For decades, growing companies followed a predictable software path. They started with a basic accounting tool such as QuickBooks, then moved into a larger enterprise resource planning system, or ERP, once the business became too complicated.
For many companies, that usually meant software such as Oracle NetSuite, Sage Intacct, SAP, Microsoft Dynamics, or Oracle Fusion. These products became some of the most important systems inside modern businesses because they stored the financial record that management teams, auditors, investors, and regulators depended on.
Now a new group of artificial intelligence companies is attacking that foundation.
And New York has become one of the most interesting places to watch the fight.
Companies including DualEntry, Basis, Tabs, Vic.ai, Trullion, Blue Onion, and Finaloop are headquartered in New York, while Rillet operates a substantial New York office alongside its San Francisco headquarters. Together, they are attacking different parts of accounting, from the general ledger and monthly close to accounts payable, revenue recognition, reconciliation, tax, audit, and ecommerce accounting.
Some are trying to make accountants much faster.
Others are trying to build AI workers that can perform accounting tasks.
And a smaller but increasingly important group is going much further: they want to replace the traditional ERP itself.
That leads to a much bigger question.
Can artificial intelligence really replace traditional ERP software?
Our analysis suggests the answer is becoming yes for certain kinds of businesses, but definitely not for every business yet.
The biggest change is not simply that companies are adding AI features to accounting software. The entire shape of the finance technology stack is starting to change.
The Short Answer: The Top AI Accounting Startups in NYC
NYC Tech Journal identified eight companies that deserve particular attention because they either have their headquarters in New York or maintain a meaningful operating presence in the city and are building technology directly around accounting and finance workflows.
| Company | NYC connection | Main product | What it is trying to replace or automate | Latest major disclosed round |
|---|---|---|---|---|
| Rillet | Major NYC office | AI-native ERP | General ledger, close, reporting, revenue accounting | $100M Series C |
| DualEntry | NYC headquarters | AI-native ERP | Mid-market financial ERP | $90M Series A |
| Basis | NYC headquarters | AI accounting agents | Human accounting work across firms | $100M Series B |
| Tabs | NYC headquarters | AI revenue platform | Billing, collections, revenue recognition | $55M Series B |
| Vic.ai | NYC operations | Autonomous finance platform | Accounts payable and spend workflows | $52M Series C |
| Trullion | NYC headquarters | AI accounting platform | Lease accounting, revenue recognition, audit | $15M financing |
| Blue Onion | NYC headquarters | Financial data platform | Ecommerce reconciliation and transaction accounting | $10M Series A |
| Finaloop | Brooklyn headquarters | AI-native accounting platform | Ecommerce accounting, inventory and operations | $35M Series A |
The most important point is that these companies are not all competing for exactly the same job.
Rillet and DualEntry are attempting to become the core financial system.
Finaloop is trying to do something similar for ecommerce and retail companies.
Tabs attacks the revenue side of finance.
Vic.ai focuses heavily on accounts payable and autonomous finance.
Blue Onion builds a highly detailed financial data layer beneath the general ledger.
Trullion focuses on difficult accounting and audit workflows.
Basis is different again. It is building AI agents that perform accounting work rather than trying to become the accounting database itself.
That difference explains where ERP disruption is likely to happen first.
Original Research: How NYC Tech Journal Built This AI Accounting Dataset
There are plenty of lists of accounting software companies online. That was not enough for this analysis.
We wanted to answer a harder question:
Which New York AI accounting companies are actually positioned to replace a traditional ERP, rather than simply making one small accounting task easier?
To answer that, NYC Tech Journal created an original dataset based on publicly available information collected through August 30, 2026.
How Companies Qualified for the Study
A company had to meet three basic tests.
First, it needed a meaningful New York City connection. That could mean having its headquarters in NYC or operating a substantial New York office.
Second, accounting or financial operations needed to be central to the product. A general AI company that happens to sell to finance departments did not qualify.
Third, artificial intelligence or advanced automation needed to be a major part of the product rather than a small feature added to otherwise traditional software.
That gave us an eight-company core sample.
The 10 ERP Capabilities We Tested
For each company, we examined public product pages and company documentation for clear evidence of ten capabilities.
| Capability | Why it matters |
|---|---|
| General ledger | The financial system of record |
| Accounts payable | Money the company owes |
| Accounts receivable and billing | Money customers owe |
| Revenue recognition | Converting contracts and sales into accounting revenue |
| Bank and cash reconciliation | Matching financial records with actual cash |
| Multi-entity consolidation | Running several legal entities or subsidiaries |
| Fixed assets | Tracking long-term assets and depreciation |
| Inventory and operations | Connecting physical business activity to accounting |
| Close and reporting | Producing usable financial results |
| Controls and auditability | Showing who did what and why |
A capability received credit only when we found clear public evidence that the company supported it.
A zero therefore does not necessarily mean that a feature does not exist. It means we could not establish it clearly enough from public information to include it in the model.
This is important because private software companies do not publish every product detail.
The NYC Tech Journal ERP Replacement Readiness Index
We then weighted the capabilities based on how important they are to replacing a traditional financial ERP.
The general ledger received the largest weight because a company cannot truly replace accounting software without becoming a financial system of record.
Multi-entity accounting, AP, AR, revenue, reconciliation, reporting, inventory, fixed assets, and controls were then included around it.
The final score runs from 0 to 100.
A score above 80 means the public product evidence suggests the company could potentially serve as a core accounting system for at least some businesses.
A score between 40 and 79 suggests a modular finance platform that can replace important ERP workflows but is more likely to operate alongside another system.
A score below 40 means the company is primarily an augmentation layer rather than an ERP replacement.
Which NYC AI Accounting Companies Look Most Like ERP Replacements?
NYC Tech Journal ERP Replacement Readiness Index
| Company | Score | Relative readiness |
|---|---|---|
| Rillet | 90/100 | ██████████████████ |
| DualEntry | 90/100 | ██████████████████ |
| Finaloop | 85/100 | █████████████████ |
| Tabs | 45/100 | █████████ |
| Blue Onion | 45/100 | █████████ |
| Basis | 25/100 | █████ |
| Vic.ai | 25/100 | █████ |
| Trullion | 25/100 | █████ |
The gap is striking.
Rillet, DualEntry, and Finaloop are building something much closer to a system of record. The remaining companies are mostly building powerful layers around that system.
That does not mean a company with a score of 25 is less important.
Basis, for example, may ultimately change accounting labor more dramatically than some ERP companies. Its lower score simply reflects the fact that Basis is designed to make accounting firms intelligent and automated rather than become the general ledger inside a business.
The Detailed Capability Matrix
The table below shows how the eight companies compare across the ten functions used in our scoring model.
| Company | GL | AP | AR | Revenue recognition | Reconciliation | Multi-entity | Fixed assets | Inventory | Close/reporting | Controls/audit |
|---|---|---|---|---|---|---|---|---|---|---|
| Rillet | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | — | ✓ | ✓ |
| DualEntry | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | — | ✓ | ✓ |
| Finaloop | ✓ | ✓ | ✓ | ✓ | ✓ | — | — | ✓ | ✓ | ✓ |
| Tabs | — | — | ✓ | ✓ | ✓ | — | — | — | ✓ | ✓ |
| Blue Onion | — | — | ✓ | ✓ | ✓ | — | — | — | ✓ | ✓ |
| Basis | — | — | — | — | ✓ | — | — | — | ✓ | ✓ |
| Vic.ai | — | ✓ | — | — | — | — | — | — | ✓ | ✓ |
| Trullion | — | — | — | ✓ | — | — | — | — | ✓ | ✓ |
Rillet publicly describes an automated general ledger, accounts receivable, accounts payable workflows, bank reconciliation, advanced revenue recognition, multi-entity consolidation, fixed-asset schedules, close automation, reporting, and financial controls.
DualEntry publicly lists general ledger, accounts payable, accounts receivable, cash management, reconciliation, multi-currency consolidation, revenue recognition, fixed assets, close management, audit trails, approval workflows, and other financial controls.
Finaloop has developed an unusually broad vertical platform for ecommerce. Its product combines a general ledger, reconciled financial statements, AP and AR support, revenue accounting, inventory and COGS tracking, reconciliations, operational data, and close workflows.
This distinction becomes critical when discussing whether AI can replace ERP.
New York Has Exactly the Kind of Economy Where AI Accounting Can Grow Fast
It is not an accident that so many finance automation companies are appearing in New York.
New York has an unusually large concentration of companies where accounting complexity appears long before factories, warehouses, or physical supply chains become the main operational problem.
Think about the industries around Manhattan and Brooklyn: software, fintech, financial services, advertising, professional services, ecommerce, media, private markets, healthcare technology, and venture-backed businesses.
These companies may have complicated billing, multiple entities, international operations, revenue recognition rules, large numbers of contracts, and demanding investors.
But many do not manufacture physical products.
That makes financial ERP a particularly attractive target.
New York Has an Enormous Finance Labor Base
Business and financial operations represent a larger share of employment in the New York metropolitan area than they do nationally.
The city also has a deep concentration of financial managers, accountants, auditors, controllers, consultants, bankers, and finance operators.
That creates unusually powerful economics for automation.
If software saves five hours a month in a low-cost administrative role, the value may be modest.
If it saves hundreds of hours across controllers, accounting managers, finance leaders, external accountants, implementation consultants, auditors, and operations staff in New York, the financial value can become enormous.
Why Accounting Automation Is Economically Attractive in New York
| NYC metro finance indicator | Latest figure used |
|---|---|
| Business and financial operations share of employment | 7.6% |
| U.S. share | 6.8% |
| NYC concentration premium | About 11.8% |
| Financial managers | 89,960 |
| Mean annual financial-manager pay | $250,570 |
| Accountants and auditors | About 111,930 |
| Median accountant and auditor pay | About $105,650 |
These numbers explain something important about New York’s AI ecosystem.
The city does not merely have software engineers who can build accounting AI. It has an enormous nearby customer base that can tell those engineers exactly where accounting breaks.
That customer density can be a major startup advantage.
Rillet: The Most Direct Attack on Traditional ERP
Rillet is one of the clearest examples of AI moving from helpful accounting feature to actual ERP replacement.
The company now describes itself as an AI-native ERP. Its platform includes a general ledger, multi-entity consolidation, revenue recognition, accounts receivable, close management, reconciliation, reporting, integrations, AI agents, and accounting controls.
The company has also grown extremely quickly.
In August 2026, Rillet announced a $100 million Series C at a $1 billion valuation, bringing total reported funding above $200 million.
Most importantly for the ERP question, customers are not simply adding Rillet on top of old software.
Some are replacing traditional accounting systems with it.
Why Rillet Matters
Traditional accounting software was designed mainly around humans entering, reviewing, and closing information.
Rillet is trying to reverse that model.
Data enters continuously. Accounting logic processes it. AI handles more of the repetitive work. Humans focus on exceptions, judgment, controls, and decisions.
That is a fundamentally different architecture.
For New York CFOs evaluating whether an AI-native ERP can genuinely replace an incumbent accounting platform, Rillet belongs near the top of the shortlist.
DualEntry: New York’s Most Direct Homegrown ERP Challenger
If Rillet shows that AI-native ERP has become a serious software category, DualEntry shows how aggressively New York founders are entering it.
DualEntry is headquartered at 7 World Trade Center and describes itself as an AI-native ERP for mid-market businesses.
It was founded in 2024 and announced a $90 million Series A in October 2025, bringing total funding above $100 million.
That is a remarkable amount of capital for such a young company.
The product is designed around exactly the point where businesses usually begin considering NetSuite, Sage Intacct, or another mid-market ERP.
DualEntry publicly lists general ledger, accounts payable, accounts receivable, cash management, purchase orders, close management, account reconciliation, bank feeds, revenue recognition, multi-currency consolidation, financial planning, fixed assets, approval workflows, audit trails, custom roles, and AI tools.
Migration May Be the Real Competitive Weapon
Traditional ERP replacements are painful partly because migration is painful.
Historical data has to be cleaned, mapped, imported, checked, and reconciled. Consultants often become deeply involved.
That can turn an accounting software project into a major business project.
DualEntry is trying to make migration part of the product.
That is strategically important because the biggest barrier protecting old ERP vendors may not be product quality.
It may be switching cost.
If AI can reduce that cost, competition becomes much more dangerous for incumbents.
Basis: AI Accountants Rather Than an AI ERP
Basis is perhaps the most important company in this article that does not score highly in our ERP Replacement Readiness Index.
That is intentional.
Basis is building something different.
The New York-headquartered company develops AI agents designed specifically for accounting firms.
Its agents can work on reconciliations, journal entries, tax preparation, audit procedures, financial analysis, and related accounting tasks.
Basis Attacks the Labor Layer
This is important because accounting has two technology problems.
The first is bad software.
The second is the enormous amount of human work required around that software.
Basis attacks the second problem.
Instead of replacing NetSuite with another general ledger, Basis can potentially place AI labor on top of the systems accounting firms already use.
That could become extremely disruptive.
An accounting firm’s traditional growth model is closely connected to employee capacity.
More clients usually mean more people doing preparation, reconciliation, testing, review, and administrative work.
If AI agents can perform the first pass of that work, the economics of an accounting firm may change even if the firm’s ERP software does not.
Tabs: Rebuilding the Revenue Side of Finance
Many finance systems work reasonably well until a company starts selling in complicated ways.
Then contracts appear.
Usage pricing appears.
Different billing schedules appear.
Credits, amendments, renewals, discounts, collections, and revenue recognition begin interacting with one another.
That is where Tabs is positioning itself.
The New York-headquartered company focuses on billing and revenue automation.
Its platform covers contract ingestion, invoicing, receivables, payments, collections, revenue recognition, reporting, cash forecasting, and related revenue workflows.
Tabs Shows How the ERP May Break Into Components
Tabs is especially important because it provides evidence for another future.
Perhaps companies will not replace a traditional ERP with one giant AI ERP.
Perhaps the giant ERP itself will be broken apart.
Tabs already integrates its revenue-recognition workflows with existing ERPs.
It has also partnered with Rillet around a headless ERP framework, where specialized systems exchange data rather than forcing every workflow into one enormous software suite.
That model could become one of the most important enterprise software trends of the next few years.
Vic.ai: Turning Accounts Payable Into an Autonomous Workflow
Accounts payable is one of the clearest early opportunities for accounting AI.
Invoices arrive in many formats.
Information has to be extracted.
Bills have to be coded.
Approval rules must be followed.
Duplicate payments have to be avoided.
Purchase orders may need to be matched.
Payments then need to be executed.
Historically, people spend a surprising amount of time moving information through this process.
Vic.ai has spent years trying to automate it.
Vic.ai Does Not Need to Replace the ERP to Create Major Value
This is a useful strategic lesson.
A company does not need to own the general ledger to become important.
If Vic.ai can automate a large percentage of AP work while feeding accurate entries into an existing financial system, the customer’s ERP may remain in place while the human workflow around it changes dramatically.
For many large enterprises, this may be the safer path to AI adoption.
Replace tasks first.
Replace workflows second.
Replace the financial core only when the technology and organization are ready.
Trullion: AI for Accounting Work That Requires Evidence
Trullion is headquartered in New York and focuses on accounting and audit workflows where the underlying evidence matters.
Its products include lease accounting, revenue recognition, reporting, audit automation, and AI-supported document analysis.
This is an important category because accounting is not simply about generating a number.
You often need to prove where the number came from.
A lease calculation needs to connect back to a contract.
Revenue accounting needs to connect back to commercial terms.
Audit work requires evidence.
Policies need to be followed consistently.
Trullion Highlights One of AI Accounting’s Hardest Problems
General-purpose AI is good at generating plausible answers.
Accounting cannot run on plausible answers.
The numbers must tie out.
The accounting treatment needs evidence.
The calculations need to be reproducible.
Someone must be able to review how the conclusion was reached.
This is why some of the most valuable accounting AI may not be the products with the flashiest chatbot.
The winners may be systems that combine AI flexibility with deterministic accounting logic, structured data, evidence trails, and human approval.
Blue Onion: Building Clean Financial Data Before AI Touches It
Artificial intelligence has another accounting problem that is easy to overlook.
Bad data.
If payments, refunds, fees, orders, deposits, and processor payouts do not reconcile correctly, putting a powerful AI model on top of the mess does not solve the problem.
It can simply produce faster confusion.
Blue Onion is attacking that layer.
The New York company describes itself as a financial intelligence platform and automated transaction-level subledger.
It reconciles data across sales channels, payment processors, banks, and ERPs so finance teams can trace amounts back to the individual transactions that created them.
Why the Subledger Could Become More Important in the AI Era
Traditional software stacks often move summarized data from one system into another.
AI agents work better when they have detailed, reliable context.
That could increase the importance of transaction-level financial infrastructure.
Blue Onion’s strategy is therefore interesting beyond ecommerce.
It reflects a broader idea: before finance teams can safely automate decisions, the underlying financial data needs to be cleaned, reconciled, and traceable.
The future AI finance stack may contain far more intelligence, but it may also demand much better data discipline.
Finaloop: A Vertical ERP Alternative for Ecommerce
Finaloop may be one of the strongest examples of why vertical AI could disrupt traditional ERP before general AI does.
The Brooklyn-headquartered company focuses specifically on ecommerce, retail, wholesale, and consumer brands.
It combines accounting software, financial operations, inventory management, reconciliations, reporting, analytics, and human accounting support.
Vertical ERP May Be Easier to Build Than Universal ERP
A generic ERP needs to support enormous variation.
A manufacturer operates differently from a software company.
A hospital operates differently from an ecommerce brand.
A construction company operates differently from a fintech startup.
Vertical software can make narrower assumptions.
Finaloop knows its customers are likely to care about Shopify, Amazon, payment processors, returns, inventory, COGS, warehouses, marketplaces, and multichannel profitability.
That allows it to build much deeper automation around a specific operating model.
This may be one of the biggest weaknesses in the traditional ERP model.
A giant horizontal ERP tries to serve everyone.
An AI-native vertical platform can be designed around the exact workflows of one industry.
Original Research: Where Is the Funding Going?
We also looked at the latest major disclosed funding round for each company in our eight-company sample.
This is not total capital raised.
Using the latest major round gives us a cleaner way to compare recent investor commitment without mixing years of historical financing.
Latest Major Funding Round in Our NYC AI Accounting Sample
| Company | Latest major round used | Relative size |
|---|---|---|
| Rillet | $100M | ████████████████████ |
| Basis | $100M | ████████████████████ |
| DualEntry | $90M | ██████████████████ |
| Tabs | $55M | ███████████ |
| Vic.ai | $52M | ██████████ |
| Finaloop | $35M | ███████ |
| Trullion | $15M | ███ |
| Blue Onion | $10M | ██ |
The total latest-round capital represented in this sample is approximately $457 million.
The three largest rounds alone, Rillet, Basis, and DualEntry, represent about 63.5% of the capital in this sample.
That concentration matters.
Investors appear particularly willing to place very large bets on two ideas.
AI can replace the financial system itself.
And AI agents can perform meaningful accounting labor.
Funding by Product Architecture
| Product model | Companies | Latest-round capital | Share |
|---|---|---|---|
| ERP and system replacement | Rillet, DualEntry, Finaloop | $225M | 49.2% |
| Workflow specialist | Tabs, Vic.ai, Trullion, Blue Onion | $132M | 28.9% |
| AI labor and agent layer | Basis | $100M | 21.9% |
Nearly half of the latest-round capital in our sample is going directly toward companies that can plausibly replace the accounting system itself for some customers.
That is a powerful signal.
Five years ago, building a brand-new general ledger would have looked like an unusually difficult venture investment.
Today, major investors are funding multiple attempts.
The market is clearly moving beyond experimentation.
Why Traditional ERP Software Is Suddenly Vulnerable
Traditional ERP systems did not become unpopular because accounting stopped mattering.
Quite the opposite.
They became vulnerable because accounting became more complicated while the user experience often stayed difficult.
Implementation Is Too Heavy
Companies can spend months selecting an ERP, designing a chart of accounts, migrating historical data, configuring workflows, connecting other systems, testing controls, training users, and fixing unexpected problems.
This makes changing systems frightening.
The incumbent’s strongest advantage often becomes the difficulty of leaving.
AI changes that equation if it can automate mapping, classification, historical data migration, system configuration, testing, and reconciliation.
That could reduce one of the biggest barriers protecting traditional ERP vendors.
Too Much Work Happens Outside the ERP
Ask a controller where the monthly close happens and the truthful answer may not be inside the ERP.
It may happen across the ERP, spreadsheets, emails, Slack messages, contract folders, bank portals, billing systems, and manually maintained schedules.
That creates the perfect environment for AI.
AI is strongest when it can gather information across several sources, identify patterns, perform repetitive reasoning, draft work, and send humans the exceptions.
The new accounting companies are being designed around that reality.
Traditional ERP Was Built Around Human Clicks
Most legacy software assumes that a person is navigating screens.
Find a transaction.
Open a record.
Select an account.
Enter a value.
Run a report.
Export the report.
Manipulate it in Excel.
Send it to someone.
An AI-native system can start from a different assumption.
The software can monitor the underlying events and perform the work continuously.
That is a much larger change than simply adding a chatbot to an old ERP.
Can AI Really Replace ERP?
For some businesses, yes.
For others, not yet.
The answer depends heavily on what the word ERP means inside the company.
A SaaS Company Has a Much Easier Replacement Path
Imagine a venture-backed software company.
Its operating stack might include Salesforce for CRM, Stripe for payments, Ramp for spend, Rippling for payroll, a bank, and an accounting platform.
The ERP may mainly be responsible for the financial system of record, revenue schedules, journal entries, multi-entity accounting, reconciliation, and reporting.
That is precisely the area where Rillet and DualEntry are becoming strong.
Replacing the legacy financial ERP in this type of business is increasingly realistic.
A Manufacturer Is a Different Problem
Now imagine a manufacturer.
The ERP may track warehouses, raw materials, bills of materials, work orders, production schedules, serial numbers, landed costs, purchasing, inventory movements, and plant operations.
That is much harder.
AI-native financial ERP is developing faster than AI-native operational ERP.
That difference will determine which industries migrate first.
The More Likely Future: ERP Becomes a Network Rather Than One Giant Suite
The traditional ERP idea is simple.
Put everything in one giant system.
Finance.
Inventory.
Orders.
Purchasing.
Sometimes payroll.
Sometimes planning.
Sometimes CRM.
The result can be powerful, but it can also become slow and complicated.
AI may enable a different model.
The Headless ERP
A headless ERP keeps a trusted financial core but allows specialized systems to handle particular workflows.
Tabs handles revenue.
A spend platform handles employee expenses and payments.
A payroll system handles payroll.
Blue Onion handles detailed ecommerce reconciliation.
An inventory platform handles inventory.
Rillet or DualEntry holds the financial record.
AI agents move between these systems and perform work.
That is much closer to how modern technology companies already operate.
The idea may sound technical, but the business logic is simple.
Use the best system for each job while keeping the financial data connected.
AI makes this model more practical because intelligent agents can increasingly handle the messy work between applications.
Accounting AI Is Also Being Pulled Forward by a Talent Problem
Software is not the only pressure changing accounting.
The profession has a people problem.
Accounting degree completions have fallen in recent years, while many firms and finance departments continue to struggle with staffing pressure.
At the same time, accountants are not expected to disappear.
The more likely change is that automation reduces routine work while making analytical and advisory work more important.
This is an important distinction.
AI accounting does not require the number of accountants to collapse for the technology to create huge economic value.
The industry simply needs accountants to produce more output per person.
That is already attractive when skilled finance workers are expensive and difficult to hire.
In New York, where finance salaries are particularly high, the incentive becomes even stronger.
What Work Will AI Remove First?
The easiest mistake is to imagine AI replacing the accountant.
Jobs are rarely that simple.
Accounting jobs consist of dozens of smaller tasks.
Some are much easier to automate than others.
Transaction Matching Is an Obvious Target
Reconciliation often involves matching records from different sources.
A payment appears in one system.
An invoice appears somewhere else.
A bank transaction appears somewhere else.
Software can evaluate thousands of possible matches much faster than a human.
That makes reconciliation one of the clearest AI opportunities.
Data Extraction Will Keep Disappearing
Accounting teams still copy important information from invoices, contracts, leases, statements, purchase orders, and tax documents.
Modern AI can read many of these documents directly.
The human role increasingly moves from entering information to reviewing exceptions.
Journal Entry Preparation Will Become More Automated
Journal entries are structured.
They require accounts, amounts, dates, explanations, and supporting evidence.
AI systems can increasingly prepare the first draft of these entries automatically.
Human accountants will still review important judgments, but the amount of manual preparation can fall sharply.
The Month-End Close Could Become Continuous
Traditional accounting operates in batches.
The month happens.
Then the accounting team spends days trying to understand it.
AI-native platforms are moving toward continuous accounting.
Reconciliations can happen as transactions arrive.
Revenue schedules can update automatically.
Unusual activity can be flagged immediately.
Supporting documents can remain attached to the transaction.
The idea is that month-end eventually becomes a review event rather than a massive reconstruction project.
What AI Should Not Control Without Strong Guardrails
Accounting automation becomes more powerful when companies understand where human judgment remains essential.
Material Accounting Policy Decisions
Software can help gather evidence and model the accounting treatment.
But difficult judgments around materiality, uncertain estimates, unusual transactions, acquisition accounting, impairment, legal interpretation, and disclosure can have major consequences.
Those decisions require accountable professionals.
Approval Authority
AI can route an invoice.
It can recommend approval.
It can find supporting evidence.
But organizations still need clear rules around who can commit company money, modify financial records, approve journals, or override controls.
The automation should strengthen separation of duties rather than weaken it.
Audit Evidence
An AI answer is not enough.
A company needs to show the source documents, logic, calculations, approvals, changes, and history behind the answer.
That is why audit trails appear repeatedly across serious accounting AI products.
AI needs to be explainable enough for finance teams, auditors, boards, and regulators to trust the output.
How New York CFOs Should Evaluate an AI Accounting Platform
The wrong way to buy AI accounting software is to begin with the question:
Does it use AI?
Almost everything will claim to use AI.
A better evaluation begins with the financial workflow.
Start With the System-of-Record Question
Ask whether the product will become your official general ledger.
If the answer is yes, the evaluation needs to be extremely strict.
Historical transactions must survive migration.
Opening balances must match.
Financial statements must reconcile.
Controls need to work.
Audit history must remain available.
Permissions need to be correct.
Every important integration must be tested.
A tool that helps automate invoices is useful.
A system holding the official financial record is mission critical.
Those are different risk levels.
Measure Automation at the Task Level
Do not accept a statement such as “we automate the close.”
Break the close into actual work.
Bank reconciliation.
Credit-card reconciliation.
Accruals.
Prepaids.
Deferred revenue.
Intercompany entries.
Fixed assets.
Flux analysis.
Journal review.
Financial statement preparation.
Management reporting.
Then measure how many minutes or hours each task takes before and after the new system.
That turns an AI demonstration into a real business case.
Make the Vendor Prove Exception Handling
The happy path is not the hard part of accounting.
The difficult transactions are what matter.
Ask the vendor to process messy examples from your own business.
A contract amendment.
A partial payment.
A refund.
A foreign-currency transaction.
A corrected invoice.
An intercompany transaction.
A backdated journal.
A failed payment.
A strange revenue arrangement.
A migration adjustment.
You learn far more from seeing how software handles ten ugly transactions than from watching a polished 30-minute demo.
A Practical ERP Replacement Scorecard
Finance leaders can use the following framework before changing their core accounting platform.
| Evaluation area | Suggested weight |
|---|---|
| Accounting correctness | 20% |
| Migration reliability | 15% |
| Controls and audit trail | 15% |
| Workflow automation | 15% |
| Integrations | 10% |
| Reporting | 10% |
| Multi-entity and global support | 5% |
| Implementation burden | 5% |
| Vendor stability | 5% |
The most important factor should remain accounting correctness.
That sounds obvious, but AI software makes flashy demonstrations easy.
A beautiful interface cannot compensate for an incorrect trial balance.
A Better 90-Day Way to Test AI Accounting
Companies should resist the temptation to replace their ERP simply because the new software looks dramatically better.
A controlled migration is safer.
Days 1–30: Build the Baseline
Measure the current system before changing anything.
Document close time, reconciliation hours, number of manual journal entries, spreadsheet dependencies, implementation costs, integration failures, reporting delays, and the number of people involved.
Without a baseline, it becomes impossible to prove that the AI platform is actually better.
Days 31–60: Run Difficult Workflows
Use real company data where security and policy allow.
Test the difficult edge cases.
Have accountants verify the results.
Ask the auditor what evidence would be required.
Measure exception rates rather than just automation rates.
A vendor claiming 95% automation may sound stronger than one claiming 80%, but the more important question is whether the remaining 5% contains the most dangerous transactions.
Days 61–90: Parallel Close
For a system-of-record replacement, run at least one controlled close against the existing environment.
Compare trial balances.
Compare financial statements.
Compare reconciliations.
Compare revenue schedules.
Compare intercompany activity.
Compare adjustment logs.
Document every difference.
The goal is not merely to prove that the new system works.
The goal is to understand exactly where it fails before the old system disappears.
Which Businesses Can Replace Traditional ERP First?
Our analysis suggests that ERP replacement readiness depends heavily on the operating model.
| Business type | AI-native financial ERP readiness |
|---|---|
| SaaS | Very high |
| Professional services | Very high |
| AI and software startups | Very high |
| Fintech | High, subject to controls |
| Ecommerce | High with vertical platforms |
| Media and advertising | High |
| Healthcare software | High, depending on integrations |
| Retail with complex inventory | Medium |
| Wholesale and distribution | Medium |
| Manufacturing | Low to medium today |
| Heavy industrial operations | Low today |
This is why New York may be unusually well suited to the first wave.
A large share of the city’s fast-growing companies live toward the upper half of this table.
Their accounting is complicated.
Their finance teams are expensive.
But their physical operating systems may not be deeply dependent on the ERP.
That is exactly where a financial-core replacement can move fastest.
The Biggest Threat to Traditional ERP May Not Be Another ERP
The most interesting result from our research is that traditional ERP faces attacks from three directions at once.
Rillet and DualEntry attack the system of record.
Tabs, Vic.ai, Trullion, and Blue Onion attack major workflows around it.
Basis attacks the human labor required to operate the entire process.
This creates a difficult strategic problem for legacy vendors.
Even if NetSuite keeps the general ledger, Tabs could take billing.
Vic.ai could take AP.
Specialized platforms could take reconciliation.
AI agents could perform the manual work accountants once did around the ERP.
The legacy system could technically remain installed while becoming less important every year.
That may be more dangerous than a direct replacement.
AI-Native ERP Could Change ERP Pricing Too
Traditional ERP economics often extend far beyond the subscription.
There may be implementation partners.
Consultants.
Custom scripts.
Integration software.
Data migration.
Training.
Support contracts.
Extra modules.
Internal administrators.
Spreadsheet processes built around missing features.
A newer AI-native platform can attack this entire cost structure.
The battle is not simply about whose accounting software costs less.
The real question is what it costs the company to operate the entire finance technology system.
That includes people.
If a more expensive AI platform eliminates hundreds of consulting and accounting hours, its total economic value can still be much stronger.
What Traditional ERP Vendors Still Have That Startups Do Not
It would be a mistake to assume decades of enterprise software development can be erased immediately.
Traditional ERP vendors still have important advantages.
Enormous Functional Depth
Large ERP systems have spent decades building obscure features because some customer somewhere needed them.
Global tax rules.
Localization.
Manufacturing.
Inventory costing.
Procurement.
Project accounting.
Payroll integrations.
Government reporting.
Industry-specific controls.
AI startups will need time to match that depth.
Huge Implementation Ecosystems
Thousands of consultants understand traditional systems.
That may make implementations expensive, but it also gives customers access to talent when something goes wrong.
Young companies have much smaller ecosystems.
Long Audit Histories
Public companies and auditors understand incumbent platforms.
A new financial system has to earn that institutional trust.
The technology race is not just about features.
It is also about trust.
What NYC Tech Journal’s Dataset Suggests About the Next Five Years
The numbers point toward several likely developments.
The General Ledger Is Becoming Competitive Again
For years, startups tended to build around the general ledger because replacing it looked too difficult.
That assumption has changed.
Rillet and DualEntry have raised enormous rounds specifically around modernizing the financial core.
That alone makes the next few years of accounting software far more interesting.
AI Agents Will Become a Normal Part of Accounting Teams
Basis is already built around this idea.
Rillet is building specialized agents into its ledger.
Tabs is introducing agents into revenue workflows.
Other companies are moving in the same direction.
The future accounting department may consist of fewer people manually moving data and more people supervising automated financial processes.
Vertical Accounting Platforms Will Become Stronger
Finaloop shows what happens when the software understands the exact economics of a particular industry.
Expect similar platforms for healthcare, real estate, construction, restaurants, logistics, professional services, and other complicated markets.
New York’s industry diversity makes it a natural place for many of those companies to emerge.
The ERP May Shrink
Instead of becoming larger, the central ERP may become smaller.
It may hold the trusted financial record while specialized applications handle workflows around it.
AP could live elsewhere.
Billing could live elsewhere.
Inventory could live elsewhere.
Payroll already often lives elsewhere.
AI agents could connect the pieces.
The result would still be an enterprise resource planning system, but it would look less like a giant application and more like a connected financial network.
Can Artificial Intelligence Replace Traditional ERP Software?
The clearest answer is yes for some businesses and not yet for others.
For a SaaS company, professional-services firm, fintech, technology startup, or some ecommerce businesses, the gap has narrowed dramatically.
The combination of a modern general ledger, native integrations, continuous reconciliation, AI-assisted close, automated revenue workflows, strong controls, and specialized operating systems can now form a legitimate alternative to older financial software.
For manufacturing companies and organizations that depend heavily on deep inventory, production, procurement, warehouse, and industry-specific functionality, traditional ERP remains far harder to replace.
But that should not make incumbent vendors comfortable.
The disruption does not require every customer to leave.
It begins when the most modern customers stop automatically assuming they need NetSuite, Oracle, SAP, or another traditional system as they grow.
That change is already happening.
The default choice is no longer as obvious as it once was.
Why New York Could Become a Major Center for AI Accounting
Accounting AI needs more than artificial intelligence talent.
It needs accounting knowledge.
It needs CFOs.
Controllers.
Auditors.
Fintech companies.
Banks.
Private-equity firms.
Accounting firms.
Enterprise customers.
Complex businesses.
Investors.
New York has all of them packed into one metropolitan area.
That density matters because the hardest accounting problems are rarely discovered in a machine-learning laboratory.
They are discovered during month-end close.
During an audit.
During an ERP implementation.
During a revenue-recognition review.
During an acquisition.
During an intercompany reconciliation.
During a board meeting where the CFO cannot explain why two systems show different numbers.
New York gives founders unusually close access to those problems.
The result is becoming visible in the companies now being built.
Basis is putting AI agents inside accounting firms.
DualEntry is building an ERP from Lower Manhattan.
Tabs is rebuilding revenue operations.
Blue Onion is cleaning ecommerce financial data.
Finaloop is building a vertical financial operating system from Brooklyn.
Trullion is automating complex accounting and audit work.
Vic.ai continues pushing accounts payable toward autonomous operation.
And Rillet is using its New York presence as part of a broader attempt to replace some of the largest accounting systems in the world.
That is no longer a collection of isolated startups.
It is beginning to look like an ecosystem.
Final Takeaway
Traditional ERP software is not about to disappear overnight.
But its biggest protection has always been the assumption that replacing it was too difficult.
Artificial intelligence is attacking that assumption.
AI can make migration faster.
It can automate transaction coding.
It can reconcile accounts.
It can interpret contracts.
It can generate schedules.
It can prepare journal entries.
It can investigate exceptions.
It can accelerate reporting.
It can help accountants review enormous amounts of information without manually moving every number.
The most advanced companies are combining those abilities with completely new accounting systems rather than simply adding AI features to old ones.
NYC Tech Journal’s original analysis of eight New York-connected accounting AI companies found that Rillet, DualEntry, and Finaloop already show broad enough public functionality to qualify as credible system-replacement candidates for the right businesses, while companies such as Tabs, Blue Onion, Vic.ai, Trullion, and Basis are removing important pieces of work from the traditional ERP environment.
Our funding analysis also shows investors putting serious money behind every layer of that change.
The latest major disclosed rounds across the eight-company sample total approximately $457 million, with almost half going to companies attempting direct system replacement.
The winners may not eliminate the ERP.
They may redefine what an ERP is.
Instead of one giant piece of software filled with screens that finance teams spend their careers learning to navigate, the next ERP may be a much smaller financial core surrounded by specialized applications and AI agents that continuously perform the work.
If that happens, New York will have played an unusually large role in building it.
