While working in the tax computation and advisory department at the Nigerian subsidiary of KPMG, the global professional services firm, Kelechi Ibe spent his days helping clients to calculate and structure their taxes. His role was to match the transaction descriptions, which were shared by clients, to the tax law. However, the process for each transaction remained the same.
By 2019, a year and a half at KPMG, that pattern had become hard to ignore. The logic was familiar, but the process remained manual, so he tried to automate it using Excel Macros, a tool for automating repetitive tasks.
“It didn’t work,” Ibe said, explaining that the task required using an intelligent tool that could read every transaction and determine the appropriate tax to apply. “This would only have been possible if every transaction were hardcoded into the database and had rules associated with each of them. It would have been tough and complicated, with a higher risk of errors,” he said
Seven years later, Ibe teamed up with Sam Ayo, a senior intelligence and machine learning engineer, to build TaxStreem, an automated tax compliance platform. Launched in March 2026, TaxStreem uses AI to automate tax computation and compliance by interpreting financial transactions in real time.
“AI understands context and nuances, and it gets better with training. A tax technology, as I had always envisaged, could only have been possible with AI,” Ibe said.
Tax compliance is not optional
Under Nigeria’s new tax regime, small businesses are required to use the e-invoicing system from July 2027 to digitally record their transactions for stricter tax compliance.
To properly issue tax-compliant invoices, businesses first need to understand the tax implications of every transaction they make, including what attracts value-added tax (VAT), what is exempt, what is zero-rated, what withholding tax applies, and why. That calculation and figuring-out layer is where TaxStreem operates.
TaxStreem is designed as an infrastructure layer for tax—a tool that sits directly on top of a business’s financial activity and interprets it in real time. The platform operates four tools that each handle a different task.
The first is TaxStreem Numens, an AI-powered tax intelligence and computation engine. The engine, according to Ibe, is trained on Nigerian tax laws and acts as the system’s core calculation tool.
He explained that when users connect their business bank accounts from providers such as GTBank, Access Bank, and fintech rails like Kuda, Paystack, and Flutterwave, Numens reads each transaction’s narration, interprets what it represents, assigns the correct tax treatment, and provides an explanation on why it took that particular decision.
Although users can still upload reconciliation documents manually to calculate their tax, Ibe noted that pulling transactions directly ensures no transactions are left out.
Its second engine, Flux, is an automated filing engine that logs into government portals, particularly the Nigeria Revenue Service (NRS) portal, to submit tax returns and retrieve proof of filing for businesses.
TaxStreem Prism, the third engine, focuses on invoicing and accounts, specifically Accounts Payable (AP) and Accounts Receivable (AR), a company’s revenue and expenditures. On the expenditure side, Prism replaces the traditional email-based invoice process with a dedicated vendor portal where vendors can upload their invoices.
Ibe noted that Prism analyses the uploaded invoice to check if there are errors in the tax calculation or tax ID, and sends a list of noted errors to both the business and the vendor that uploaded the invoice.
On the revenue accounts side, Prism allows businesses to generate invoices directly within the platform. Over a call, Ibe demo-ed how it works: businesses only need to input their product, and the system automatically determines the correct tax treatment. For example, entering ‘diapers’ would automatically be recorded as VAT-exempt, as classified under Nigeria’s tax laws. The invoice is then generated with the correct tax structure already applied.
Martina, TaxStreem’s AI chatbot, is the fourth layer. Beyond answering tax questions, it sits atop all transaction data and interprets business performance by offering insights for decision-making. TaxStreem combines its internal AI model, which is trained on Nigerian tax rules, with frontier models like Google’s Gemini to run minor agentic tasks.
“What we did is create a custom domain-driven model that is proprietary to us and leverage frontier models to run some mundane tasks for some of the things that would not require user data to protect data privacy within the financial sector,” co-founder Ayo noted.
The business of compliance
TaxStreem has two revenue streams, targeted at both small and medium enterprises (SMEs) and larger businesses. It charges a ₦25,000 ($18.49) monthly subscription and has an enterprise plan without a labelled pricetag. TaxStreem also sells its Application Programming Interface (APIs) to other platforms, particularly business banking and payments tools.
The startup operates in a market with existing expense management solutions for businesses, including FlexFinance, Bujeti, Duplo, and Bumpa, as well as accounting and bookkeeping tools like QuickBooks, FreshBooks, and Xero.
However, Ibe noted that he views such tools as partners that could integrate with the platform rather than competitors. For him, TaxStreem’s edge lies in its automated filing on government portals.
TaxStreem has raised an undisclosed friends and family round that Ibe estimated to be “tens of thousands of dollars.”
“I have not raised institutional funding, and it was very deliberate,” he noted. “I didn’t want to raise until we had launched and had some traction.”
He explained that TaxStreem’s next step is to add state-level tax filing on the Lagos Internal Revenue Service (LIRS) portal. Ibe also noted current API integrations with fintech platforms he chose not to name, against a backdrop of increasing enforcement.
“The government is going to enforce tax laws, and more businesses are going to suffer if they do not comply,” he said. “However, there is a technology alignment. Everything came together to give us a perfect time to build a solution like this.”
