Every software development company seems to be “AI-powered” now. The term is on websites, in pitch decks, in sales calls. Most of the time it means very little.
I want to explain what it actually means in my practice, and give you a way to evaluate any development company’s AI claims against reality.
What “AI-Powered” Actually Should Mean
At minimum, a development company using AI tools effectively should:
Use AI coding assistants in daily work. Tools like Cursor, GitHub Copilot, or Amazon CodeWhisperer integrated into the actual development workflow — not just mentioned in marketing.
Be able to build AI-integrated products for clients. Can they build a system that connects to OpenAI, Anthropic, or other AI APIs to add intelligent capabilities to your software?
Use AI for workflow automation. Can they help automate your business processes using AI-enhanced tooling?
If the answer to any of these is no, “AI-powered” is a marketing term, not a capability description.
How I Use AI in My Practice
I’ll be direct about what this looks like in my work:
Cursor as my primary IDE. Cursor is an AI-native development environment built on VS Code. It has deep integration with Claude and GPT-4 for inline code generation, explanation, and review. I use this every day.
Claude for architecture conversations. When I’m working through a complex design decision, I describe the problem and constraints to Claude and use the response to stress-test my thinking. This isn’t outsourcing the decision — it’s getting a second perspective before committing.
AI for building AI-integrated products. I build software that uses AI APIs to add capabilities: natural language search, document analysis, automated categorization, intelligent form processing. This is a service I offer, not just a tool I use.
n8n for AI-enhanced automation. I build automation workflows that incorporate AI for classification, summarization, and routing. A document comes in, AI extracts the relevant data, the workflow routes it appropriately.
What AI Can Build for Your Business
If you’re a mid-market business or manufacturer, here are concrete AI integrations that deliver real value:
Document processing. Purchase orders, invoices, quality reports — AI can extract structured data from unstructured documents with high accuracy. This replaces manual data entry.
Customer communication. AI-assisted email drafting, inquiry routing, and response triage. Not replacing human relationships — removing the administrative burden.
Inventory and demand forecasting. Patterns in your historical data that humans would miss, surfaced by AI analysis.
Natural language reporting. Let operations staff query your data in plain English instead of waiting for a developer to write a new report.
Automated quality control. Image-based defect detection on production lines. Computer vision has gotten remarkably capable.
How to Evaluate an AI-Capable Development Shop
Questions to ask any development company claiming AI capabilities:
“What AI tools do your developers use daily?” Vague answers about “cutting-edge AI” without specifics are a red flag. The answer should name tools.
“Show me an AI integration you’ve built for a client.” If they can’t point to a real system they’ve built, their AI capabilities are theoretical.
“How do you handle AI output quality and reliability in production systems?” Any developer who’s shipped AI-integrated software seriously has thought about this. AI outputs are probabilistic — how do you handle the cases where the AI is wrong?
“What’s your approach to AI vendor lock-in?” A good developer builds AI integrations with abstraction layers so you can switch providers without rewriting everything.
The Honest Limitation
AI makes certain kinds of software faster and better. It doesn’t make all software better. For highly deterministic systems where every output needs to be predictable and auditable, AI integration adds complexity without benefit.
I’m not going to recommend AI integration because it’s trendy. I’m going to recommend it when it solves a real problem faster or better than a traditional approach. Sometimes that’s the right tool. Sometimes it’s not.
If you’re trying to figure out whether AI integration would actually help your business, reach out for a conversation. I’ll give you an honest assessment.