Product / Answers from your systems

Answers from your systems

Replies written from your own data

The customer asks. Messago asks your system. The reply comes back composed from live values — their policy, their order, their appointment — and hands off when it cannot answer.

messago.xylm.ai / flows / premium-due / step 3 · API call
1

You name the fields you want out of the response. Those names are usable in every later step.

2

The failure path is part of the step, not an afterthought — it hands the chat to a named team.

3

The reply is assembled from the values that came back, and previewed before it is published.

How it works

01

Point it at your endpoint

Give the step a URL, a method and whatever authentication your system expects. Credentials are stored encrypted, not typed into a message.

02

Send what you know

The customer’s phone number is already there. Anything the conversation collected — an order number, a policy number — can go with the request.

03

Name the fields you want

Map values out of the response and give them names. Those names are available to every step after this one.

04

Write the reply around them

Compose the message using those names, add buttons, and preview it against a real test response before publishing.

What it does

Reads orders, billing or bookings

Anything with an HTTP endpoint: an order system, a billing ledger, a clinic’s appointment book, a dealer portal.

A different reply per customer

The same question produces a different answer for each person, because the answer is built from their record rather than picked from a list.

Values persist through the chat

Once fetched, a value can be used in later messages, passed into another API call, or shown to the agent in the contact panel.

Missing data hands off

A failed call, a timeout or an empty result routes to a person with the conversation and everything collected so far attached — it does not stall or guess.

Nothing is stored that you did not map

Only the fields you name are kept against the conversation. The rest of the response is used and dropped.

Also here

AI answering, on your own key

For the questions that are not a lookup — what does this plan cover, do you deliver to my pin code — connect a model and give it your own material.

Your account, your key

Connect OpenAI, Anthropic or Google with your own API key. It is stored encrypted and the usage is billed to you by them, not marked up by us.

Give it your material

Policies, plans, delivery rules, the wording you want used — added as contexts the model answers from.

It hands off when it cannot answer

We will not claim it never invents anything. It answers from the material you give it, and anything outside that goes to a person — which is the part that matters.

What the customer sees

Their own numbers, in the first reply, without giving a policy number or waiting for office hours. The second customer asking the same question gets a different answer.

Figures illustrative — values come from your system.

Talk to a person

Tell us what your team is dealing with.

We will walk your own use case through the product on a call. Pick a plan and your workspace is live as soon as payment clears, or book a demo and we will pick it up from there.