DAVID AI

How do you know the AI consulting/implementation company you're hiring actually knows what they're doing? I'm genuinely concerned about how many companies are investing in AI agents, custom models, and complex implementations without properly vetting the people they're hiring. It's easier than ever to sound knowledgeable about AI. Anyone can ask Claude to generate a list of discovery questions, audit frameworks, or technical recommendations before a sales call. That doesn't necessarily mean they understand how to architect, deploy, secure, or maintain the systems they're selling. My advice: Interview your AI company. Yesterday, I was in a pre-kickoff meeting where the client essentially interviewed me. They asked detailed questions about middleware, integration architecture, our AI agent harnesses, and why we've chosen NVIDIA and Apple hardware for certain local, on-prem deployments. Here are a few of the technical questions that came up: "When would you choose an Enterprise Service Bus (ESB), an API gateway, or direct service integrations, and why?" "How does middleware manage the request and response lifecycle?" "How do you handle scoped dependencies when middleware is instantiated as a singleton?" "How do you enforce encryption in transit between application and service layers?" "How would your AI agent harness operate in a fully air-gapped warehouse environment?" I loved these questions. They required us to explain our architectural decisions, the tradeoffs involved, and how our systems actually operate. And I think more companies should be doing this. If someone is going to have access to your company's data, internal processes, customer information, and critical infrastructure, you should understand what they're building and why. Ask them to diagram their architecture. Ask what happens when something fails. Ask how they handle security, observability, and maintenance. Ask them why they chose one technology over another. And if the answers don't make sense, keep asking. You don't need to be an engineer to hold an AI implementation company accountable. Please be willing to ask good questions and expect clear, defensible answers. If you're evaluating an AI company and don't know what to ask, I'd be happy to send you my technical vetting checklist. You can use it with any vendor, including us. And if we end up working together, even better :) @getdavidai getdavid.ai