Client Pulse pilot rollout has started at the first banks. Client Pulse is TreasurUp’s AI client intelligence for relationship managers, delivered inside the bank’s own business banking channel. Bank relationship managers can monitor all of their clients automatically, prepare for every meeting faster, and keep a full record of each relationship.
This article sets out why the bank risks losing sight of its own clients, how the three signal groups work across a portfolio, what goes into the one-click meeting brief, and the trust principles Client Pulse is built on.
In brief
Client Pulse monitors every client automatically each week and surfaces the right insight at the right time, via a weekly summary and a chat interface. Every signal is measured against the client’s own baseline. Client Pulse prepares; the sales manager always makes the final call.
The bank risks becoming invisible in its own relationships
Two pressures are pulling the daily relationship away from the bank.
Workflows are leaving the bank, one at a time
SMEs quietly route cross-border payments and FX conversion to fintechs, and cash visibility to ERP and accounting tools. Each workflow that leaves is a daily relationship the bank no longer owns.
A third party may become the preferred channel
Company owners increasingly ask AI assistants about their finances. If that assistant becomes the first interface for financial decisions, the bank is mediated, one step further from the owner.
Both pressures point to the same fix: give relationship managers the insight to act, not just react. That is a capacity problem as much as a data problem. Relationship managers cover more clients than they can watch, and pre-call preparation is manual, so it gets skipped or eats the time the call was meant to have.
AI client intelligence: one view of the book, and an agent to ask
The starting screen is the portfolio overview. It shows trading volume, hit ratio and active clients against the previous period, then lists the client priorities for the week, grouped by signal.
Portfolio overview in Client Pulse
Next to the list, the Portfolio Agent writes the weekly briefing. In the example, it names 7 clients ready to automate as your biggest opportunity this week. The priority list says where to begin: Start with Velmora Industries. Six weeks without a trade. A good time to reconnect.
Portfolio-level insights: three signal groups
Client trading behaviour is tracked and automatically categorised into three signal groups: at risk, ready to automate and ready to expand.
Signals catch when a client shifts from their own normal, not a fixed threshold. A built-in agent answers portfolio-level questions and generates weekly summaries.
| Signal group | Purpose | What it spots |
| At risk | Protect the relationship | Spot early when clients start going passive. |
| Ready to automate | Help them work smarter | Recognise when clients are doing manual, repetitive work. |
| Ready to expand | Grow the relationship | Spot growth early and be ready to help clients grow further. |
Client-level insights: a one-click meeting brief
Key trading metrics per client measured against their own historical baseline: volume, trade frequency, hit ratio, tenor, active users.
Pattern changes are detected automatically and explained in plain language.
A one-click meeting brief covers client snapshot, active signals and talking points. Notes, reminders and every generated brief are saved as a relationship history.
Client workspace with the Client Agent meeting brief. Design concept · Reference data.
In the example, the Client Agent turns two active signals into three next steps: reconnect after 6 weeks without a trade, explore automated FX batches and review the gap in hedge cover.
How AI client intelligence differs from CRM data
A CRM records what someone typed in after the fact. Client Pulse reads what the client actually did in the channel: a batch saved but never executed, a hit ratio that started slipping, a regularly traded pair that quietly disappeared.
That is why signals are based on each client’s own baseline: pattern changes, not arbitrary thresholds. Client Pulse complements the CRM rather than replacing it. For the wider case, read how agentic AI gives relationship managers their week back.
What banks get from Client Pulse
- At-risk clients flagged before volume drops
- Automation and expansion opportunities identified from trading behaviour
- Signals based on each client’s own baseline: pattern changes, not arbitrary thresholds
- Consistent quality of preparation across the whole sales manager team
- Client knowledge retained in the platform when people move on
Built for trust
The sales manager always makes the final call, no automated or client-facing actions. Signal logic is explainable, not a black-box model. Client data stays within the bank, never shared across banks.
Client Pulse is the bank-side agent in TreasurUp’s agentic business banking roster. It prepares briefings, ranked priorities and suggested talking points. It does not decide, trade or contact clients. An approval gate sits between every proposal and any action, and the bank owns the agent: branding, data, model choice and deployment topology, with its own LLM across managed SaaS, single-tenant or hybrid deployment.
Where the pilot stands
The pilot rollout has started at the first banks. It starts from behavioural data the bank’s channel already produces, so a first group of relationship managers can work with the weekly priority list and meeting briefs in weeks, not quarters.
The rollout runs in four steps: connect the channel data, baseline the signals per client, pilot with a first group of relationship managers, then extend to the full team. The current build reads FX behaviour, where TreasurUp’s FX solution is live inside bank channels today. The signal model is not FX-specific, and TreasurUp is extending Client Pulse to payments, liquidity and cash flow forecasting.
Frequently asked questions
Does Client Pulse take any action on its own?
No. Client Pulse prepares: it assembles briefings, ranks priorities and suggests what to raise in the next conversation. The sales manager always makes the final call, no automated or client-facing actions. An approval gate sits between every proposal and any action.
What data does Client Pulse look at?
Key trading metrics per client measured against their own historical baseline: volume, trade frequency, hit ratio, tenor, active users. It also reads channel behaviour such as logins, saved-but-unexecuted batches, quote-to-trade ratios and module adoption. Other bank sources, such as trade or client databases, can join through the TreasurUp connection centre over MCP.
Is our client data shared with other banks?
No. Client data stays within the bank, never shared across banks. The bank owns the agent, including its data, model choice and deployment topology, and can bring its own LLM.
Does Client Pulse only cover FX?
Today it runs on FX, where TreasurUp is live inside bank channels and the behavioural data is richest. The signal model itself is not FX-specific, and TreasurUp is extending Client Pulse to payments, liquidity and cash flow forecasting.
How quickly can our relationship managers see this on real portfolio data?
Weeks, not quarters, for a pilot, since Client Pulse starts from behavioural data the channel produces today. Full TreasurUp solutions typically go live in 3 to 9 months.
Want to see Client Pulse on your own client book?
Get in touch to book a one-hour working session: your relationship manager workflow and the Client Pulse signals, running on realistic portfolio scenarios.
Curious where the agent should sit at your bank?
Book a meeting