Q Bridge connects the AI agents your teams already use to TraiQ's treasury engine. The tools and calculations behind Q become available inside Claude, OpenAI's Codex, GitHub Copilot, Gemini, Cursor and any agent built on the open Model Context Protocol — so TraiQ fits into the way your organisation already works, not the other way around.
Treasury work doesn't happen in one application. It runs through the TMS, the ledger, spreadsheets and committee papers — and, more and more, through the AI assistants and automations your organisation has already approved. Q Bridge lets those tools draw on TraiQ directly, instead of someone copying numbers between windows.
Built into every TraiQ install. Ask the book anything, run scenarios and draft the committee pack — with the dashboards right beside you.
For the agents and automations your organisation already runs. They call TraiQ's tools directly and get back figures they can cite.
A connected agent gets TraiQ's read-only toolset for the book it's pointed at — banking or corporate. Every answer comes from the same engines that drive the dashboards and Q, so the figure your agent receives is the figure on the screen.
Positions, exposures, limits, counterparties and maturity profiles — filtered and aggregated the way the question needs.
Break a ratio into its drivers, rank limits and covenants by headroom, see what changed between any two dates, and trace every figure to its deals.
The full stress packs — IRRBB, liquidity, FX and reverse stress on a bank book; covenant, cash trough, funding cost and hedge ratio on a corporate one.
Send figures from your TMS, ledger or a draft return. TraiQ sends back only the breaks beyond tolerance, traced to the deals — and keeps nothing it was sent.
Ask whether the book is fit to quote: how fresh it is, what the integrity checks found, and anything held out of the totals.
Run any of your saved report definitions fresh and get the figures back as data, ready for the agent's own analysis.
Reports as Excel, PowerPoint or PDF, decks with native editable charts, a stress test's full workbook — by a single-use link that expires in minutes.
Every Q brief is available as a ready-made prompt, so your agent can produce the daily or committee brief in its own workspace.
Agents never change anything in TraiQ — every call is read-only, on every book.
Four ways a treasury team puts TraiQ to work inside the tools it already uses.
A scheduled agent sends last night's TMS extract to TraiQ and surfaces only what doesn't tie.
An analyst's agent pulls the figures and the deck from TraiQ, then writes the commentary in your house style.
Before a return is filed, an agent compares its figures with TraiQ's, metric by metric.
The first questions of the morning, answered in the agent your team already has open.
An agent is only as trustworthy as the data it cites. Every result Q Bridge returns is stamped with where it came from — so a number that ends up in an email, a memo or a committee paper can always be traced back to the book that produced it.
| Deal | TMS | TraiQ | Break |
|---|---|---|---|
| MM-0887 | A$412.0m | A$418.6m | A$6.6m |
| FXF-2214 | US$38.5m | US$36.5m | US$2.0m |
| IRS-0412 | A$150.0m | A$125.0m | A$25.0m |
Agents can compute and fetch, never change. Anything that would write is refused inside every call — not merely left off the menu.
If the data is stale, failed validation, or the date asked for isn't held, the agent is told why — never handed an answer from something else.
Commentary an agent adds to a TraiQ file prints under its own label, and every figure in it is checked against TraiQ's numbers.
Each analyst issues their own key per device. Keys expire after 90 days, can be revoked at any time, and stop the moment an account is disabled.
Every call is logged — which tool, which agent, how long it took and, over your network, whose key — without ever recording the question or the data.
Q Bridge ships switched off. Each deployment decides whether to enable it, and which families of agent it admits.
Q Bridge follows the same rule as the rest of TraiQ: your book stays on your infrastructure, and you decide which AI sees what.
On your premises, TraiQ never opens a public entry point for agents. Your analysts' agents reach it over your own network, each call carrying that analyst's key.
Results go to the AI vendor your organisation chose, under your organisation's own account and its data terms. TraiQ never sees the conversation.
Q Bridge comes with lock-down guidance for each supported agent — its own web and shell access switched off in TraiQ sessions — so data only moves where you've approved.
Q Bridge speaks the Model Context Protocol, so it isn't tied to any one AI vendor. Your deployment decides which agents are admitted.
| Agent | Status | How it connects |
|---|---|---|
| Claude Code · Anthropic | Supported | On the analyst's machine, or over your network with a personal key |
| Codex · OpenAI | Supported | On the analyst's machine, or over your network with a personal key |
| Claude Desktop | Supported | On the analyst's machine, or over your network with a personal key |
| Cursor | Supported | On the analyst's machine, or over your network with a personal key |
| VS Code · GitHub Copilot | Supported | On the analyst's machine, or over your network with a personal key |
| Gemini CLI · Google | Supported | On the analyst's machine, or over your network with a personal key |
| Other MCP agents | Admitted by you | Any MCP-capable agent can be admitted by your deployment |
Q Bridge was designed to pass a security review, not to work around one.
We'll walk through how Q Bridge would connect to the agents and systems your team already uses.