Inherent Blog

Retrieval, memory, and inference in production.

Practical writing on the systems that make AI agents dependable — for the people who decide and the people who build.

57 posts

Start here

2026-05-247 min readAI, ExplainedPrimer

What Is AI in Business? A Plain-English Guide

Artificial intelligence in business means software that helps predict, classify, generate, or act. Learn the basics before buying AI tools.

2026-08-186 min readInference & CostPrimer

Inference Engineering for CEOs: Train or Rent?

Should you train and serve your own AI model, or pay per token via an API? A GPU cost breakdown, a breakeven formula, and the exact ICP for each path.

2026-08-1710 min readInference & CostDeep dive

Disaggregated Inference: When the KV Cache Has to Cross the Network

Splitting prefill and decode onto separate GPU pools is now production, not research. It also turns the KV cache into a network problem — here's the tradeoff.

2026-08-138 min readInference & CostDeep dive

Inference Engineering: TTFT, TPOT, and Two Clocks

TTFT and TPOT measure two different things in LLM inference. Optimize the wrong one for your workload and you fix nothing. Here's the engineer's map.

2026-08-0710 min readInference & CostDeep dive

The KV Cache Is Your Concurrency Ceiling

Model weights are a fixed cost. The KV cache is the variable one, and its arithmetic decides how many users one GPU can serve at the same time.

2026-08-0613 min readInference & CostDeep dive

Prefill and Decode: Why One Model Has Two Bottlenecks

LLM inference runs in two phases with opposite limits — prefill is compute-bound, decode is memory-bound. Where they collide is where your latency spikes.

2026-08-0114 min readInference & CostDeep dive

Continuous Batching Is How One GPU Serves a Crowd

Static batching leaves your GPU idle waiting for the slowest request. Continuous batching swaps requests in and out every step — up to 23x more throughput.

2026-07-2314 min readInference & CostDeep dive

Speculative Decoding: Faster Tokens, Identical Answers

A small draft model guesses tokens, a big model verifies them — 2-3x faster LLM generation with mathematically identical output. Here's the catch.

2026-07-219 min readInference & CostDeep dive

The Serving Layer Sets Your RAG Latency Floor

vLLM, TensorRT-LLM, and SGLang all fix the same problem — GPU throughput. None of them fix a RAG system that answers fast from the wrong context.

Inherent on Substack

Keep yourself updated on the latest in AI news and trends.

Everything you need to know about AI, delivered to your inbox. Every week.

2026-07-1816 min readRetrieval & RAGApplied

Retrieval, memory, and inference in production.

What Is AI in Business? A Plain-English Guide

Inference Engineering for CEOs: Train or Rent?

Disaggregated Inference: When the KV Cache Has to Cross the Network

Inference Engineering: TTFT, TPOT, and Two Clocks

The KV Cache Is Your Concurrency Ceiling

Prefill and Decode: Why One Model Has Two Bottlenecks

Continuous Batching Is How One GPU Serves a Crowd

Speculative Decoding: Faster Tokens, Identical Answers

The Serving Layer Sets Your RAG Latency Floor

Keep yourself updated on the latest in AI news and trends.

Long Context Won't Replace RAG Retrieval

Quantization Is the Cheapest Way to Cut Your Inference Bill

Batch Ingestion Is Why Your RAG Answers Go Stale

Model Routing and Fallbacks Without Silent Degradation

RAG Prompt Injection: Your Corpus Is the Attack Surface

LLM Observability for RAG: Trace the Regression

LLM Observability: A Trace You Can't Replay Is a Screenshot

Prompt Caching Cuts Cost. Semantic Caching Cuts Corners.

RAG Evaluation: Prove a Retrieval Change Actually Helped

AI Compliance Starts at Ingestion

KV Caching Is Not Deterministic Retrieval

Why We Built Inherent: The Production Memory Problem

Inherent vs pgvector: Extension You Run or Pipeline You Buy?

Inherent vs Pinecone: Vector Database or Context Layer?

The 30-Minute AI Context Audit for Leaders

Introduce a Company Context Graph Without the Jargon

Vector Database and Retrieval: What Matters Operationally

Knowledge Management AI for Teams That Need Reliable Answers

AI Context in Live Operations: What Must Stay Live

The Continuous Alignment Loop for AI Agents

Agent Memory Without Risk: A Plain-English Guide

AI Memory in Operations: What Should the System Remember?

Enterprise Search vs AI Search for Ops Teams

AI Context: The Missing Layer Behind a Useful Answer

Data Ingestion Pipeline Basics for Operations Leaders

AI Operations: Why Good Models Still Fail in Live Workflows

Context Engineering Is Your AI Strategy: A CEO Playbook

Enterprise AI Without the Hype: Lessons From Jet Engines

Vector Search and Vector Database Basics for CEOs

Semantic Search for CEOs: When AI Search Can Be Trusted

RAG Pipeline: A CEO Guide to Reliable AI Answers

RAG vs MCP: A Practical Guide

RAG Architecture Tradeoffs in Plain English

AI PRD Basics: Writing Requirements for Wrong AI Features

AI Agent Architecture for Product Managers

AI Readiness Assessment: Is Your Business Ready to Scale AI?

RAG for CEOs

Knowledge Base AI: Why CEOs Should Care

What Is Context Engineering? A CEO-Level Guide [2026]

Organize Company Knowledge for AI

What Is Agentic AI?

AI Agents vs Workflows

Enterprise AI Without the Hype

RAG Ingestion Pipeline: A Field Guide for FDEs [2026]

What Is a Context Engine for AI Agents? [2026]

The AI-Ready Knowledge Base Checklist for Better PRDs

What is a Knowledge Base in the AI Agent World?

Production RAG Needs Truth and Memory