PastAGI

Practical AI guides, honest tool reviews, engineering deep dives, real-world use cases, and sharp analysis that cuts through the hype.

Recent

ChatGPT Business Premium pricing decoded through token math, revealing how premium seat costs reflect real agent consumption economics.

News • Aug 11, 2026 •11 min read

ChatGPT Business Premium Pricing Decodes Agent Token Math

ChatGPT Business Premium pricing at $125 reveals the real cost of agentic AI. Reverse-engineer the token math to set your own agent price floor.

#ai-agents #token-economics +4

Local AI inference setup testing a compressed 30B multimodal model on consumer hardware within tight VRAM limits.

Tools • Aug 10, 2026 •12 min read

Muse Glimmer Local Review Tests 30B Agents Under 20GB

This Muse Glimmer local review analyzes how 30B models compressed under 20GB VRAM trade latency, tool accuracy, and context for local agent workflows.

#muse-glimmer-local-review #local-ai-agent-models +6

Structured output enforcement with local LLMs balances JSON schema compliance against throughput and production reliability.

Guides • Aug 09, 2026 •12 min read

Structured Output Local LLM Tactics That Survive Production

Structured output local LLM enforcement means choosing JSON mode, grammar decoding, or tool-calling. Each trades latency, throughput, and reliability.

#structured-output #local-llm +4

Security team conducting an AI agent cyber security evaluation to identify potential system vulnerabilities.

News • Aug 08, 2026 •9 min read

AI Agent Cyber Security Evaluation After the Astra Slowdown

AI agent cyber security evaluation matters now. OpenAI Astra hit a critical cybersecurity threshold. Learn what this gate means for agent deployments.

#ai-agent-cyber-security-evaluation #autonomous-offensive-cyber-capability +6

Agentic loop token costs compound across turns as static payloads and conversation history are re-sent on every API call.

Engineering • Aug 07, 2026 •12 min read

Agentic Loop Token Costs Are an Architecture Problem

Agentic loop token costs come from context accumulation, tool bloat, and retries. Learn to map hidden API waste before downgrading your model.

#agentic-loop-token-costs #llm-api-token-optimization +6

Megakernels in LLM inference fuse separate GPU kernel launches into one persistent execution context to eliminate host overhead during low-latency token decoding.

Engineering • Aug 06, 2026 •10 min read

Megakernels in LLM Inference When Fusion Actually Wins

Megakernels in LLM inference trade off launch overhead against SM occupancy. Learn when fused kernels beat CUDA graphs for low-latency agentic workloads.

#megakernels-in-llm-inference #fused-kernels-vs-cuda-graphs +6

Stacked pull requests decompose large AI-generated diffs into a chain of smaller linked branches for incremental code review.

Use Cases • Aug 05, 2026 •13 min read

Stacked Pull Requests Relocate AI Mega-PR Review Cost

Stacked pull requests relocate AI mega-PR cost into rebase cascades and multiplied CI runs. They win above a measurable threshold and lose below it.

#stacked-pull-requests #ai-code-review +3

Prefill decode disaggregation isolates compute phases across separate GPU pools in a datacenter to optimize large language model serving.

Engineering • Aug 03, 2026 •11 min read

Disaggregated GPU Inference Hits the KV Cache Wall

Disaggregated GPU inference splits prefill and decode for higher throughput, but each request moves 2.6 GB of KV cache across the datacenter.

#llm-serving #gpu-inference +3

Document mediated prompt injection turns shared document stores into self-propagating attack surfaces when an AI agent both reads from and writes to the same files.

News • Aug 01, 2026 •11 min read

How The Copilot Prompt Injection Worm Spreads

The Copilot prompt injection worm proves prompt injection can self-propagate through shared documents. Learn why AI security fails and how to adapt.

#ai-security #prompt-injection +3

LLM-native recommendation architecture using generative language models to replace hand-crafted feature pipelines in production recommendation systems.

Use Cases • Jul 31, 2026 •13 min read

LLM-Native Recommendation Architecture After Netflix GenRec

Netflix's GenRec replaces thousands of hand-crafted ML features with an LLM-native recommendation architecture. Explore the engineering trade-offs for builders.

#llm-architecture #recommendation-systems +4

MCP servers for AI agents provide a standard protocol that bridges fragmented operational data silos without per-system custom connectors.

Use Cases • Jul 30, 2026 •13 min read

How MCP Servers for AI Agents Bridge Fragmented Data

MCP servers for AI agents dissolve fragmented operational data silos, shifting the bottleneck from integration to query planning and context engineering.

#mcp-servers-for-ai-agents #model-context-protocol +6

Controlling reasoning effort in LLM APIs means setting explicit parameters that cap how many tokens a model spends on internal chain-of-thought before producing visible output.

Guides • Jul 28, 2026 •13 min read

Control Reasoning Effort LLM APIs in Production

Control reasoning effort LLM APIs across OpenAI, DeepSeek, and Anthropic. Practical routing rules to cut cost and latency without losing accuracy.

#llm-engineering #api-optimization +3