Your Tech Stack: How these execs decide what’s worth paying for

Editor’s note: Welcome to the first installment in Your Tech Stack, an occasional series about, well, your tech stack. How do you build it? Should you go with an all-in-one provider or take a mix-and-match approach? What’s the best bang for your buck?

Those are the types of questions we’ll aim to answer. Have a specific challenge? Send it our way: [email protected].

When teams at the AI accounts receivable startup Daylit consider whether to invest in a new piece of software, they go through a three-step checklist, according to co-founder and CEO Jared Shulman.

Would the software automate or help with an existing documented process? Is there readily accessible data for it to use? And would there be a human in the loop to evaluate the results?

“On the go-to-market side, we’re buyers of lots of different tools to get automation in our go-to-market efforts—Valley is an example of one that’s pulling data from LinkedIn,” Shulman said. “But before we buy that product, we need to know, is there a process in place? What do we do on the go-to-market side to actually reach out to these customers?”

Between a flood of new agentic tools and services, a SaaS industry in flux, and an ever-shifting raft of new AI features, it can be harder than ever for companies to know what’s essential in the modern tech stack. Should you choose best-in-breed or all-in-one? When do you vibe-code and when do you outsource? We asked a few execs how they think about these choices and where the market is headed.

For Saurabh Giri, chief product and technology officer at neo-cloud platform Lightning AI, the choice between a multivendor and bundled provider in traditional SaaS comes down to four factors: speed, simplicity, cost, and control. Some of that will depend on the stage of the company, as well.

“The cost can be less when you go with an all-in-one provider in the near term, but in the long term, usually the cost factor becomes slightly more nuanced,” Giri said. “Once we reached the point where we hit the threshold of [having] the team and the expertise to do things in-house, we then realized that we could save a lot of money by doing many of these things in-house.”

Those are the criteria for more traditional business software, but the rules are changing when it comes to new agentic offerings, he said. Giri recommended avoiding too much lock-in.

“For a lot of enterprises that I talk to, the question is, do you stick with a vendor that is injecting the AI, or do you build your own agentic AI in-house that can make it cheaper?” Giri said. “And that is where it becomes slightly more nuanced, only because the AI perimeters are still evolving. But the long-term answer is still the same: If you have the team to do it, if you have the time to do it, it is best done in-house...You would not go all-in on OpenAI or Gemini or Anthropic. You would try to be vendor-agnostic.”

Rimini Street CIO Joe Locandro tends to favor a best-in-breed strategy when assembling the enterprise software support firm’s tech stack, as long as the platforms are interoperable. Data integrity and reliability, as well as security tools, are the two essential buckets he said are the most valuable parts of the stack.

The company uses API integration platforms like Boomi and Microsoft tools to coordinate between them. But that integration layer is also changing fast with the rise of Anthropic’s Model Context Protocol (MCP), an open-source standard for business tools to communicate with agents, Locandro said.

“An agent that consistently goes into a finance system quickly learns that there’s financial data in there—book value data—but there’s nothing about customer complaints or things like that. So it starts knowing and building up a data dictionary, and it knows where to go to get it after repeatedly going in,” Locandro said. “APIs are still there. They will be there for the time being, but the new technology is eroding some of that advantage as you start doing changes and moving along the path.”

As someone building a tech stack for an audio media company that targets Hispanic users, Nueva Network Chief Innovation Officer Rene Alegria said buying software that fulfills his specific language needs more or less requires a best-in-breed approach. Even new AI-powered translation tends to fall short, he said.

“We get these demos…They’re super impressive…only for us, six months later, to realize that there’s a glitch with the language preference,” Alegria said. “So oftentimes we use best-of-breed because it’s tried and true. We tend to use best-of-breeds for all things close to revenue.”

These days, Alegria said his employees are increasingly vibe-coding their own tools, as well. The company has a tech-themed WhatsApp group where they share different articles for inspiration and ideas for experiments.

“Can we build better for our unique self-interest, for our community?” Alegria said. “We do a lot of vibe-coding, just to make sure we still can, and to keep up with what’s happening.”