---
title: "Discussion on \"When Personalisation Costs 15 Seconds: Redesigning an LLM Feature for Reliability\" | Hashnode"
description: "Discussion on \"When Personalisation Costs 15 Seconds: Redesigning an LLM Feature for Reliability\". Taking the LLM off the Critical Path: Parlato’s Cache-First Redesign"
url: "https://hashnode.com/posts/redesigning-an-llm-feature-for-reliability/6abaddbf7b74fb3693c520a7"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [When Personalisation Costs 15 Seconds: Redesigning an LLM Feature for Reliability](https://gonzalosimon.hashnode.dev/redesigning-an-llm-feature-for-reliability?utm_source=hashnode&utm_medium=feed)

Taking the LLM off the Critical Path: Parlato’s Cache-First Redesign

Sep 28 · 12 min read · gonzalosimon.hashnode.dev  
Read the article

-   [#typescript](https://hashnode.com/tag/typescript)
-   [#postgresql](https://hashnode.com/tag/postgresql)
-   [#nextjs](https://hashnode.com/tag/nextjs)
-   [#performance](https://hashnode.com/tag/performance)
-   [#optimization](https://hashnode.com/tag/optimization)

## Written by

[Gonzalo Aguilar](https://hashnode.com/@gonzalosimon)

@gonzalosimon · Cordoba, Argentina

  
[View profile](https://hashnode.com/@gonzalosimon)

Software Engineer

Full-stack engineer with 6+ years building scalable backend systems and web platforms. Specializes in database optimization, API architecture, and performance engineering with TypeScript, Python, and Node.js.

## Responses 2

Markdown supported · ⌘↵ to post

[Evelyn Wyatt](https://hashnode.com/@Evelynwyatt)

[Sep 29](https://hashnode.com/posts/redesigning-an-llm-feature-for-reliability/6abaddbf7b74fb3693c520a7/comment/6abbb1aecbd61aa17ea053ab)

The separation between core functionality and optional generation is the strongest part of this redesign. At IT Path Solutions, we’ve found that removing an LLM from the critical path is only half the reliability problem; the fallback still needs a clear product contract. If curated and generated content differ in difficulty, ambiguity, or expected behavior, the system can remain technically available while becoming inconsistent for users. Defining what the fallback must guarantee makes the architecture much stronger: the background model can improve the experience when it succeeds, but it should never change the guarantees of the core interaction. That turns AI into a true enrichment layer rather than a hidden dependency.

[Brian · AI News](https://hashnode.com/@brianainews)

[Sep 29](https://hashnode.com/posts/redesigning-an-llm-feature-for-reliability/6abaddbf7b74fb3693c520a7/comment/6abb537380fb4ffeb7532236)

Tool schemas that say almost anything get almost anything back. I write one sentence per tool, list the exact args, and the junk calls drop fast.
