Emergent AI on why small businesses should build their own software

There was a time when only the biggest companies could afford software built just for them. Custom builds started at around $20,000, going up to $100,000, and would take the better part of a year to ship. Everyone else would make do with the generic tools on the market.

Emergent AI wants to change that. Founded by brother who have been building things together as kids, the platform lets anyone describe a workflow and have an AI agent build it, all with no coding required.

We spoke to Mukund Jha, Co-Founder and CEO, Emergent, about why business software keeps failing small companies, how to know when to build instead of buy, and why the person who understands the problem best might not need a developer at all.

Can you tell us about your background and what led you to build Emergent?

My brother Madhav and I have been building things together since we were children. We once asked our dad for a computer game, and he gave us a C++ CD and told us to make our own. That curiosity never left us and continues to push us today.

Later, while I was scaling engineering at Dunzo, one of India’s largest hyperlocal commerce companies, I saw how much skilled engineering time was being swallowed by testing. When AI began accelerating, our original idea was to automate that bottleneck.

During Y Combinator, we decided to tackle a much bigger problem: could we build an agent capable of creating complete software? That’s how Emergent AI was born. We wanted to give people without a technical background the ability to turn an idea or operational problem into working software with just a few keystrokes.

What issues is Emergent solving that no one thought about before?

It’s actually very interesting – when we first launched, we expected to see people replacing SaaS subscriptions or moving away from spreadsheets to automate certain operational activities. Some, of course, did, but our most recent research shows one in three of the 300 operators had built something that had never previously existed for them.

You see, their problem was often too specific, their market was too small or the commercial opportunity was not attractive enough for a traditional software vendor. Custom development was also beyond their budget. They had simply carried on without the right tools. But AI has shifted the balance, and now the person closest to the problem can describe what they need and build it themselves. Software can be created because one business needs it, rather than because a vendor believes thousands of companies will buy it.

Why has so much business software failed to meet the needs of startups and smaller companies?

Traditional software companies tend to move upmarket because enterprise customers can support larger contracts and higher acquisition costs. Smaller businesses are left with generic products designed to serve the broadest possible audience, and they just have to make do. That often forces founders to alter their processes around the software. They add spreadsheets, manual workarounds and disconnected tools to fill the gaps. For a small team, that can quickly become an administrative burden.

For years, custom software meant hiring engineers or paying an agency, followed by months of development and ongoing maintenance. Just imagine, the median price for a basic build was $20,000! Some had been quoted as much as $100,000, with timelines stretching beyond a year.

Most small businesses cannot justify that cost, particularly when they are trying to solve a focused operational problem. So, AI makes it economically possible to serve much narrower requirements. For instance, a construction firm, independent healthcare provider, or a poultry farm may each need completely different software. Those industries might never justify a dedicated SaaS product, but the business can now create a tool that fits how it already works.

How should a startup decide whether to buy software or build its own?

A startup should buy when an existing product genuinely fits the way the business works. If founders start bending their processes around generic software, adding workarounds or accepting that a tool only solves part of the job, then the software is not right for them.

They should therefore consider building their own tools. Founders often understand their operational problems better than anyone else because they have been handling them for years. They do not need to know how to code; they need to be able to describe the workflow clearly enough to create something around it.

However, they should also test whether software reliably removes work from the team. If it requires constant supervision or becomes another system to manage, it has failed regardless of whether it was bought or built.

What mistakes are startups making when adopting AI?

The biggest mistake is chasing every new tool without being clear about the work it should remove. That leaves a small team with another subscription or system that they need to learn to use.

Founders can also place too much weight on how polished an AI output appears. Reliability is a better test. Does the system complete the task consistently? Does it require constant supervision? Has it genuinely reduced the amount of work sitting with the founder or the wider team?

See Also

AI is making failure affordable

AI should give founders time back for decisions that still require their judgement: setting direction, building partnerships, understanding customers and deciding where the company goes next. If it creates more supervision than it removes, the use case needs to be reconsidered.

AI adoption among UK businesses is growing quickly, yet many still struggle to identify a clear use case. Where should founders begin?

British Chambers of Commerce research found that 54% of UK firms were actively using AI by March 2026, up from 35% in 2025. At the same time, 71% remained unable to identify a clear use case, while 60% cited limited AI skills as a barrier.

With that in mind, founders do not need to begin with a broad AI strategy. They should look at where time is being lost inside the business. That could be customer follow-ups, stock management, scheduling, reporting or moving information between different systems. If you think about it, the strongest use cases often come from tasks the founder already understands in detail because they have been doing them for years. Knowing the workflow and understanding where it breaks is now more valuable than knowing how to code, because you can get straight to fixing it.

What is next for Emergent, and how do you see the platform developing?

Our mission is to help entrepreneurs and small and medium-sized businesses truly scale. Around 90% of businesses globally are SMEs, and they account for more than half of employment, yet they have often been poorly served by custom technology.

We want Emergent to support the full journey from identifying a problem and building an application through to running the software once it is live. That includes the infrastructure behind it and agents that can take responsibility for ongoing work.

Our autonomous AI Agent Wingman is part of that next stage. It applies the same agentic engineering foundation to activities such as finance, scheduling, lead generation and social media management. The aim is to help founders build the software their company needs and then use AI to reduce the operational load of running it.