AI cash flow forecasting and treasury software compared
Most treasury software now mentions AI. The differences are in what sits underneath: how wide the platform is, how it connects to your banks, how the forecast is built, how much the AI may do on its own, and what it costs.
Start with the job you need done. If you need a full treasury management system with payments, hedging and in-house banking, look at the broad platforms first. If your main problem is knowing where your cash is today and where it will be in thirteen weeks, a focused platform can get you there faster and for less.
We build NineAnts, so read this page with that in mind. We have tried to describe every product fairly, including where others are the better choice. Last checked: 28 October 2026. Competitor facts come from each vendor's public website or press coverage (see sources below). Where a vendor does not publish something, we say so.
At a glance
| Criterion | NineAnts | Kyriba | HighRadius | Ripple Treasury (GTreasury) | Trovata | Atlar | Nilus | Panax | Palm | Bond |
|---|---|---|---|---|---|---|---|---|---|---|
| Best for | Mid-market and European groups that want cash visibility and an AI forecast without a full TMS | Large enterprises that need a broad liquidity platform | Mid to large companies, especially with an order-to-cash focus | Corporates that want a full TMS, including digital assets | Teams that want API-based bank data, cash visibility and forecasting, with a TMS option | Multi-bank, multi-entity companies that want cash, payments and forecasting in one platform | Finance teams that want AI agents across treasury, AR and AP | Mid-sized companies with small finance teams that want AI agents across cash operations | Large companies that already have a TMS or ERP and want an AI forecasting layer on top | Enterprises that want AI agents to execute payments, FX and investments within policy |
| Scope | Connectivity, cash visibility, categorisation, forecasting, Treasury AI Agent | Treasury, payments, risk, working capital | Treasury and risk, plus order-to-cash | Full TMS: liquidity, forecasting, netting, payments, in-house bank, risk, reconciliation | Cash, TMS, bank data infrastructure | Cash, payments, forecasting, reconciliation, investments, debt | Nine modules from cash and payments to FX, debt and investments | Cash positioning, categorisation, forecasting, reporting, reconciliation, payments, cash accounting | Categorisation, forecasting, liquidity positioning and reporting on top of your TMS or ERP | Payments, trading, forecasting, reconciliation, risk, liquidity, in-house bank, debt |
| Forecasting approach | Bottom-up, with a model chosen per cash flow category and accuracy tracked against actuals | AI forecasting within the platform | AutoML choosing from 100+ ML models | Cash forecasting module with AI insights | Short and long-term forecasts from history, ERP and projected activity | Forecasting module and Forecasting agent | 13-week forecast from actual transactions | Always-on, multi-scenario forecasting; states 96% accuracy | 25+ AI models on ERP and bank data; states over 60% lower variance | Built from ledger, open AR and AP, payroll and bank actuals per entity; variance measured weekly |
| AI agent | Treasury AI Agent, autonomy set per action type | TAI, on Kyriba's embedded LLM | Agentic AI features | GSmart agents: manual, semi-autonomous or autonomous per workflow | Trovata AI, agents demo library | Ready-made agents, no action without approval | Pre-built agents per module | Core agents for categorisation, reconciliation, forecasting and reporting, plus a no-code Agent Builder | Agents run analysis on a schedule or on demand, with a full trace of every run | Ten agents, from optimisation, FX and payments to forecasting and policy checks; escalates above set mandates |
| Bank connectivity | Bank APIs, SFTP (CAMT, MT940), PSP feeds, ERP connectors | Open API, direct bank file connections, SWIFT | “All major banks”, bank files | 300+ pre-built bank connectors | API-based multi-bank connectivity | 100+ banks, built in-house | 20,000+ bank and PSP connections | 10,000+ bank, ERP and cash platform connections, API-first | Reads from TMS (e.g. Kyriba, FIS, ION), ERP (e.g. SAP) and bank portals | 15,000+ banks via API, SFTP and SWIFT, plus 250+ PSPs |
| Data residency | Microsoft Azure, EEA only, including AI processing | Not stated on the pages we reviewed | Not stated on the pages we reviewed | Not stated on the pages we reviewed | Not stated on the pages we reviewed | Not stated (SOC 2, ISO 27001, GDPR, DORA listed) | Not stated (SOC 2 Type II, GDPR listed) | Not stated on the pages we reviewed | Not stated (SOC 2, GDPR listed) | Not stated (SOC 2 Type II, ISO 27001, GDPR, DORA listed) |
| Published pricing | Yes: from €4,000 a year | Not published | Not published | Not published | Base package $24k a year | Not published | Not published | Not published (quote-based plans) | Not published | Not published |
| Stated time to go live | Days to a few weeks for connectivity | Not published | Under 12 weeks | 90 days for liquidity, forecasting and netting | Not published | Under 90 days | 30 to 60 days on average | Not published | Days to a live forecast | Not published |
The vendors in short
NineAnts
NineAnts is an agentic AI platform for treasury and finance teams, built by Automation Boutique in Utrecht. It connects to banks, PSPs and ERP systems, categorises every transaction with readable rules that AI proposes and people approve, and builds a bottom-up cash flow forecast with a model chosen per category. The Treasury AI Agent looks for excess cash, shortages and FX exposure, and you decide per action type whether it acts, asks first or does nothing. Data and AI processing stay in Microsoft Azure in the EEA. Prices are public: Starter €4,000, Professional €12,000 and Enterprise from €24,000 a year, with forecasting and the agent in Enterprise.
Kyriba
Kyriba is one of the broadest platforms in the market, covering treasury, payments, risk management and working capital. In 2025 it launched TAI, an agentic AI built on its own embedded language model and more than 20 years of liquidity data, without third-party LLM integrations. It suits large, international groups that want one platform for most of the finance function and have the budget and team for a larger implementation.
HighRadius
HighRadius offers cash forecasting as part of its treasury and risk products, aimed at mid to large companies. Its forecasting uses AutoML to pick from more than 100 machine learning models, and it states up to 95% forecast accuracy. It is a strong fit if you also use, or plan to use, its order-to-cash products. Implementation is stated at under 12 weeks.
Ripple Treasury (GTreasury)
GTreasury was acquired by Ripple in 2025 and is now marketed as Ripple Treasury. It is a full treasury management system: liquidity, forecasting, netting, payments, in-house banking, FX and interest rate risk, hedge accounting and reconciliation, plus digital asset accounts. Its GSmart agents propose actions with the relevant policy clause attached, and each workflow can run manually, semi-autonomously or fully autonomously. It fits corporates that want a complete TMS from one vendor.
Trovata
Trovata is built around normalised bank data via APIs. It offers Trovata Cash for visibility and forecasting, a TMS on the same data foundation, and bank connectivity as a service. It is one of the few vendors with a published entry price: a base package at $24k a year. A good fit for teams, often in the US, that want API-first bank data and room to grow into a TMS.
Atlar
Atlar, from Stockholm and now also in New York, combines cash management, payments, forecasting, reconciliation, investments and debt. It builds its own connections to more than 100 banks and offers ready-made agents that take no action without approval. It states that most teams go live in under 90 days. It suits companies that want cash and payments in one modern platform.
Nilus
Nilus covers treasury, accounts receivable and accounts payable in nine modules, each with pre-built AI agents, and lists more than 20,000 bank and PSP connections. It states an average deployment of 30 to 60 days. A good fit for finance teams that want agents across treasury and payables or receivables, not only for treasury.
Panax
Panax, based in Tel Aviv and New York, calls itself an AI-native treasury and cash operations platform. It connects bank and ERP data and runs AI agents for categorisation, reconciliation, forecasting and reporting, with a no-code Agent Builder for custom agents. It lists more than 10,000 bank, ERP and cash platform connections and states 96% forecast accuracy. It has focused on mid-sized companies with small finance teams, and also covers payments and cash accounting.
Palm
Palm, from London, positions itself as an AI intelligence layer on top of an existing TMS or ERP. It categorises transactions, forecasts cash with more than 25 AI models on ERP and bank data, and shows liquidity by entity, currency and pool. Its agents run treasury analysis on a schedule and keep a full trace of each run. It reads data from systems such as Kyriba, FIS, ION and SAP, and says teams get to a live forecast in days. A good fit for large companies that already run a TMS and want stronger forecasting on top.
Bond
Bond, based in London and New York, calls itself an AI execution platform for enterprise finance. Its agents cover optimisation, FX, payments, investments, reconciliation, risk, policy checks, forecasting, reporting and debt, and they execute within mandates you set, escalating anything above them for approval. It lists more than 15,000 banks via API, SFTP and SWIFT, and names SOC 2 Type II and ISO 27001. It suits enterprises that want agents to analyse and also move money, trade FX and place investments.
When NineAnts is not the right fit
- You need a full TMS. NineAnts does not do payment execution, hedge accounting, in-house banking or debt and investment management. Kyriba, Ripple Treasury, Atlar or Bond cover these.
- You are a large global group with hundreds of banks across many regions. The broad platforms have more pre-built connections and more years of global rollout behind them.
- You want treasury, AR and AP in one product. Nilus or HighRadius cover more of the finance workflow.
- You need US payment rails and US-based support first. Several of the vendors above are US-based or US-focused.
We will tell you this ourselves in a first call. We have implemented traditional TMS systems for clients and know when one serves you better.
How to choose: a checklist
- Write down the job. Cash visibility, a 13-week forecast, payments, risk, or all of them?
- List your banks and formats. Ask each vendor how they connect to your specific banks (API, SFTP, SWIFT) and how long each connection takes.
- Ask how the forecast is built. One model for everything, or a method per cash flow category? Can you see why a number is what it is?
- Ask what the AI may do on its own. Can you set autonomy per action? Is every AI action logged?
- Ask where your data lives. Which cloud, which region, and does AI processing stay in that region?
- Ask for a price. Including users, bank connections and implementation.
- Test on your own data. A demo on sample data tells you little. Ask for a proof on a few of your real accounts.
Frequently asked questions
What is AI cash flow forecasting?
It uses your historical transactions and known future cash flows to predict where your cash will be, by category, entity and currency. Good systems choose a method per cash flow category and measure their own accuracy against actuals.
Do I need a treasury management system to use AI forecasting?
No. A full TMS adds payments, risk and in-house banking. If your main need is visibility and forecasting, a focused platform such as NineAnts works on top of your banks and ERP, or next to an existing TMS.
What is a treasury AI agent?
Software that can analyse cash data, choose the right method for a question and recommend or carry out actions within limits you set. Most vendors above now offer one. The differences are in how much autonomy you can configure and how actions are logged.
Which tools publish their prices?
Of the vendors on this page, NineAnts (from €4,000 a year) and Trovata (base package $24k a year) publish prices. The others work with quotes.
Where is my data stored with NineAnts?
In Microsoft Azure data centres inside the European Economic Area, including all AI processing. Customer data is never used to train AI models. Automation Boutique is ISO 27001 certified.
Want to see how NineAnts handles your own banks and cash flow categories?
Book a demoSources
- Kyriba: Kyriba unveils agentic AI TAI (Financial IT), Kyriba enterprise connectivity fact sheet
- HighRadius: HighRadius cash forecasting software
- Ripple Treasury (GTreasury): Ripple acquires GTreasury, Ripple Treasury
- Trovata: Trovata, Trovata pricing
- Atlar: Atlar, Atlar pricing
- Nilus: Nilus
- Panax: Panax, Panax connectivity, Panax Series A (TechCrunch)
- Palm: Palm, Palm AI agents, Palm integrations
- Bond: Bond, Interview with Bond (treasuryXL)