---
title: "Enterprise AI | Dhruv Doshi"
description: "Adoption strategy, production patterns, and operating discipline for AI systems in large organisations."
canonical: https://doshidhruv.com/topics/enterprise-ai/
author: Dhruv Doshi
---

[Skip to content](#main-content)

[Dhruv Doshi](/)

[Work](/projects)[Experience](/resume)[Notes](/notes)[Guides](/guides)[Research](/research)[Investing](/investing)[About](/about)[Contact](/contact)[Search /](/search)dark mode

# Enterprise AI

Adoption strategy, production patterns, and operating discipline for AI systems in large organisations.

- ["Agent orchestration patterns: planner-executor, routing, and supervision"](/notes/agent-orchestration-patterns-planner-executor-routing-and-supervision) · Note · Most teams building AI agents start with a single agent loop: a model, a prompt, and a pile of tools. It works for the demo. It falls apart when the task is too long, too varied, or too…
- ["Choosing a vector database: what actually differentiates them"](/notes/choosing-a-vector-database-what-actually-differentiates-them) · Note · The vector database market has a marketing problem: every product claims to be the fastest, most scalable, most accurate option, and the benchmarks they publish are engineered to prove…
- [Sequence enterprise AI adoption by risk, not by hype](/notes/sequence-enterprise-ai-adoption-by-risk-not-by-hype) · Note · Most enterprise AI programs fail in one of two ways: they start with the highest risk use case and stall in governance review, or they scatter dozens of disconnected pilots that never…
- ["Fine-tuning vs RAG vs prompting: how to actually decide"](/notes/fine-tuning-vs-rag-vs-prompting-how-to-actually-decide) · Note · Teams building on LLMs face a recurring decision: should we fine tune a model, build retrieval augmented generation, or just engineer better prompts? The discourse around this is noisy…
- ["FinOps for AI workloads: controlling GPU and inference spend"](/notes/finops-for-ai-workloads-controlling-gpu-and-inference-spend) · Note · AI workloads have a cost profile unlike anything else in the enterprise cloud bill: GPU compute that is expensive per hour and often poorly utilized, inference costs that scale with…
- ["Shadow AI: discovering and governing unsanctioned AI use"](/notes/shadow-ai-discovering-and-governing-unsanctioned-ai-use) · Note · Shadow AI is the use of AI tools and services without IT or security approval: employees pasting customer data into public chatbots, teams building workflows on unvetted AI APIs,…
- [When to bring in enterprise AI consulting](/notes/when-to-bring-in-enterprise-ai-consulting) · Note · Most organizations do not need an AI consultant. They need a clear problem, an empowered owner, and the patience to run the sequencing that works: low risk internal wins first, the…

**Dhruv Doshi** · Toronto, Canada · [work@doshidhruv.com](mailto:work@doshidhruv.com)

[Resume](/resume)[Notes](/notes)[Guides](/guides)[Topics](/topics)[Search](/search)[Research](/research)[Investing](/investing)[LinkedIn](https://www.linkedin.com/in/dhruvdoshi25071999)[GitHub](https://github.com/DhruvDoshi)

[Sitemap](/sitemap.xml)[RSS](/feed.xml)[LLMs](/llms.txt)

© 2026 Dhruv Doshi