# What three years of watching AI in production taught us

DevFeed: [What three years of watching AI in production taught us](<https://devfeed.tech/articles/what-three-years-of-watching-ai-in-production-taught-us-31111.md>)

Original publisher: [Read original article](<https://www.mintlify.com/blog/why-we-joined-mintlify>)

Author: Cole Gottdank

Published: 2026-03-11T00:00:00Z

Content type: opinion

Language: en

Sources: [Mintlify Blog](<https://devfeed.tech/sources/mintlify-blog.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [observability](<https://devfeed.tech/topics/observability.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Documentation](<https://devfeed.tech/topics/documentation.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [docs](<https://devfeed.tech/tags/docs.md>), [observability](<https://devfeed.tech/tags/observability.md>), [production](<https://devfeed.tech/tags/production.md>), [software](<https://devfeed.tech/tags/software.md>)

## AI overview

Helicone's founders reflect on three years of observing AI systems in production and argue that the knowledge layer--the documentation and context supplied to models--can matter as much as model quality. They explain how outdated product information can cause AI systems to provide confidently incorrect answers, and connect this insight to their decision to join Mintlify.

## Source excerpt

Learn what three years of watching AI in production taught us about the knowledge layer and why that realization led us to join Mintlify.