
Nilus is an AI treasury platform for multi-entity, mid-market finance teams that want a live cash position, a 13-week forecast, bank reconciliation and payment approvals run by pre-built agents across their banks and ERPs.
Independently scored, one dimension at a time. Every dimension is sourced, and we say whether a claim is published, reported, or estimated.
Written by the Audit Friendly research team. No vendor edits, no sponsored placement.
You are buying a cash operations layer with agents built in. Nilus pulls balances and transactions over Swift, host-to-host, APIs and SFTP, tags them, reconciles them to the ERP, and builds a 13-week forecast that corrects itself against actuals. Nine modules are sold separately, from Cash and Liquidity to Debt and Investments. A library of 60 pre-built agents does the routine work, and a person approves anything that moves money. Nilus cites a 30 to 60 day average deployment (homepage, checked 1 November 2026).
The catch is maturity and openness. Pricing is quote only, and no public benchmark exists. The Payments module still produces a CSV file that someone uploads to the bank portal; direct delivery over Swift is listed as near-term roadmap. No public API or MCP server is documented, and the subscription agreement bars automated access without written consent. The published SLA promises 96 percent monthly availability on a reasonable-efforts basis. In May 2026 Nilus itself wrote that it is not the natural fit for a full enterprise TMS with complex FX, debt and hedging work, and none of its eight public case studies covers those modules.
Before signing, get these in writing: the annual fee for each module and a cap on increases after the initial term; which of your banks connect directly over Swift or host-to-host and which run through aggregators such as Plaid or MX; the ship date for direct payment delivery; a data export window after termination, since the standard terms end access to your data on termination; and support response times, which the public SLA does not set.
Jobs, mapped to how finance teams actually buy.
Forget the 40-row feature grid. Pick a finance workflow and see how this tool and its closest peers actually do it, with our verdict on who wins for whom.
Here's what it really costs: published rates where a vendor lists them, aggregated quote data where they don't, plus a path to your own number.
Scored on what an outside agent can actually reach: the documented API, how it authenticates, and whether the vendor ships an MCP server.
Exists: No. Nilus documents no MCP server and no public API, and none appears in the official MCP registry or on GitHub (checked 1 November 2026). API Evangelist's 17 October 2026 profile says Nilus publishes no public developer API, OpenAPI contract or developer portal. Its free Delta academy runs a Claude for Treasury video course, which teaches Claude on its own and adds no Nilus connector.
Scope: Not applicable outside the app. Inside Nilus, agents read bank, ERP and payment data and propose actions, and the agents FAQ says no agent executes a payment, moves cash or takes an irreversible step without explicit human approval.
Turnkey: Not applicable for outside agents. The subscription agreement bars any automated means of accessing the platform without Nilus's prior written consent, so scripted or agent access needs a written carve-out. Inside the app, the Agent Builder creates custom agents from a plain-language description, with no code.
In-product AI: the Agent Fabric, launched 7 June 2026, now lists 60 pre-built agents (24 automation, 27 analysis, 4 reporting, 5 prediction) with a builder, scheduler, workflow manager and report builder (agents page, checked 1 November 2026). The vendor claims forecast accuracy gains of more than 30 percent and auto-match rates of 95 to 99 percent; no independent account confirms them. According to Nilus, deterministic logic runs approvals, sanctions screening and payment files, while LLMs handle messy-memo matching and narratives.
Implementation
The homepage cites a 30 to 60 day average deployment, and a 21 June 2026 post lays out a four-week plan: banks and ERP in week 1, guardrails in week 2, a parallel run in week 3 and the first forecast in week 4. A one-day file-based onboarding covers banks without a direct integration. Taboola's case study lists a 3-month implementation, so plan for longer with many entities.
ERP & data connectivity
Banks connect over Swift, host-to-host, APIs or SFTP, with a one-day file-based fallback, and the homepage cites 20,000+ bank and PSP connections. The subscription agreement names aggregators such as Plaid, Finicity, MX and Salt Edge for some account data. The Made In case study describes a direct NetSuite integration and the agents FAQ names SAP, Oracle and NetSuite (checked 1 November 2026), while Freightos's case study lists a direct NetSuite link only as a plan. Dynamics 365, Sage and Workday appear as logos, and no connector documentation is public.
Compliance depth
The homepage shows SOC 2 Type II, SOC 1, GDPR and SAML SSO badges (checked 1 November 2026). The Payments module documents maker-checker approval with amount thresholds, per-user limits and multi-tier chains, plus fraud, sanctions and duplicate screening, and the Connectivity and Controls page describes an immutable audit trail and segregation of duties. No ISO 27001 certificate, SOC report period or data residency is stated on its public pages.
Modeling depth
The Cash and Liquidity module builds a 13-week rolling forecast from actual transactions that self-corrects against forecast-versus-actual variance, with stress tests for delayed collections, a drawdown or an FX move, plus runway and forecast-by-bank views (checked 1 November 2026). The Debt module projects covenant headroom against the forecast. Longer horizons, P&L and headcount planning are not described on the module pages.
Scalability
Nilus cites 100+ finance teams and 20,000+ bank and PSP connections (homepage, checked 1 November 2026). Its largest published case, Taboola, runs 16 banks, 22 entities and 18 currencies on the platform at $2B+ revenue. The base is far smaller than Kyriba's 4,000 customers, and the SLA commits to only 96 percent monthly availability, so uptime and support are what to test first as volume grows.
AI / agent readiness
Inside the product, the Agent Fabric (launched 7 June 2026) offers 60 pre-built agents, a plain-language Agent Builder, a scheduler and a workflow manager, and nothing that moves money runs without human approval. Outside the product there is nothing to connect to: no public API, developer docs or MCP server are documented, API Evangelist's 17 October 2026 profile rates it human-only, and the subscription agreement bars automated access without written consent.
Support
Nilus publishes no support hours, channels or response times. Its SLA sets 96 percent monthly availability as a reasonable-efforts target, applies only 90 days after provisioning, and pays credits on the next annual renewal invoice. Case studies describe a hands-on implementation team, the careers page lists a forward deployed engineer role, and the free Delta academy (announced 28 January 2026) trains finance staff on agentic AI.
Pricing transparency
There is no pricing page, and Nilus's own 26 May 2026 buyer's guide lists its pricing as custom, set by entities, accounts, integrations, workflow scope and implementation. A 2 June 2026 Nilus guide puts modern agentic platforms at $35,000 to $90,000 a year, a category range with no Nilus figure. The subscription agreement lets Nilus change fees after the initial term with notice, with termination as the only remedy.
Nilus pulls balances and transactions over Swift, host-to-host, APIs or SFTP, tags each transaction by cash-flow category, and rolls accounts up by entity and currency. The daily grid flags days that fall below a target buffer. Taboola's case study reports 95 percent of transactions auto-tagged (vendor case study).
The forecast projects inflows and outflows by category and adjusts against forecast-versus-actual variance as data lands. Scenarios stress delayed collections, a drawdown or an FX move. The vendor claims accuracy gains of more than 30 percent; no independent source confirms it.
Nilus matches expected items from the ERP, payroll and uploads against bank transactions, explains each suggested match for approval, and posts journal entries back to the ERP. About 95 percent of transactions are categorized automatically, according to Nilus. Made In's case study cites an 85 percent cut in reconciliation time.
Payments are created from expected transactions, POs or invoices and routed through amount thresholds, per-user limits and multi-tier approvals. Today the output is a bank-ready CSV file for ACH, wire, SEPA or RTP that someone uploads to the bank portal. Direct delivery over Swift is listed as near-term roadmap.
The Liquidity Agent, launched 31 March 2026, tracks entity balance floors, payroll readiness, revolver headroom, FX exposure caps, bank concentration and payment account balances against thresholds the team sets. Breaches come with a suggested next step, and tables and stress tests export to CSV.
AI in Treasury: Real Use Case for Faster Close & Better Forecasting
A summit session featuring customer Flare, which runs 25+ bank accounts across several countries and currencies, on using Nilus for close and forecasting. Useful for a mid-size multi-currency team. About 29 minutes, posted 26 November 2025, around 40 views; recorded at a Techsommet event and posted by Nilus. Transcript not read.
Nilus | Scaling Cash Visibility ft. Raluca Dusa, VP Treasury and Risk at FCG
Flow Control Group's treasury head joins Nilus's CEO and sales lead to discuss tracking cash through more than 120 acquisitions in five years. The closest public reference for a roll-up buyer. About 36 minutes, posted 29 October 2026, under 20 views; hosted by the vendor. Transcript not read.
Nilus - Agentic AI in Treasury
A Strategic Treasurer webinar with a Nilus speaker on how agentic AI applies to treasury work. A third-party host gives broader context than vendor clips. About 57 minutes, posted 10 March 2026, around 300 views; the session features Nilus, so it leans toward its approach. Transcript not read.
Meet the Nilus Analyst Agent
A one-minute introduction to the Analyst Agent, which builds pivots and narrative drafts from treasury data. Relevant to analysts who answer ad hoc CFO questions. Posted 11 December 2025, around 220 views; vendor-made and brief, with no setup shown. Transcript not read.
Introducing Nilus Liquidity Agent
The launch video for the Liquidity Agent, showing visibility across entities, banks and currencies and suggested actions on breaches. Helps a treasurer judge how policy thresholds are set up. About 4 minutes, posted 31 March 2026, around 150 views; vendor launch content. Transcript not read.
AI Agents in Treasury: Live Workflows Demo
Nilus's head of AI runs a live session on agents that handle reconciliation and cash forecasting. The longest look at the product in action, useful before a proof demo. About 52 minutes, posted 1 April 2026, around 300 views; run by the vendor, so it shows its chosen workflows. Transcript not read.
When your company grows 2X your treasury complexity grows 10X
A clip of Nilus's CEO on a treasury podcast, arguing that complexity grows faster than revenue. Frames the multi-entity problem Nilus targets. Under 1 minute, posted 18 February 2026, around 80 views; vendor opinion with no product shown. Transcript not read.
TMS helps treasurers with audits
A treasury practitioner explains how a treasury system helps during audits. Relevant to controllers who care about audit trails. About 30 seconds, posted 18 February 2026, about 1,900 views; posted by Nilus and general to any TMS, so it shows nothing specific to Nilus. Transcript not read.
One Source Of Truth - How Frontline Turned Their Siloed Data Into Actionable Insights
Frontline's finance lead explains how Nilus pulled siloed bank and ERP data into one view. Useful for buyers weighing a consolidation layer over several banks. About 30 seconds, posted 12 March 2026, around 160 views; made by Nilus as a customer clip. Transcript not read.
From days to minutes - How Frontline Solved Reconciliation & Forecasting
Frontline's finance lead describes moving from manual data entry to automated reconciliation and forecasting in Nilus. Relevant for teams whose reconciliation eats days each month. About 30 seconds, posted 12 March 2026, around 100 views; a vendor-cut testimonial with no numbers shown. Transcript not read.
The 7-Day Advantage - Tapping Into Real-Time Cash Positioning
A finance leader at customer Frontline describes seeing the cash position a week earlier with Nilus. A short customer view for buyers who want a reference voice. About 30 seconds, posted 12 March 2026, around 80 views; produced by Nilus, so it is a selected testimonial. Transcript not read.
Beyond Chatbots
Nilus's CEO explains how the company separates AI bolt-ons from agents that do work. Useful for a CFO sorting vendor AI claims. Under 1 minute, posted 4 March 2026, about 1,000 views; vendor-made positioning with no product walkthrough. Transcript not read.
The Nilus Liquidity Agent
Nilus's CEO briefly describes the Liquidity Agent, which watches balances and policy limits across entities. A one-minute primer for a treasurer deciding whether to book a demo on buffer and covenant monitoring. About 1 minute, posted 18 February 2026, around 70 views; made by Nilus to promote the agent, so expect the vendor's framing only. Transcript not read.
Compliance depth
Homepage lists SOC 2 Type II, SOC 1, GDPR and SAML SSO; maker-checker payment approvals, sanctions screening and an immutable audit trail.
Scalability
Cites 100+ finance teams and 20,000+ bank and PSP connections; Taboola runs 16 banks, 22 entities and 18 currencies on it (case study).
AI / agent readiness
60 pre-built agents and a plain-language Agent Builder, with human approval before any payment or cash move; no public API or MCP server documented (1 Nov 2026).
Support
No published support hours or response times; the SLA targets 96 percent monthly availability on a reasonable-efforts basis.
Implementation
Cites a 30 to 60 day average deployment and a published four-week plan; Taboola's own case study lists a 3-month implementation.
Pricing transparency
Quote only with no pricing page; nine modules sold separately, and fees may change after the initial term with notice.
Modeling depth
13-week rolling forecast built from actual transactions that self-corrects against variance, with stress tests, runway and covenant headroom.
ERP & data connectivity
Banks via Swift, host-to-host, API or SFTP, some through aggregators such as Plaid; direct NetSuite integration; SAP and Oracle logos shown.
AF Score
Audit Friendly score for Nilus across eight dimensions, 5 November 2026.
Bank and PSP links
Bank and payment provider connections Nilus cites on its homepage (vendor figure, checked 1 November 2026); its terms name aggregators such as Plaid and MX as data sources.
Customers
Finance teams Nilus says it works with, per its homepage, checked 1 November 2026 (vendor figure).
Nilus platform (nine modules)
Priced on entities, accounts, integrations, workflow scope and implementation. Modules are sold separately and can be added over time.
The Delta academy
Free video courses, a handbook and webinars on agentic AI in treasury, open to anyone. It is general training, separate from product onboarding.
Connect banks and ERP, load 12 to 24 months of history
Set policy guardrails and approval workflows
Activate agents and run in parallel with current process
First 13-week forecast, first board pack, handoff
Payments still leave as a file
The Payments module page says Nilus generates a bank-ready instruction file covering ACH, wire, SEPA and RTP for upload to the bank portal. Direct delivery over Swift and partners, settlement status and automatic invoice matching are listed as near-term roadmap (checked 1 November 2026). The homepage describes multi-rail delivery matched back to invoice, so confirm what ships on your go-live date. If payments are in scope, put the delivery date in the order form.
Data access ends at termination
Nilus's subscription agreement says that on termination or expiry the customer loses all access to Customer Data and Output Data in the platform, and the customer alone must download it beforehand (checked 1 November 2026). Customer Data stays the customer's property, but no export window or format is promised. Negotiate a post-termination export period and the formats, including forecast history and audit logs.
Fees can change after the first term
The subscription agreement lets Nilus change its fees at any time after the initial subscription term, by email or in-platform notice. The customer's sole remedy is to ask to terminate before the change takes effect (checked 1 November 2026). No renewal cap is published. Put a cap and multi-year prices in the order form.
A thin, self-measured SLA
The SLA sets 96 percent monthly availability on a reasonable-efforts basis, which allows about 29 hours of downtime in a 30-day month (Audit Friendly estimate). It applies only 90 days after provisioning, Nilus's own measurement is the sole basis for disputes, and credits need a ticket within 24 hours and land on the annual renewal invoice. Downtime from third-party bank accounts and keys is excluded. Ask for a signed SLA with support response times.
Some bank feeds run through aggregators
The subscription agreement says account data can come through third-party APIs such as Salt Edge, Plaid, Finicity and MX, and that the customer accepts those providers' terms. The SLA excludes third-party accounts and keys from downtime. For each of your banks, ask whether the feed is Swift, host-to-host, a direct API or an aggregator, and what intraday data it carries.
FX, debt and investments are newer
In its 26 May 2026 buyer's guide, Nilus wrote that it is not the natural fit if a buyer needs a full enterprise TMS for complex FX, debt, hedging, payment factory or global risk workflows. By 1 November 2026 its site lists FX with hedge accounting, Debt, Investments and Bank Account Management modules. All eight case studies on its site cover cash positioning, forecasting, reconciliation or accounting automation. Ask for a live reference on each newer module you plan to buy.
Agent Fabric launched
Nilus launched the Agent Fabric: an agent library, Agent Builder, scheduler, workflow manager and report builder across its modules, with human approval required before any payment, cash move or irreversible step.
Liquidity Agent released
A monitoring agent that tracks balance floors, payroll readiness, revolver headroom, FX caps and bank concentration against set thresholds and suggests next steps.
The Delta academy opens
Nilus announced The Delta, a free virtual academy with three video courses, a handbook and webinars on agentic AI for treasury staff.
Analyst Agent introduced
An agent that builds pivots and narrative variance drafts from treasury data on request and keeps past analyses searchable.
Ranked on LinkedIn Top Startups in Israel
Nilus says it ranked number 9 on LinkedIn's 2025 Top Startups list for Israel.
Nilus is quote only. Its own 26 May 2026 buyer's guide says pricing depends on entities, accounts, integrations, workflow scope and implementation needs, and there is no pricing page or public Vendr benchmark (checked 1 November 2026).
For a rough frame, a 2 June 2026 Nilus guide puts modern agentic treasury platforms at $35,000 to $90,000 a year, and a 4 May 2026 post by its CEO prices an AI treasury agent at about $60,000 a year. Neither is a Nilus quote. Modules are sold separately, so the number depends on how many you start with.
Under the public subscription agreement, fees are invoiced in advance, due within 30 days and non-refundable, for an initial term set in the order form. After that term Nilus may change fees with notice, and the remedy is to terminate before the change.
Customer Data remains the customer's property, but on termination the customer loses all access to its data and outputs in the platform and must download them first. Negotiate an export window and formats, including forecast history and audit logs.
Nilus cites a 30 to 60 day average deployment and publishes a four-week plan: banks and ERP in week 1, guardrails and approvals in week 2, a parallel run in week 3, and the first 13-week forecast and board pack in week 4.
The Nilus team runs onboarding, with bank administrators and an ERP admin on the customer side. Taboola's case study lists a 3-month implementation across 16 banks and 22 entities, so plan for longer at that scale.
Not today. Nilus documents no public API and no MCP server, and the subscription agreement bars automated access to the platform without its prior written consent (checked 1 November 2026).
The AI runs inside Nilus instead: 60 pre-built agents, an Agent Builder and a scheduler, with human approval before any payment or cash move. Claude can work only with files exported from Nilus, such as Liquidity Agent CSVs.
Answered only from our own published research on this tool, never from general internet noise. If we cannot answer it well, our research agents will dig in and publish a sourced answer.