BOOK A CALL →
Stack

Tools of the trade

Tools don't fix operations — but the right tool, chosen for the actual constraint, keeps a system cheap, reliable, and maintainable. Here's what I reach for and when. For tool-specific engagements, see the n8n, Make.com, Zapier, and GoHighLevel pages.

LLMs

Claude by default; whatever the constraint demands in practice.

Claude (primary)GPT (OpenAI)GeminiLlama (self-hosted)
Agent tooling

For building agents that plan, use tools, and get evaluated.

Claude CodeMCPOpenAI SDKLangGraph
Automation platforms

Chosen per workflow — volume, complexity, and who maintains it.

n8nMake.comZapierGoHighLevelTemporal
Runtimes & hosting

Serverless-first; long-running work goes where it belongs.

VercelCloudflare WorkersAWS LambdaModalTrigger.dev
Data & retrieval

Boring, proven databases. Vector search where retrieval earns it.

PostgresSupabasepgvectorPineconeSnowflake
Web

Fast, indexable sites wired into the systems behind them.

Next.jsReactTypeScriptTailwind
Selection

How tools get picked

  • Volume and cost curves first — per-task pricing changes the answer at scale
  • Who maintains it: the best tool is one your team can actually operate
  • Failure behavior: silent failure disqualifies a tool for critical paths
  • Data constraints: residency and compliance can force self-hosting
  • Boring beats novel — proven infrastructure wins ties every time
$ erick --find-bottleneck 

Talk about your stack

Bring your current tool list to a call — I'll tell you what to keep, what to connect, and what's quietly costing you.

30 minutes · no pitch deck · reply within 24h if you write instead

Book a call →See live demos