Newsletter

Updates and long-form essays from the Nori team on AI agents and the agent ecosystem.

• Agentics: agent evals considered harmful
  most people are doing evals wrong, and its making their AI dumber
  October 21, 2026
• Agentics: Agent identity is a really hard problem
  This is why AI hasn’t transformed the economy yet
  October 14, 2026
• Nori and Subconscious Partner on Long-Running AI Agents
  Durable cloud infrastructure from Nori, fast and efficient long-horizon inference from Subconscious. Hit enter and close your laptop.
  October 7, 2026
• DevLog #24: Get Push Notifications, Open Long Chats Fast, and Start a Trial Without a Card
  Get a push notification when an agent finishes, open long conversations in under a second, start a 30-day trial without a card, use GPT-6.1 Sol, fix the GitHub App from the Integrations page, and request admin access from Discord.
  October 5, 2026
• DevLog #23: Resume Released Sessions, Watch the Agent’s Browser, and Chain Automations
  The September roundup. Resume released sessions with their files, watch and take over the agent’s browser, chain automations, connect API providers, run OpenCode and pi, and have Slack agents reply only when tagged.
  October 1, 2026
• DevLog #22: Attach Files in Web Chat, Run Slash Commands, and Use Claude Opus 5.5
  Work in the redesigned web chat, attach files and images, run commands from a slash menu, install chat on your phone, use Claude Opus 5.5, and trigger GitHub Actions workflows from a session.
  September 28, 2026
• Agentics: what is developer process automation?
  The only sane way to actually automate parts of the developer role.
  September 24, 2026
• How Miravoice builds and tests voice agents from Slack
  From a phone to a cloud development environment in ten minutes. Nori now handles more than 25% of Miravoice’s merged PRs.
  September 18, 2026
• DevLog #21: Save Multiple Accounts, Control Slack Replies, and Check Subscription Usage
  Save multiple Claude and ChatGPT accounts, check reported subscription usage, choose when agents reply in Slack, connect API providers, and manage automations from one panel.
  September 21, 2026
• DevLog #20: Edit Your Triggers, Resume Saved Work, and Keep Your CLI Draft
  Edit scheduled triggers, recover missed cron runs, reconnect integrations, resume saved conversations and repositories, see session details in chat threads, and keep your draft while the Nori CLI starts.
  September 14, 2026
• Agentics Meetup Videos and our first AI Takes Series
  I suppose newsletters are supposed to also be about the news sometimes?
  September 10, 2026
• Agentics: Agent Optimized Codebases
  What we look for when we turn existing codebases into agent-friendly ones
  September 7, 2026
• DevLog #19: Fire One Trigger When Another Finishes, Run OpenCode or pi, and Sign In to Codex from the CLI
  This week: fire a trigger when another trigger finishes, keep scheduled runs through a broker restart, run OpenCode or pi as your coding agent, pick Claude Fable 5.1, find every integration on one page, read the real reason a queued task failed, and sign in to Codex from inside the Nori CLI.
  September 7, 2026
• DevLog #18: Drive a Running CLI Session from Another Machine, Get One Reply from Slack, and Pick Models Your Agent Does Not List
  This week: turn on remote control in the Nori CLI and drive a running session from another machine, get one Slack reply per mention, ask a session to shut itself down and still get the answer, read transcripts that showed up blank in the browser, connect PostHog with read-only access, and pick models your coding agent does not advertise.
  August 31, 2026
• Agentics: How we use Cloud Agents to syndicate content across social media channels
  You can avoid a lot of brain rot if your agents go on social media for you
  August 25, 2026
• DevLog #17: Connect Your Provider on One Screen, Switch Models While You Chat, and Run Any Model ID
  This week: connect Claude or Codex on a single onboarding screen with three sign-in methods, sign in to Codex before your workspace exists, switch model, skillset, and provider from the web chat composer, open Billing as an org admin, run any model ID from the Nori CLI, and fixes for provider usage limits, MCP write access, Slack replies, and terminal keys.
  August 24, 2026
• Agentics: understanding the agent stack
  Understanding the agent stack from the user’s keyboard all the way to the datacenter and back
  August 12, 2026
• Agentics: how to build an agent to automate your on call
  A walk through of how you might set up an auto-on-call, why these things can fail, and how to think about agentic loops
  August 5, 2026
• DevLog #16: Track Unread Conversations, Pin the Agent Behind a Trigger, and Manage Cloud Sessions from the Terminal
  This week: see and clear unread conversations from the browser and see who started each session, pin which coding agent runs a trigger and keep triggers running through a credential rotation, manage cloud sessions from your terminal, and fixes that restore the Codex Spark model, start Gemini on Modal, and return tools from the Stripe MCP integration.
  August 3, 2026
• DevLog #15: The July Roundup
  July across Nori Sessions and the CLI: take a new org from login to its first pull request in the browser, keep conversations alive across machines and share them live, control which agent runs your scheduled triggers, and run agents headless with nori exec from scripts and CI.
  August 1, 2026
• Agentics: Personalizing Agents
  Videos and slides from our Agentics spotlight event on modifying an AI’s personality.
  July 30, 2026
• DevLog #14: Onboarding Grows a Front Door, Skillsets Go Git-Native, and Opus 5 Moves In
  Onboarding becomes a real in-app hub that walks you from login to your first agent-authored pull request, the chat history learns to remember what you have read, a single session streams live to every window watching it, Claude Opus 5 joins the model catalog, and the Skillsets CLI goes fully Git-native. nori-lint reaches week thirteen of stillness.
  July 27, 2026
• Agentics: Coding agents cannot remove complexity
  LLMs are excellent senior engineers and terrible staff engineers.
  July 21, 2026
• Agentics: Thoughts on Open Models and Kimi K3
  Kimi K3 changes some critical assumptions in the tech industry and meaningfully challenges US dominance in AI
  July 21, 2026
• DevLog #13: The Chat Opens a Window, Vim Gets Serious, and Conversations Outlive Machines
  Conversations become durable across machines and interfaces, the web chat opens live previews and gets smoother streaming, completion alerts, pull-request cards, and a tidier composer. Slack renders richer replies, integrations get more dependable, the CLI’s Vim mode grows up, and legacy configuration paths retire. nori-lint completes a full quarter of stillness.
  July 20, 2026
• DevLog #12: Slack Grows a Roster, the Web Chat Talks Back, and Logins Learn to Heal
  Slack grows a roster of custom per-org agents you mention by name, each with its own model and skillset, and every thread shows a live thinking/working status. The web chat learns bang commands, quick actions, and a hide-the-tool-calls toggle. Expired provider, MCP, and Codex logins heal themselves, GPT-5.6 joins the model bench, the registrar stops losing tarballs, and nori-skillsets finishes its namespace cleanup. nori-lint reaches week eleven of doing nothing.
  July 13, 2026
• Notes from an AI Conference
  My thoughts after attending AI Engineer World’s Fair: skill sprawl, evals, background agents, and why human review still matters.
  July 10, 2026
• Agentics: Design Principles for Building an Agent-First CLI
  What does ‘agent first’ mean in practice? Principles for CLI tools built for agents as their primary consumer.
  July 9, 2026
• DevLog #11: The Cloud Reaches the CLI, Slack Stops Stuttering, and Skillsets Breaks Its Silence
  Cloud sessions land inside the CLI with a real picker, resume, and close. Slack stops double-posting replies and mangling links, a stack of reliability fixes keeps one bad machine from wedging the fleet, triggers grow a test command, the registrar tightens its permissions, and nori-skillsets ends its silence. nori-lint reaches week ten of doing nothing.
  July 7, 2026
• Agentics: Sometimes free isn’t cheap enough
  On buffalomaxxing.
  July 5, 2026
• Agentics: Memorizing Session Transcripts Isn’t Useful
  Keep track of artifacts, not scratch. Session transcript search gave our agents zero performance benefit on SWE tasks — and automatic memory may make models worse.
  July 2, 2026
• Agentics: HTML is All You Need (for Agents to Make Graphics)
  An experiment in pushing coding agents as far as possible to make visual media.
  June 30, 2026
• DevLog #10: Conversations Follow You Everywhere, a Snowflake Agent Joins, and AWS Comes Aboard
  Start a session on one surface and resume it live on another, with new !resume and !catchup commands. Snowflake Cortex Code joins as a selectable agent, AWS becomes a one-form integration, checkpoints grow to 2GB, the registrar gets a skillset review desk, and nori-lint reaches week nine of doing nothing.
  June 29, 2026
• Agentics / Tech Things: Tokenmaxxing is dead, long live tokenmaxxing
  Alt title: The reports of tokenmaxxing’s death are greatly exaggerated.
  June 27, 2026
• Agentics: the modern company won’t have bullshit jobs
  The upside case for AI is that this is a freeing and joyful thing.
  June 25, 2026
• Agentics: Cost to Implement vs Cost to Verify
  Models have compressed the cost to implement. The cost to verify has not moved. That gap is the new bottleneck.
  June 22, 2026
• DevLog #9: Cursor Joins the Crew, Chat Moves to the Browser, and Sessions Survive a Deploy
  Cursor joins as a first-class agent, the broker gets a built-in chat app as its home page, and you can switch model or provider mid-thread from Slack. Reliability fixes keep sessions alive across deploys and broker restarts, the CLI's loop mode rides through rate limits, and nori-lint extends its zero-PR streak into an eighth week.
  June 22, 2026
• DevLog #8: Agents Improve Themselves, a New Brain Named Fable, and Admins Get the Keys
  Sessions ships an opt-in Skill Improvement Loop that drafts skillset edits for human approval, adds the Claude Fable 5 model and Brevo, Grafana, and Better Stack integrations, hands admins self-service idle timeouts and unlimited invites, and fixes the reliability bug that left sessions thinking forever.
  June 15, 2026
• Agentics: Using Coding Agents for Browser Automation
  Playwright is rapidly becoming my default laptop browser.
  June 10, 2026
• DevLog #7: The CLI Wakes Up, Agents Get a Browser, and Triggers Learn Time Zones
  The CLI returns from sabbatical with cloud sessions you can launch from the terminal, an agent-controlled browser, and persistent thread goals. Sessions ships time-zone-aware triggers, an in-browser skillset builder, a Render integration, and a stack of Slack and Discord fixes.
  June 8, 2026
• Agentics: local coding agents are inherently unsafe
  Easy, powerful, secure. Pick two.
  June 6, 2026
• DevLog #6: New Default Brains, Slack Bouncers, and a Goodbye to Goose
  Claude Opus 4.8 becomes the default model, Slack gains Open/Guarded/Strict access modes, the dashboard adds a Users page with invites and an owner role, the MCP catalog hits 58 servers, and Goose and Cline retire as first-class agents.
  June 1, 2026
• DevLog #5: Reminder Triggers, Discord Grows Up, and Agents Configure Themselves
  Sessions ships one-time triggers, Discord gets three reliability fixes, Handroll gains ACP remote control, and agents can now modify their own org environment.
  May 25, 2026
• Agentics: How Corvus uses AI to remove three hidden taxes on development
  Background agents helped a four-person team in chemical manufacturing feel like twelve
  May 19, 2026
• DevLog #4: Session Transcripts, Bang Commands, and Your CLI Learns New Tricks
  Sessions ships a transcript reader. Slack gets four new bang commands. The CLI composer gains shell mode, skill discovery, and syntax-highlighted diffs. Skillsets adds symlink support and settings.json backup.
  May 18, 2026
• Agentics: 1100 people RSVP’d to learn about coding agents
  Pictures, videos, and slides from Agentics NYC: May 2026.
  May 13, 2026
• DevLog #3: Public API Tokens, Docker in Sessions, and Slack Stops Saying “Sprite”
  Public registrar API tokens land end-to-end across Registrar, Skillsets, and Sessions. Docker daemons are first-class on sprites. A new sessions landing page with Slack permalinks. An OAuth incident becomes proactive token refresh.
  May 11, 2026
• DevLog #2: Multi-Agent MCP, Durable Resume, and One-Click Google Workspace
  MCP works across Claude, Codex, and Gemini at the same time. Sessions resume a Slack thread after the original VM is gone. The CLI gains a transcript-backed nori resume.
  May 4, 2026
• Agentics: Using Metacognition to Get A Model Upgrade
  The industry has coalesced around a standard best practice for using coding agents. We’re calling it SPACE development.
  April 30, 2026
• DevLog #1: Skillset-Aware Sessions, MCP Catalogs, and Honest Token Counters
  Sessions gets a real MCP catalog, skillset-aware triggers and Slack threads, built-in CLIs on every VM, and Skillsets fixes its token math.
  April 27, 2026
• Agentics: AI enablement requires managed agent runtimes
  AI enablement in enterprise requires nothing less than fully managed agent runtimes.
  April 23, 2026
• DevLog #0: Multi-Provider Support, MCP Integrations, and 47 Bugfixes
  Sessions supports Claude, Codex, and Gemini. MCP integrations are live. The CLI now lets you bring your own ACP agent.
  April 20, 2026
• Agentics: I hate making powerpoints, coding agents make it better
  Using coding agents to make slide decks ~25x faster than before.
  March 31, 2026
• Agentics: 6 emerging agent dev tool categories, market map, and meditations on the AI dev tool market
  Reinventing dev tools in the age of AI.
  March 24, 2026
• Agentics: How to Effectively Use Coding Agents
  A talk at NYU Langone’s OLAB on how Nori uses coding agents.
  March 13, 2026
• Agentics: We wrote a linter for SKILL.md files
  Parsing through a sea of snake oil.
  March 5, 2026
• Agentics: PMs Should Vibe Code
  AI is making product people the MVPs of any business.
  March 4, 2026
• Agentics: Agent orchestrators are bad
  Agent orchestrators cost more, produce worse outputs, and feel productive without being productive. Here’s why.
  February 19, 2026
• Agentics: Your agent skills are all slop
  We believe in only the finest, hand crafted, barrel aged, artisanal skills.
  January 18, 2026
• Agentics: People are seeing the light on coding agents
  Year 0 of the agentic coding revolution.
  January 7, 2026
• Agentics: Coding agents suck at microservices
  Somewhat contra humanlayer on docs.
  December 10, 2025
• Agentics: Debugging when everything is on fire
  Using AI to debug critical system failures.
  November 20, 2025
• Agentics: Coding Agents can Manage Other Coding Agents
  Test Driven Development is all you need
  November 14, 2025
• Agentics: Averaging 10+ PRs a day with Claude Code
  If you use Claude Code, try nori-ai, it’s really good
  October 30, 2025
• Agentics: We’re all senior engineers now
  All problems are search problems and how that applies to coding agents
  October 1, 2025