Novo launches AI banking tools with approval-based controls
Novo launched AI-assisted banking, payroll, tax and credit tools on November 17, while its developer API remains a closed beta requiring scoped access and owner approval.

Why it matters
- Novo's November 17 launch combines AI-assisted banking, payroll, tax preparation and credit for small businesses.
- Customer spending on AI tools rose 493% from November 2024 to June 2026, while the number of paying businesses rose 145%.
- Novo's Banking Agent cannot move money or make changes without owner approval.
- The Novo Agent API remains a closed beta with individually scoped keys, short-lived tokens and no sandbox.
- AI-paying customers held twice as much cash at $250,000 in revenue and four times as much above $500,000.
By Marcus Bennett
Novo says monthly customer spending on a group of artificial intelligence (AI) tools rose 493% between November 2024 and June 2026, while the number of businesses paying for those tools rose 145%.
Novo based those figures on an analysis of anonymized, aggregated customer data. Its November 17 launch brings AI-assisted banking, payroll, tax preparation and credit into a more connected platform for small businesses.
Novo says natural-language access to financial information, transaction labeling and analytics can reduce the administrative work owners handle across separate applications. The aim is to make financial software respond more like an assistant that understands the business.
What changed on November 17?
Novo’s November 17 launch connects three operating workflows with a broader set of payment, integration and credit products.
- Account intelligence: plain-language search, transaction labeling, spending analytics and personalized recommendations.
- Payroll: employee and contractor payments, payroll tax withholding, filing, paystubs and year-end forms.
- Tax: federal and state preparation and filing for eligible sole proprietors and limited liability companies (LLCs).
Novo says the launch connects these services through its platform. The products do not create one automatic pool of credit; Novo determines eligibility for each financial product separately.
What can the Novo Banking Agent do?
Novo AI lets owners ask questions in plain language, locate transactions, examine spending patterns and receive personalized recommendations. An owner can describe an expense by its approximate amount, purpose and timing instead of recalling a merchant’s exact name.
Novo’s Banking Agent product page says answers draw on Novo’s help material and the customer’s account, with links to sources or relevant account screens. Novo says a general-purpose chatbot can explain cash flow, but useful account assistance needs access to the actual payments and balances behind the question.
Novo gives owners these access choices:
- Help content
- Help and navigation
- All account data
The in-app Banking Agent can prepare tasks and suggest settings. It cannot independently move money or make changes without owner approval. It can also direct users to human support.
“The businesses growing with AI run leaner and move faster, and they’re going to expect the same from their bank,” said Jackson Barnes, Novo’s CEO, in the announcement.
Novo identifies itself as a financial technology company rather than a bank. Middlesex Federal Savings, F.A. provides its banking services.
What can developers build now?
The Novo Agent API (application programming interface) remains a closed beta, a limited test restricted to Novo account owners, rather than unrestricted general availability.
The beta supports reading account information, transactions, bookkeeping data and card details. Developers can also freeze a card. Access uses individually scoped keys and short-lived tokens.
Developers work through REST (Representational State Transfer) requests and JSON (JavaScript Object Notation), common formats for requesting information and receiving structured responses.
Novo also marks externally written text as untrusted. The company says that distinction matters when an AI tool reads merchant descriptions or other free-text records: retrieved information should not automatically become an instruction to the agent.
Novo labels money movement and payee operations, meaning actions involving payment recipients, as planned capabilities. The current design requires owner approval on a phone.
The beta does not include an official Model Context Protocol (MCP) server, webhooks (event notifications) or a sandbox (isolated test environment). These limits separate what developers can build now from Novo’s broader agentic banking roadmap.
How does payroll connect to available cash?
Novo Payroll, powered by Gusto, brings W-2 employee and 1099 contractor payments into the account interface. It includes payroll tax withholding and filing, paystubs and year-end forms.
Novo says existing business information can prefill setup, reducing repeated entry across systems. The company presents the practical benefit as seeing a payroll obligation beside the money available to cover it. Connecting those records can make an approaching cash shortfall easier to spot before a payment falls due.
What changes for business taxes?
Novo Tax uses april’s embedded tax-filing technology. Novo says eligible sole proprietors and LLCs can prepare and file federal and state returns without filing fees, with Novo covering those costs.
Novo uses categorized income and expenses to help prefill the return. Its AI guide works through follow-up questions, and users review the information before filing.
Novo excludes transactions left uncategorized from the transfer into the return. Filing for the 2026 tax year opens in January 2027.
Which payment and credit products join the launch?
Novo’s wider launch adds instant Novo-to-Novo payments, connected external accounts and more than 10 native integrations. It also expands credit card features and adds installment loans and revenue-based funding.
Novo says eligibility operates separately across those products. Access to a payment method, card feature, loan or funding program does not make every other credit product automatically available.
Novo’s business savings account offers a variable 2% annual percentage yield (APY). It requires a $50,000 opening minimum and a $50,000 end-of-day balance to earn interest.
The announcement’s savings rate therefore applies only to businesses able to maintain that balance, not every checking customer.
Does customer AI spending correlate with higher cash?
The 493% spending increase and 145% rise in paying businesses came from Novo’s anonymized, aggregated customer analysis. Novo also compared cash positions between customers paying for AI tools and businesses that did not pay for them.
Novo’s comparison found that businesses with at least $250,000 in annual revenue and AI-paying customers held twice as much cash by the end of the Q3 2025-to-Q3 2026 comparison period. For businesses above $500,000 in revenue, the difference was fourfold.
Those comparisons show association, not proof that AI spending produced the additional cash. The data stops short of causation.
Novo says businesses adopting the tools may differ in size, sector, growth or financial management. The announcement provides no experimental design that isolates AI’s contribution.
What will determine whether the platform works?
Novo’s launch addresses a practical question: whether connected financial records can make everyday business administration less fragmented. The company combines account-grounded answers, embedded service providers and controlled access for outside software.
Novo says the platform’s usefulness will depend on whether owners can turn those connections into faster, more accurate decisions while retaining control of their money. The immediate test runs through a closed developer beta, owner approval for banking actions and a January 2027 start for 2026 tax filing.
Those boundaries will determine whether the service can deliver useful automation without giving AI unilateral authority over consequential financial actions.
Source: unite.ai
More from Marcus Bennett
Show full bio
Senior reporter covering companies, deals and strategy in artificial intelligence and machine learning at AI In Context.
718 articles
Related articles
- Iron Mountain says only 34% of financial data is AI-ready
- Connected financial data matters more than the AI model
- Only 34% of Financial Services Content Is Connected, Iron Mountain Finds
- AI Pricing in 2027: Buyers Will Pay Only for Completed Work
- Intuit and OpenAI sign $100M+ deal to bring finance apps into ChatGPT