Blog — AI Agents, MCP Tools & Build Logs by Amit Kumar
TECHNICAL WRITING ON PRODUCTION AI AGENTS, MCP TOOLS, SELF-HOSTED AUTOMATIONS, MULTI-AGENT WORKFLOWS, AND PRODUCT SHIPPING.
New to self-hosting? Start with the complete guide →
Latest
- How I Cut AI Agent Memory Usage by 70% in 3 Weeks Using These 3 Techniques09-10-2026
How I reduced AI agent memory usage by 70% through dynamic context management, tool call caching, and gradient checkpointing - saving $87,880 annually while improving performance.
- Brotli Test09-10-2026
A test for Brotli handling
- Debug Test09-10-2026
A debug test
- Minimal Test09-10-2026
A minimal test
- How I Cut AI Agent Memory Usage by 70% in 3 Weeks Using These 3 Techniques09-10-2026
How I reduced AI agent memory usage by 70% through dynamic context management, tool call caching, and gradient checkpointing - saving $87,880 annually while improving performance.
- How I Cut AI Agent Memory Usage by 70% in 3 Weeks Using These 3 Techniques09-10-2026
How I reduced AI agent memory usage by 70% through dynamic context management, tool call caching, and gradient checkpointing - saving $87,880 annually while improving performance.
Multi-agent systems
Agent frameworks
Self-hosting & infrastructure
- How to Build Your Own MCP Server (in About 90 Minutes)18-08-2026
- AI Agent Budget Guardrails: How I Stop Runaway Loops From Draining the Budget31-07-2026
- Self-Hosting AI Agents: Why I Moved Off APIs (and the Stack I Run)13-07-2026
- Best VPS for Self-Hosted AI Agents in 2026 (Hetzner, Hostinger, DigitalOcean Compared)09-07-2026
- How to Deploy Hermes Agent on Hetzner — A Complete 2026 Walkthrough09-07-2026
- Hostinger VPS for Self-Hosted AI Agents — What Works in 202609-07-2026
- How to Deploy Hermes Agents to a VPS the Right Way: A Production Playbook (Under $10/month, 9 Phases, Zero Public Ports)26-05-2026
Shipping to production
- How I Cut AI Agent Memory Usage by 70% in 3 Weeks Using These 3 Techniques09-10-2026
- How I Cut AI Agent Memory Usage by 70% in 3 Weeks Using These 3 Techniques09-10-2026
- Simple Test09-10-2026
- Test Post for Memory Optimization09-10-2026
- Test Post for Memory Optimization09-10-2026
- Test Post for Memory Optimization09-10-2026
- AI Agent Observability: Stop Flying Blind in Production21-08-2026
- Context Engineering: The Discipline That Makes AI Agents Actually Reliable14-08-2026
- AI Agent Failures: Model or Harness? How to Localize the Real Fault07-08-2026
- AI Agent Tool-Call Verification: How to Prove Your Agent Actually Did the Work04-08-2026
- How to Actually Test AI Agents Before They Break in Production18-07-2026
- Prompt Injection Is the New SQL Injection: How I Harden My AI Agents14-07-2026
- Why Your AI Agent Can't Use Tools Safely (And How MCP Fixes It)23-06-2026
- Beyond the Context Window: How AI Agent Memory Systems Actually Fail in Production22-06-2026
- Why Your AI Agent Pilot Never Makes It to Production (And How to Fix It)19-04-2026
- How to Build Enterprise-Grade, Production-Ready AI Agents26-03-2026
More writing
- How I Cut AI Agent Memory Usage by 70% in 3 Weeks Using These 3 Techniques09-10-2026
- How I Cut AI Agent Memory Usage by 70% in 3 Weeks Using These 3 Techniques09-10-2026
- The Real Cost of Building an AI Agent That Actually Works: Lessons from 3 Failed Projects06-10-2026
- Building Transparent Reasoning Traces in AI Agents: A Production Guide29-09-2026
- How I Evaluated 7 AI Agent Frameworks (and What Actually Matters)22-09-2026
- How to Build AI Agents That Get Cited in AI Search: A Practical Guide11-09-2026
- When to Use AI Agents (and When a Plain Script Beats Them)25-08-2026
- How to Get Cited by AI Search (Not Just Ranked by Google) — A Builder's Guide to GEO in 202621-07-2026
Read on Substack
Get the next build note before it becomes a blog post.
Founder notes, product experiments, and practical AI systems breakdowns from the workbench.