Treasury teams don't struggle because they lack data. They struggle because that data is fragmented, heavy, and constantly shifting. Hundreds of accounts, banks, entities, forecasts, and risks spread across the globe. The answers are always there. Getting to them just takes too long.
That changes today.
We're launching Pulse, Palm's treasury AI analyst that lets you query your treasury data in plain language, get actionable answers in seconds, and stay ahead of risks before they surface.
Not a chatbot. A decision interface.
Every TMS vendor is rushing to add conversational AI. We took a fundamentally different approach, and we've been at it longer than most realise. Our first prototype dates back over a year, when LLMs weren't yet capable of the reasoning treasury demands. We constantly iterated behind the scenes, until both the scale of our data infrastructure and the increased capabilities of LLMs perfectly lined up.
Pulse reasons across all your Palm data simultaneously: cash positions, forecasts, investments, debt, FX exposures, and intercompany flows. It's powered by frontier AI models hosted in our cloud environment with row-level security enforced per customer at the infrastructure level. Every query runs against your data and only your data, and only read operations are permitted, enforced at the infrastructure layer.
Ask where your opportunities are for more efficient cash balances, and it doesn't just surface numbers. It provides rationale, identifies specific entities, and quantifies the opportunity with actionable recommendations, connecting the dots across your entire operation. Ask for a breakdown and get a rendered chart, built from your live data, right inside the conversation. Need to investigate further? Upload a document like a bank fee schedule or pricing agreement, and Pulse cross-references it against your actual transaction data to surface what matters.
That depth is what sets it apart.
Explainable by design
In treasury, a wrong number leads to a wrong decision. That's why we built Pulse around a schema-first workflow: before answering any question, the AI retrieves the actual structure of your data (table names, column definitions, relationships) and then executes queries against your real records. It cannot invent account balances or fabricate transaction totals. Every number in every response comes directly from your data warehouse.
We didn't stop at architecture. An automated evaluation system reviews every Pulse conversation daily, grading each response for factual accuracy (does every number match the retrieved data?), data retrieval quality, and correct tool usage. Each response receives a pass, warning, or fail score. Failures trigger investigation and feed a learning process that improves Pulse's accuracy over time, on a per-customer basis. The result is an AI you can verify, not just trust.
Proactive, not just reactive
Most AI features in treasury software wait for you to ask. Pulse does not.
We rebuilt the Palm homepage from the ground up around a personalisable daily digest: an AI-generated briefing that greets you every morning. It highlights notable transactions, flags confirmed forecasts needing review, and surfaces accounts with significant variance or low balances, zooming into what happened the past week. Each section includes a narrative summary so you get context, not just numbers, and you can ask follow-up questions in-line without leaving the page.
Beyond the digest, Pulse AI agents continuously work across your data, analysing patterns in cash positioning, flagging anomalies, and surfacing insights that would take hours of manual investigation. You focus on decisions, not data validation.
One of our customers described the experience well: the platform automates the data validation and analysis, then leaves the decisions with the treasury team. That's exactly the design principle behind Pulse.
Built to meet you where you work
Pulse lives inside the Palm platform today, but we engineered it to scale across every surface where treasury teams operate.
The same intelligence powers our MCP server and Slack integration. Whether your CFO wants a quick cash summary in Slack or your team queries treasury data through their existing AI tools, Pulse delivers consistent, context-aware answers. One infrastructure. Multiple interfaces. No fragmentation.
What this means for your team
Pulse eliminates the manual work that consumes treasury teams every week: pulling reports, reconciling data across systems, building ad-hoc analyses, and chasing down answers across multiple dashboards. Ask a question. Get an answer with the chart to back it up. Upload a document and get the analysis. Take action. That's the workflow now.
We're proud of what the engineering team has built, and this is just the beginning. Pulse will continue to evolve with deeper reasoning, broader agent capabilities, and more proactive intelligence to keep your treasury running at peak performance. It builds up context on your preferences and what matters to you as a user, aiming to create a truly personalised AI experience.
Check your Pulse. Your treasury will thank you.
