Home/Features
Prox AI

Ask your business a question and get the answer

Prox sits on every page with your real data behind it — the clients, tickets, projects, hours and invoices — and answers in plain English, or does the thing, with your permissions enforced on every request.

Start free trialSee all features
app.aspirepro.io
Prox, the AspirePro AI assistant, answering a question about a ticket from the ticket page
On this page

Most AI assistants are a chat window bolted onto a product: they can talk about your data only if you paste it in. Prox is the other kind. It runs inside AspirePro with hundreds of actions over the real records, so “which clients give us the poorest return?” is a question it can actually answer — and “create the follow-up ticket” is an action it can actually take, after you say yes.

Your data

Answers, not guesses

Prox reads the live records — deals, tickets, hours, return on time — rather than a summary someone pasted in.

Your permissions

Checked on every request

Prox can only reach what your role allows. Asking it nicely does not widen your access.

Your call

It asks before it changes anything

Anything that changes data is shown to you first. Nothing is created, edited or sent behind your back.

Context

It already knows what page you're on

Open Prox from a ticket and it already has that ticket. Open it from a company and it has the client — their agreements, open work, WIP and health. You ask the short question a colleague would understand, not a prompt with the context pasted in.

  • "What is this ticket about and what should I do next?"
  • "How is this client doing right now?"
  • "Summarise where this project stands for the status call."
app.aspirepro.io
The Prox panel open on a company record, answering how the client is doing with real figures
Actions

It can do the work, once you say yes

Prox is not read-only. It can create a ticket, log time, move a deal, assign a task, draft a reply. Every action that changes something is presented for confirmation before it runs, and it is recorded in the audit log under your name — because it was your decision.

  • Create, update and assign across the modules you have access to
  • Draft replies and summaries you edit before they go anywhere
  • Confirmation step on every write, and a full audit trail after
The Prox assistant panel open over a dashboard, mid-conversation
It can do the work, once you say yes
Across modules

Questions that cross the whole business

The questions that matter rarely live in a single module. Which clients give the lowest return on your team's time. Who is overloaded next month and still gets more tickets than anyone. What did we promise in that meeting and did it become a task. Prox can reach all of it at once because it is one system underneath.

  • Cross-module questions answered without exporting anything
  • Cites the records it used, so you can check the working
  • Popout the conversation into its own window when you are working through something
app.aspirepro.io
The Prox popout window showing a ranked answer about the clients with the lowest return on time, with basis notes
Your own AI

The same tools, for the assistant you already use

Everything Prox can do is available to Claude, ChatGPT or any other assistant that speaks the open Model Context Protocol (MCP). Add AspirePro's one connector address, sign in and pick an organization, and it gets the same tools, the same permission checks and the same audit trail — you are not choosing between our AI and yours.

  • One connector address for every organization, with OAuth sign-in and no tokens to paste
  • The same permission checks apply to external clients
  • Every tool Prox gains becomes available there automatically
app.aspirepro.io
The MCP connector settings showing the connector URL
How it works

How Prox stays trustworthy

1

It runs as you

Every tool call is executed with your permissions. There is no service account with wider reach behind it.

2

It shows its basis

Where a number is approximated rather than measured, it says so instead of presenting an estimate as fact.

3

It asks before writing

Any change is confirmed by you first — nothing is created, edited or sent without an explicit yes.

4

It leaves a trail

Actions land in the audit log attributed to you, so an AI-assisted change is as reviewable as any other.

Connected

What Prox AI Assistant connects to

Nothing here is a standalone tool — each of these reads the same clients, people and permissions.

Service DeskSummarize a ticket, draft the reply, or find every open ticket for a client.Billing & InvoicingAsk which clients are not worth your team's time, and why, straight from the profitability engine.Client ManagementAsk for the state of a client — work, money, health — in one answer.IntegrationsThe same tool surface is available to external assistants over MCP.Dashboards & ReportingWhen the question becomes a standing one, turn it into a dashboard widget instead.
Questions

Frequently asked questions

What can Prox actually do?
It reads and acts across every module you have access to: finding and summarising records, answering questions that span modules, drafting replies and summaries, and creating or updating work — tickets, tasks, time entries, deals and more. It is a working assistant inside the app rather than a chat window that can only talk about what you paste in.
Can Prox see data I am not allowed to see?
No. Every request runs under your own permissions, so Prox reaches exactly what you would reach by navigating the app yourself. Modules that are switched off, and records your role excludes, are simply not available to it.
Will it change things without asking?
No. Anything that writes to your data is presented for confirmation first, and once it runs it is recorded in the audit log attributed to you.
Does it know what I am looking at?
Yes — Prox is page-context aware. Opened from a ticket, a company or a project, it already has that record, so you can ask a short question instead of describing the context.
Can I use my own AI assistant instead?
Yes. AspirePro has one connector address for the open Model Context Protocol, covering every organization you belong to. Add it to Claude, ChatGPT or any assistant that supports it, sign in and pick an organization, with no tokens to paste, and it can use the same tools. The same permission checks and audit trail apply.
Can we switch AI off?
Yes. Under System → Features → AI an administrator has two organization-wide switches, both on to start with. AI features in AspirePro covers Prox, ticket AI, AI meeting notes and the AI steps in automations: turn it off and nothing is sent to our AI provider and the AI buttons disappear. Connect AI clients (MCP) controls whether people can connect their own AI app: turn it off and connected apps are refused on their next request. The two are independent, nothing is deleted, and each change is recorded in the audit log.
Is my data used to train a model?
No. Prox sends the context needed to answer your question to our model provider, xAI, for that request only. xAI doesn't train on your organization's data and, with Zero Data Retention on, keeps nothing we send it: each request is processed and discarded. The security page sets out the details of how data is handled.
Documentation

Read the documentation

The full how-to for this module, written for the people who use it every day.

Bring your whole services business together

Stop stitching together five tools. Run sales, delivery, support, and billing from one organization — with an AI assistant on every page.

Start free trialSee pricing