, and place an order within limits the supplier has set. It's early, and most businesses aren't doing it yet. Kwessa prepares businesses in stages: facts an agent can read, the actions it's allowed to take, your rules for what needs a person, a way to check who is on the other side, and then a first live link with a real partner.
A few large companies are, and the tools are maturing quickly. For most businesses it's early, which is why we start with the stages that are useful today.
A connector lets a person's assistant use your business. Agent-to-agent is your business's own agent dealing with another business's agent, usually for trade orders, quotes and bookings.
Your rules stop an agent agreeing to something bad. Limits are built into the system, not left to the agent's judgement, and anything outside them goes to a person.
The readiness review is a fixed piece of work, quoted after a short chat. Each stage after that is quoted on its own.
Tell us what you buy or sell, and who you trade with.
A real person reads every note. There's no obligation, and we'll tell you honestly if we're not the right people for it.
Also called: Custom MCP connector
Also called: Agent integration
Also called: MCP security review