run a 2-person custom dev agency and we just shipped our own saas, so ai is basically our co-founder at this point.
i currently maintain active subscriptions for claude pro and google ai pro. honestly, the roi is completely undeniable when you factor in how much time it saves us on boilerplate code, debugging, and drafting initial architectures. we also sprinkle in some midjourney and elevenlabs for asset generation when we need it.
for a small team trying to ship fast, the $40-$60/mo we spend on core llms is by far the cheapest overhead we have.
I was actually facing a huge friction problem when it came to closing clients—both at my agency and in a wholesale business I run with my family—because the whole quoting and payment process was really outdated. So, I built a software tool to eliminate that friction; it’s like a live Notion-style interface where the salesperson and client can update the quote and communicate via a public link. That same link allows for approval and payment—using cards or bank transfers, obviously via the Stripe API—and handles everything related to invoicing and sales. It ended up being really useful for my agency, too, allowing me to send clients more professional quotes with a detailed project breakdown so they can clearly see what the work entails and understand the pricing. Honestly, I’ve definitely noticed an improvement in sales.
There is a massive competitive advantage right now in building 'boring' infrastructure manually instead of rushing to wrap an AI API.
While the trend is to build a conversational interface for everything, I've found that real businesses are still bleeding time on core sales friction. For example, moving away from static PDFs to interactive live sales quotations requires rock-solid, hand-coded database architecture to handle real-time pricing, currency locking, and instant approvals. You simply can't prompt an LLM to reliably handle that level of strict state management.
Taking the time to write the core logic yourself makes the system predictable. Businesses don't pay for the AI novelty; they pay for seamless operational execution
i currently maintain active subscriptions for claude pro and google ai pro. honestly, the roi is completely undeniable when you factor in how much time it saves us on boilerplate code, debugging, and drafting initial architectures. we also sprinkle in some midjourney and elevenlabs for asset generation when we need it.
for a small team trying to ship fast, the $40-$60/mo we spend on core llms is by far the cheapest overhead we have.