Decide, a Nigerian AI Startup With No Funding, Ranks Fourth Globally for Spreadsheet Accuracy

A small Nigerian AI startup, Decide, has unexpectedly climbed into global conversation after placing fourth in SpreadsheetBench, one of the most trusted benchmarks for testing AI agents on real spreadsheet challenges. The result has drawn attention partly because Decide has no funding, a team of only three, and launched publicly just a few months ago.

The benchmark compares AI agents on tasks people struggle with every day in Excel, and Decide’s score places it beside heavily financed global players. Yet the startup sits far behind no giant. Nobie Agent ranked first, Shortcut.ai followed closely, and Qingqiu Agent came after — all far bigger companies with stronger financial muscle. Shortcut.ai, for instance, was created by Fundamental Research Labs, a company with more than $30 million in funding, while Qingqiu Agent has more than 5,000 employees and a valuation near $5 billion.

However, Decide’s founder, former Flutterwave engineer Abiodun Adetona, said the performance proves that good product thinking can still break through in a crowded AI marketplace. “Rather than testing AI on artificial or simplified problems, SpreadsheetBench is built from real Excel questions sourced from online forums where users seek help with complex spreadsheet issues,” he explained.

The benchmark’s “Verified” section, which Decide was tested on, contains 400 carefully selected tasks. Decide solved 330 of them, earning an accuracy score of 82.5 per cent. Adetona said this subset ensures the results are consistent across all agents and reflect how AI performs in real workplace scenarios.

Why the Ranking Matters

SpreadsheetBench was introduced at NeurIPS 2024, a leading global conference on AI research, and was developed by teams from Tsinghua University and Renmin University of China. Its emphasis on real-world tasks explains why AI researchers and companies treat its rankings seriously.

Decide itself was inspired by the tedious hours knowledge workers spend cleaning data, debugging spreadsheet formulas, or navigating dozens of sheets. The startup quickly caught early interest — gaining 1,000 users within 24 days of launch. According to Adetona, its user base has now passed 3,000, with some already paying for access.

The rise has also sparked discussion within Africa’s tech community about whether lightweight, sharply focused AI tools can compete globally without the backing of deep-pocketed investors.

A Growing AI Movement Across Africa

The achievement comes as Africa’s AI scene gains momentum, supported by initiatives such as the Google for Startups Accelerator Africa. Google recently opened applications for the tenth cohort, which focuses on AI-first solutions across the continent. Head of Startup Ecosystem Africa, Folarin Aiyegbusi, said the programme aims to equip startups with the technical foundation needed to build advanced solutions.

“Africa’s tech landscape is seeing a vibrant shift toward deep tech innovation,” Aiyegbusi said while discussing the evolving scene. He added that this year’s accelerator focuses on health and broader societal impact by turning selected startups into “research labs of the continent”.

The accelerator, launched in 2018, has already supported over 180 African startups that have raised more than $350 million cumulatively and created over 3,700 jobs.