AI tools
The AI tools we build with, and how we choose
There is no single best AI tool. There is the right one for the job. Here is the stack we build on for small and mid-sized businesses, and the honest trade-offs between them.
Most businesses do not have a tool problem. They have a choosing-and-connecting problem. The market is full of capable AI models and automation platforms, and the hard part is picking the right ones for your workflow and wiring them together so they actually run.
We stay deliberately model-agnostic and platform-agnostic. We build on the tools below, pick the best fit per job, and keep that choice swappable so you are never locked into one vendor. Here is how we think about each layer.
AI models
The large language models that do the reasoning, writing, and understanding. We build with OpenAI (ChatGPT) and Anthropic (Claude) and stay model-agnostic, picking the right one per task.
Automation platforms
The engines that run multi-step workflows across your tools. We default to n8n for its economics and power, and use Make or Zapier where they fit better.
CRM and business apps
The systems you already run, like HubSpot, kept in sync and extended with AI so your CRM does the busywork instead of your team.
Custom agents and infrastructure
When off-the-shelf tools are not enough, we build custom agents on Cloudflare primitives, with the observability and guardrails production work needs.
Tool guides
Plain-language guides to the tools businesses ask us about most.
What we build with n8n (n8n automation agency)
Why n8n is our default automation platform for client builds, and the workflows we ship with it.
Read the guideHubSpot AI automation and integration
Automating and integrating HubSpot so your CRM does the busywork instead of your team.
Read the guideCustom GPTs vs custom AI agents: which does your business need?
The real difference between a custom GPT and a custom AI agent, and how to choose.
Read the guideClaude vs ChatGPT for business: which should you deploy?
How we choose between Claude and ChatGPT (and why we use both) when building for clients.
Read the guideHow we choose your stack
We do not start from a tool. We start from the work you want done, then pick the model and platform that fit, and design the build so the choice stays swappable. A well-designed workflow on a good-enough tool beats a sloppy one on the best tool, every time.
That decision happens in a fixed-fee automation audit, which maps your workflows and recommends the right tools and the highest-payback place to start. See how we price it.
See what is worth automating first
Book a call and we will map where your hours leak, then rank the fixes by payback.