Publication

Graph Praxis

@graph-praxis

articles: 15
total claps: 1,339
top claps: 364
avg read: 11.1 min
paywalled: 14
active: 2026-01-18 → 2026-07-12

Wiki topics

LLM · Large Language Models · 10 RAG · RAG & Retrieval · 10 AGT · AI Agents · 7 PHI · Philosophy · 6 FT · Fine-tuning & Adaptation · 3 BIZ · Business Strategy · 3 ECO · Economy · General · 3 🔧 · Data Engineering · 2

Top tags

#knowledge-graph ·10 #graphrag ·9 #ontology ·5 #agentic-ai ·4 #llm ·4 #robocorpco ·2 #obsidian ·2 #claude-code ·2 #harness-engineering ·2 #local-llm ·2

AI LIF

AL Alexander Shereshevsky · Graph Praxis

Your Obsidian Vault Is a Knowledge Graph. Here’s How to Make It Think (quickly).

Following the yes LLM Wiki / no LLM Wiki debate, I want to share my experience from the last five years of building a second brain in…

Apr 11, 2026 · 12 min read · 364 claps paywalled #knowledge-graph #obsidian #claude-code

Your Obsidian Vault Is a Knowledge Graph. Here’s How to Make It Think (quickly).

TCH AI

AL Alexander Shereshevsky · Graph Praxis

Building Knowledge Graphs with Local LLMs: From Benchmarking to Fine-Tuning

For the production GraphRAG pipeline, the extraction step is where it quietly breaks.

Mar 23, 2026 · 13 min read · 194 claps paywalled #graphrag #local-llm #knowledge-graph

Building Knowledge Graphs with Local LLMs: From Benchmarking to Fine-Tuning

ECO AI SOC

AL Alexander Shereshevsky · Graph Praxis

Multi-LoRA for Knowledge Graphs: One Model, Many Domains

Our extraction model speaks fluent Wikipedia. Our documents speak legal, medical, and financial. Here’s how we fixed that.

Mar 24, 2026 · 14 min read · 163 claps paywalled #graphrag #knowledge-graph #fine-tuning-llm

Multi-LoRA for Knowledge Graphs: One Model, Many Domains

AI HUM

AL Alexander Shereshevsky · Graph Praxis

From Domain Expertise to Autonomous Agents: Introducing the Agentic Graph Framework

How we’re enabling domain experts to transform decades of knowledge into executable AI agents — without writing a single line of code

Jan 18, 2026 · 8 min read · 97 claps paywalled #ontology #knowledge-graph #robocorpco

From Domain Expertise to Autonomous Agents: Introducing the Agentic Graph Framework

AI ECO

AL Alexander Shereshevsky · Graph Praxis

Your Agent Doesn’t Have a Memory Problem. It Has a Recompute Problem.

Why local graph memory is the right primitive in 2026 — and what I learned wiring a minimal implementation into a real agent

Jun 04, 2026 · 10 min read · 92 claps paywalled #harness-engineering #graphrag #memory-management

Your Agent Doesn’t Have a Memory Problem. It Has a Recompute Problem.

ECO HUM MDA AI LIF TCH

AL Alexander Shereshevsky · Graph Praxis

Your Obsidian Vault Is a Knowledge Graph. Here’s How to Make It Think.

I watched Boris Cherny say “coding is solved” and thought of an essay from 1945. That collision — between the creator of Claude Code and a…

May 17, 2026 · 15 min read · 91 claps paywalled #knowledge-management #knowledge-graph #memex

Your Obsidian Vault Is a Knowledge Graph. Here’s How to Make It Think.

AI ECO HUM

AL Alexander Shereshevsky · Graph Praxis

Everyone Is Writing Skills for Their Agents. Almost Nobody Can Say When They’re Complete.

Simon Willison called Skills “a bigger deal than MCP.” He’s right — but a skill file is only as good as the expert knowledge inside it…

Jul 05, 2026 · 8 min read · 79 claps paywalled #knowledge-management #agentic-ai #llm

Everyone Is Writing Skills for Their Agents. Almost Nobody Can Say When They’re Complete.

AI TCH

AL Alexander Shereshevsky · Graph Praxis

The Phase Transition in Your Knowledge Graph: Why Agent Memory Suddenly “Clicks”

Every practitioner has felt it. Your GraphRAG system is useless for weeks — hallucinating, missing obvious connections, and underperforming…

Apr 14, 2026 · 13 min read · 67 claps paywalled #knowledge-graph #llm-wiki #mempalace

The Phase Transition in Your Knowledge Graph: Why Agent Memory Suddenly “Clicks”

HUM

AL Alexander Shereshevsky · Graph Praxis

Three Lies and a Whale

How I tricked my kids into learning to reason — and turned my day job into a detective game. Play it at lieinthegraph.com.

Jun 05, 2026 · 5 min read · 65 claps #knowledge-graph #games #critical-thinking

AI TCH HUM

AL Alexander Shereshevsky · Graph Praxis

Causal Inference on Knowledge Graphs: The Fourth Layer of Context Blindness

Extending ontology-grounded GraphRAG and agent memory from association to intervention and counterfactuals — why agents that act in the…

Apr 25, 2026 · 18 min read · 45 claps paywalled #graphrag #causal-inference #knowledge-graph

Causal Inference on Knowledge Graphs: The Fourth Layer of Context Blindness

AI ECO HUM

AL Alexander Shereshevsky · Graph Praxis

The GraphRAG Cost Cliff: How $33,000 Became $33 in Eighteen Months

In early 2024, indexing a single dataset for GraphRAG cost $33,000. By mid-2025, Microsoft Research had reduced that to 0.1%. The cost…

Mar 07, 2026 · 14 min read · 36 claps paywalled #graphrag #ontology #enterprise-ai

The GraphRAG Cost Cliff: How $33,000 Became $33 in Eighteen Months

ECO MDA TCH

AL Alexander Shereshevsky · Graph Praxis

What Music Recognition Can Teach Anti-Money Laundering

An audio fingerprinting algorithm from 2003 turns out to be an unreasonably good fit for transaction monitoring.

May 06, 2026 · 6 min read · 29 claps paywalled #knowledge-graph #aml #network-analysis

What Music Recognition Can Teach Anti-Money Laundering

AI ECO HUM

AL Alexander Shereshevsky · Graph Praxis

RAG FinOps: A 12-Month Postmortem on Where the Dollars Actually Go

Compute collapsed. Maintenance ate the savings. The five stages of a 2026 Graph-RAG bill, what changed in 12 months, and a budget…

May 02, 2026 · 11 min read · 15 claps paywalled #llmops #graphrag #ontology

RAG FinOps: A 12-Month Postmortem on Where the Dollars Actually Go

AI ECO HUM

AL Alexander Shereshevsky · Graph Praxis

Why Enterprise AI Agents Are Failing

The Hidden Tax on Enterprise Intelligence

Jan 19, 2026 · 9 min read · 1 claps paywalled #ontology #knowledge-graph #robocorpco

AL Alexander Shereshevsky · Graph Praxis

How to Know Your Agent Skill Is Done

An evals harness for skills — because 47,000 of them shipped and almost none have a test suite.

Jul 12, 2026 · 9 min read · 1 claps paywalled #agentic-ai #harness-engineering #llm-evaluation