Bangladesh must build AI foundations, not just chase trends: Vivasoft CEO

As artificial intelligence rapidly reshapes the global technology landscape, Bangladesh's software sector faces the dual challenge of adopting AI tools and fostering an ecosystem capable of producing AI products, rather than relying on outsourcing. Shafiul Hasan Tareq, CEO and co-founder of software company Vivasoft, emphasizes that Bangladesh's long-term competitiveness in the AI era hinges on strengthening its educational and infrastructural foundations.

“

There are multiple segments. For core model development, you must be highly competent. If you want to work at places like OpenAI or train your own models, you need a very sharp IQ and dedication.

Tareq advises that while cutting-edge AI model development requires exceptional expertise and resources, most graduates entering Bangladesh's software industry should prioritize learning how to use AI effectively. He suggests focusing on mastering coding agents and prompt engineering, coupled with the critical ability to evaluate AI-generated responses. "The main problem we see is people simply copy-pasting AI answers as absolute truths," Tareq stated. "Instead, you should challenge the AI using your own research and knowledge to get more accurate, filtered results."

“

We lack the infrastructure and GPU capacity required for this.

He points to significant structural limitations within Bangladesh, particularly a lack of necessary infrastructure and GPU capacity for advanced AI development. Instead of pursuing frontier model creation, Tareq advocates for a curriculum that emphasizes strong programming fundamentals, analytical thinking, and emotional intelligence. "Our current AI syllabuses are lagging 15 to 20 years behind global standards," he observed, attributing this lag to the slow pace of regulatory approval processes in public institutions.

“

The main problem we see is people simply copy-pasting AI answers as absolute truths. Instead, you should challenge the AI using your own research and knowledge to get more accurate, filtered results.

Universities, Tareq argues, must move beyond constantly chasing new AI trends. "We should not just chase AI trends. We must focus on strong fundamentals in coding and thinking. Students also need critical thinking and high emotional intelligence (EQ)," he recommended. He believes that while AI can be used to build agents or software, these skills will likely be considered "low-end" in the future. The ability to critically engage with AI, understand its limitations, and integrate its outputs with human expertise will be paramount for success.

“

Our current AI syllabuses are lagging 15 to 20 years behind global standards.