HostingerSimon Lim8 min readintermediate
What four people at Hostinger actually do with AI all day (and what happens when you have an agentic beef)
Summary
Hostinger staff use custom AI agents to automate daily tasks—from code reviews to influencer lead sourcing—shifting their work from doing the work to managing the agents. Building reliable “harnesses” (prompt contexts, constraints) consumes most of the engineering effort, and agents still hallucinate, repeat work, or suggest unsafe fixes, so human oversight remains essential.
- AI agents can replace hours‑long manual workflows (e.g., invoice generation, lead sourcing, debugging) with a few clicks, turning engineers into “managers of agents.”
- The bulk of the effort is in building and maintaining the harness around LLMs—prompt engineering, context handling, permission management—to keep agents reliable.
- Hallucinations, duplicate suggestions, and unsafe code fixes are common; teams mitigate by breaking workflows into smaller steps and keeping human validation loops.
- Human expertise is still required for high‑level architecture, nuanced partner evaluation, and creative brainstorming—areas where models lack domain intuition.
Shows a realistic, mid‑scale deployment of LLM‑driven agents in a SaaS company, highlighting both productivity gains and the practical limits of current AI—valuable for teams planning to automate internal workflows.
5/10
