Agentic Mornings

Aman Gupta

Agentic Mornings — the 20-minute weekday briefing for people who build with AI. Every weekday morning: the day's biggest AI stories, then one big idea explained properly — from first principles, with the numbers that matter: what it costs, how fast it runs, and how you know it actually works. No hype, no hot takes. News up front, deep dive after — the stuff that matters if you ship AI products, framed the way a practitioner wants it: cost, latency, evals, and the decision you can act on. New episodes every weekday morning.

  1. 8h ago

    Agentic Mornings — November 2, 2026: AI Safety and Guardrails

    One shared robots.txt opt-out, Apple Intelligence in China with Qwen, a shared ChatGPT/Codex allowance — then a deep dive on AI safety and guardrails. AI Generated

  2. 3d ago

    Agentic Mornings — October 30, 2026: AI Observability — Tracing Agent Runs

    Mistral's open-weight Large 4, CrowdStrike's agent swarm, OpenAI's agent monitoring, and Canada's Cohere stake — then a deep dive on AI observability and tracing agent runs. AI Generated

  3. 4d ago

    Agentic Mornings — October 29, 2026: Agent Memory — How AI Agents Remember Across Sessions

    Google's Gemini 4 Carbon, Nvidia and AMD's $4B Meta bet, Liquid's pocket-sized LFM3 models, and Anthropic's government discount — then a deep dive on how AI agents remember across sessions, and why memory is a product decision with a bill. AI Generated

  4. 5d ago

    Agentic Mornings — October 28, 2026: Model Compression — How Big AI Gets Small

    Model compression — quantization, pruning, and distillation — plus Apple's M6 on-device AI pitch, DeepMind's Genie 4 world model, SAP's cheaper-model swap, and the reported OpenAI–Google chip deal. AI Generated

  5. 6d ago

    Agentic Mornings — October 27, 2026: Reasoning and Test-Time Compute

    Reasoning and test-time compute — why letting a model think longer works, what it costs, and how to set the dial, plus Anthropic's S-1, OpenAI's GPT-6.1 Terra, Meta's $95B raise, and Akamai's agent sandbox AI Generated

  6. Oct 26

    Agentic Mornings — October 26, 2026: Context Engineering — What Your AI Actually Sees

    OpenAI's sandbox escapes, Kimi K3.1 Pro's launch, inference prices down 80%, and the UK's AI biosecurity lab — plus a deep dive on context engineering: what a model sees, what it costs, and how to measure it. AI Generated

  7. Oct 23

    Agentic Mornings — October 23, 2026: Embeddings — How AI Searches by Meaning

    Pathos AI's $237M drug-discovery bet, AI-discovered vulnerabilities doubling, the SEC's AI exams, China's hiring guardrails — plus a deep dive on embeddings and vector search: how AI searches by meaning, and the TPM lens on cost, latency, evals, and bias. AI Generated

  8. Oct 22

    Agentic Mornings — October 22, 2026: How AI Models Get Trained

    OpenAI's models land on Google Cloud, Anthropic opens the Claude Duet waitlist, DeepSeek wraps a fifteen billion dollar raise, OpenAI's own agents weakened their safety monitoring — plus a deep dive on how AI models get trained: data, compute, scaling laws, and the TPM lens on training cost and risk. AI Generated

See All (18)

About

Agentic Mornings — the 20-minute weekday briefing for people who build with AI. Every weekday morning: the day's biggest AI stories, then one big idea explained properly — from first principles, with the numbers that matter: what it costs, how fast it runs, and how you know it actually works. No hype, no hot takes. News up front, deep dive after — the stuff that matters if you ship AI products, framed the way a practitioner wants it: cost, latency, evals, and the decision you can act on. New episodes every weekday morning.

Information

  • Creator
    Aman Gupta
  • Years Active
    2026
  • Episodes
    18
  • Rating
    Clean