5 days ago
4 Tips For Companies Building Internal AI Tools
Startup founder building custom internal applications with AI coding application.
The AI video editing startup, Detail, ditched an expensive software-as-a-service subscription for a custom application that their founder created using AI coding and automation tools for their customer service chatbot. 'In about a week of work, we've replaced Intercom with our own homegrown customer support tools,' Paul Veugen, founder and CEO of Detail, wrote in a LinkedIn post.
While this is not widespread yet, Detail is not alone. There is an emerging trend among startups that are building custom applications using AI coding tools like Replit and V0, often dubbed 'vibe coding.' While cost savings are one variable, usually they are building these custom applications when standard SaaS tools don't fully solve painful problems. One of the biggest differences in how these tools are being created is that they are frequently created and maintained by non-technical founders and employees. Previously, the only options for internal tools were to pay for SaaS products, or pay a lot of money for a company to make a custom tool for you.
In an interview with me, Veugen clarified his reasoning for creating a custom application. It was about creating a customer service experience that worked well with the nature of his customers, their needs, and for how his startup operates.
'I cannot give twenty-four seven worldwide support with a team of 10 people. Right? It's not gonna work,' he said. 'If you do human chat in particular, there's expectations about the interaction that are absolutely suboptimal.'
So Veugen spent a day building an app that connects to Detail's public customer service articles and YouTube videos so their chatbot can answer customer questions with written and video directions at any time.
Brannon Skillern, a fractional Chief People Officer with no previous engineering experience, built a timesheet app for her agricultural tech startup client. It saves 10 -12 hours a month for herself and the CFO and has increased the accuracy of the time recorded. Due to the nature of the manual labor their research and development team does, how people tracked their time varied. But it didn't make financial sense for the startup to shell out for a SaaS product at this phase.
'I could have forced a new tool, and it just would have been paying a bunch of money and I would have low compliance, and I would still have to manipulate the data in several ways to get it to what I need it to be,' Skillern told me in an interview.
Even businesses not directly in tech are beginning to create their own custom tools that didn't exist before. For example, Cyndi Coon, founder and CEO of Laboratory5 Inc, built a research workbook application for their research workshops. 'We don't have IT, or stuff that larger companies have that allows them to test run stuff. We don't have the bandwidth as a team. Now it's amazing the power that we have in the hands of [our]
people,' Coon shared with me.
Previously, they would use surveys or forms, which she described as limiting for the level of creativity required of research participants.
While some companies are saving time and money with these applications, what makes them successful isn't solely those reasons.
There are four keys to success in building your own custom applications with AI coding applications:
As AI adoption increases, we will likely see more businesses experimenting with creating their own custom applications with vibe coding. With all new things, it is some level of risk. Businesses still need to consider legal and privacy implications of the apps they are creating. And some businesses may push the capabilities of these AI coding tools to their limits. It's important for any business to remember what they are really solving for, assess any risks, and consider paying for help or leveraging an existing tool.
However, the potential rewards outweigh these manageable risks for some companies. For startups and small businesses operating with limited resources and unique needs, the ability to create exactly what you need, represents a shift in how businesses can operate. The key is approaching custom application development thoughtfully: start with clearly defined problems, build simple solutions first, and iterate based on real usage. For startups willing to experiment, this technology offers the opportunity to move from being constrained by what exists to being empowered by what's possible.
