AI News · Daily
Every morning we scan Julian Goldie's SEO channel for the five most useful new AI videos, write our own short summary in UK English, and tie each one back to the NordSys service it affects. No hype. No transcripts. Just the bit a business owner actually needs.
Latest
The most recent summaries, in date order. Each article links out to Julian Goldie's original video and back to the NordSys service it connects to.
AI news · AI agents & tooling · 22 August 2026
Hermes Desktop v0.20.3 shipped Bot Mode as default in August 2026 — agent profiles become named Bots with their own model, memory, and skills, able to message each other directly or meet in shared collaboration rooms of two to six. Julian Goldie walked through the update. Here is what changed and who benefits.
Read the summary →AI news · AI agents & tooling · 22 August 2026
Julian Goldie's August 2026 video covers five new capabilities added to the Hermes Agent OS stack — the headliner being a /learn command that generates Python skills from plain English and refines them with use. Oracle, Jarvis voice, and an automated lead agent round out the update. Here is what each does in practice.
Read the summary →AI news · AI agents & tooling · 21 August 2026
DeepSeek released their MIT-licensed agent runtime on 14 August 2026 — a plugin-based framework where models, tools, sandboxes, and the agent loop itself are all swappable from configuration. Julian Goldie covered why it changes how coding agents are built. Here is what the harness actually does and where the current limits sit.
Read the summary →AI news · SEO & AI ranking · 21 August 2026
Julian Goldie's mid-August 2026 video covers the latest Hermes Agent OS additions — Claude, Hermes, and Fugu gaining a self-building loop, expanded SEO automation via Oracle and Firecrawl, and wake-word voice control. Here is what changed and where the limits still are.
Read the summary →AI news · Programming & automation · 21 August 2026
Julian Goldie's comparison video tests Prime Agent against Hermes on coding tasks. The result depends on task type: Prime Agent's self-editing notebook gives it the edge on isolated repeated coding work; Hermes wins when shared memory across a wider agent stack matters more. Here is why most comparisons frame this wrong.
Read the summary →AI news · SEO & AI ranking · 13 August 2026
Julian Goldie's August video shows Hermes Agent handling keyword research, content drafting, and publishing in a single uninterrupted loop — no manual prompt at each step. Here is what the automation actually covers and where a human still needs to step in.
Read the summary →AI news · Web, content & design · 13 August 2026
Julian Goldie used Claude Code to write the site structure and Seedance 2.5 to generate motion video for the visuals — building a finished website in one session. Here is what the two-tool workflow involves and what it cannot yet do alone.
Read the summary →AI news · AI agents & tooling · 12 August 2026
Prime Intellect released Prime Agent on 6 August 2026 under an MIT licence — a terminal coding agent that keeps a self-editing notebook of its own problem-solving steps. Each session, it reads what it wrote before and builds on it. Julian Goldie covered the release. Here is what the tool does and where the limits currently sit.
Read the summary →AI news · AI agents & tooling · 7 August 2026
Julian Goldie's latest video covers a significant step in the Agent OS stack: the combination of Claude, Hermes v0.20 Herald, and Sakana Fugu can now write, test, and deploy improvements to its own configuration. Here is what the full stack looks like today — Oracle, Studio, GLM, Open Montage, shared memory — and what self-improvement actually means in practice.
Read the summary →AI news · AI agents & tooling · 6 August 2026
Julian Goldie's 5 August video covers what Buzz recently fixed in how its autonomous Fizz agents handle errors and loops, then identifies a structural limitation in the Nostr-based architecture that the update does not address. Here is what changed and what it means for any business thinking of using Buzz in a real workflow.
Read the summary →AI news · AI agents & tooling · 5 August 2026
Nous Research shipped Hermes Agent v0.20.0 — the Herald Release — on 3 August 2026. It adds streaming voice with barge-in, an on-device wake word, A2A v1.0 agent-to-agent communication, signed webhooks, and a grounded-citations skill. Julian Goldie walked through the full release on 4 August.
Read the summary →AI news · AI agents & tooling · 4 August 2026
Julian Goldie's latest video covers three deployment options for Agent OS: running Hermes locally on your machine and reaching it from your phone via a Tailscale tunnel, hosting on a VPS for always-on reliability, and wiring in the Higgsfield MCP so your agent generates 4K images and cinematic video from plain-English prompts.
Read the summary →Category
Claude Code, Cursor, MCP, autonomous workflows — stories about the tools that actually do the work. Related service: AI Agents.
AI news · AI agents & tooling · 22 August 2026
Hermes Desktop v0.20.3 shipped Bot Mode as default in August 2026 — agent profiles become named Bots with their own model, memory, and skills, able to message each other directly or meet in shared collaboration rooms of two to six. Julian Goldie walked through the update. Here is what changed and who benefits.
Read the summary →AI news · AI agents & tooling · 22 August 2026
Julian Goldie's August 2026 video covers five new capabilities added to the Hermes Agent OS stack — the headliner being a /learn command that generates Python skills from plain English and refines them with use. Oracle, Jarvis voice, and an automated lead agent round out the update. Here is what each does in practice.
Read the summary →AI news · AI agents & tooling · 21 August 2026
DeepSeek released their MIT-licensed agent runtime on 14 August 2026 — a plugin-based framework where models, tools, sandboxes, and the agent loop itself are all swappable from configuration. Julian Goldie covered why it changes how coding agents are built. Here is what the harness actually does and where the current limits sit.
Read the summary →AI news · AI agents & tooling · 12 August 2026
Prime Intellect released Prime Agent on 6 August 2026 under an MIT licence — a terminal coding agent that keeps a self-editing notebook of its own problem-solving steps. Each session, it reads what it wrote before and builds on it. Julian Goldie covered the release. Here is what the tool does and where the limits currently sit.
Read the summary →AI news · AI agents & tooling · 7 August 2026
Julian Goldie's latest video covers a significant step in the Agent OS stack: the combination of Claude, Hermes v0.20 Herald, and Sakana Fugu can now write, test, and deploy improvements to its own configuration. Here is what the full stack looks like today — Oracle, Studio, GLM, Open Montage, shared memory — and what self-improvement actually means in practice.
Read the summary →AI news · AI agents & tooling · 6 August 2026
Julian Goldie's 5 August video covers what Buzz recently fixed in how its autonomous Fizz agents handle errors and loops, then identifies a structural limitation in the Nostr-based architecture that the update does not address. Here is what changed and what it means for any business thinking of using Buzz in a real workflow.
Read the summary →AI news · AI agents & tooling · 5 August 2026
Nous Research shipped Hermes Agent v0.20.0 — the Herald Release — on 3 August 2026. It adds streaming voice with barge-in, an on-device wake word, A2A v1.0 agent-to-agent communication, signed webhooks, and a grounded-citations skill. Julian Goldie walked through the full release on 4 August.
Read the summary →AI news · AI agents & tooling · 4 August 2026
Julian Goldie's latest video covers three deployment options for Agent OS: running Hermes locally on your machine and reaching it from your phone via a Tailscale tunnel, hosting on a VPS for always-on reliability, and wiring in the Higgsfield MCP so your agent generates 4K images and cinematic video from plain-English prompts.
Read the summary →AI news · AI agents & tooling · 30 July 2026
Jack Dorsey's Block launched Buzz, a free open-source Slack alternative with AI agents built into every channel. Julian Goldie tested it against Hermes Agent. The two tools do different jobs — Buzz is the workspace, Hermes is the worker — and they connect more naturally than the comparison title suggests.
Read the summary →AI news · AI agents & tooling · 29 July 2026
Hermes Agent 2.0 — version 0.20 — introduces a Model Portal that connects over 200 AI models and routes different tasks to the right one automatically. One gateway can send your SEO work to Qwen, your coding to Sonnet 5, and your outreach to GPT, all from the same Hermes installation. Julian Goldie walked through the full release. Here is what changed over the v0.19 Quicksilver build, what the multi-agent routing does in practice, and whether a small business team actually needs it.
Read the summary →AI news · AI agents & tooling · 28 July 2026
Alibaba announced Qwen 3.8 Max at the World Artificial Intelligence Conference in Shanghai on 19 July 2026. The model has 2.4 trillion parameters using a mixture-of-experts architecture, handles text, images, video, and documents in a single call, and is claimed to rank second only to Fable 5 on the benchmarks Alibaba ran. Julian Goldie covered the release and its implications for AI agents. Here is what the numbers mean and what is still unconfirmed.
Read the summary →AI news · AI agents & tooling · 28 July 2026
Hermes Cloud takes a Hermes Agent and moves it into a cloud container you do not have to manage. The setup takes roughly 60 seconds: sign in to the portal, name the agent, pick a model, deploy. From that point the agent runs continuously — handling tasks, watching schedules, monitoring feeds — whether your laptop is open or not. Julian Goldie called the setup "TOO EASY". Here is what it actually does and what it still requires from you.
Read the summary →AI news · AI agents & tooling · 28 July 2026
Hermes Agent has a voice mode now. You speak, it replies the moment you stop talking, and then it acts — no wake word, no push-to-talk button. Julian Goldie built a full website in roughly 90 seconds without touching his keyboard. Here is what the setup actually does and whether it is useful beyond the demo.
Read the summary →AI news · AI agents & tooling · 26 July 2026
Released on 20 July 2026, Hermes v0.19 Quicksilver cuts time-to-first-token from 4.3 seconds to 0.9 seconds and writes every job to a database so tasks finish even if you close the application. Julian Goldie walked through the full release in detail — here is what actually changed and whether it matters for a working business.
Read the summary →AI news · AI agents & tooling · 23 July 2026
Julian Goldie walked through the full Hermes Agent OS stack powered by Kimi K3 — a free, 2.8-trillion-parameter model with a 1-million-token context window. The combination runs Oracle (trending keyword to published SEO post), Astros (competitor monitoring), Jarvis (voice control), and a shared Obsidian memory layer from one dashboard.
Read the summary →AI news · AI agents & tooling · 22 July 2026
Kimi K3 is free at a basic tier but rate-limited until you add $20 to a Moonshot API account. Julian Goldie walked through the full build: API key, Hermes Agent config, OpenClaw setup, and what changes when both tools run Kimi K3 as their model brain.
Read the summary →AI news · AI agents & tooling · 22 July 2026
Moonshot AI launched Kimi K3 on 16 July 2026 — 2.8 trillion parameters, an open-source licence, a 1-million-token context window and a Swarm variant built for parallel agent tasks. Julian Goldie's ten-minute walkthrough covers the essentials. Here is what UK businesses running Hermes Agent or OpenClaw actually need to know.
Read the summary →AI news · AI agents & tooling · 21 July 2026
Google's Gemini Spark agent received its most substantial update in mid-July 2026 — over 50% faster on long tasks, direct editing of shared Google Docs and Slides, Google Tasks and Keep integration, and real-time topic monitoring. UK and EEA businesses are still locked out. Julian Goldie walked through all the changes.
Read the summary →AI news · AI agents & tooling · 17 July 2026
Anthropic has extended Claude Fable 5 free access to 19 July 2026 — the third deadline change in three weeks. Claude Code's 50% rate limit extension runs to the same date. Julian Goldie's recommended approach: use the remaining time to build reusable workflows and prompt libraries before pay-per-token billing kicks in.
Read the summary →AI news · AI agents & tooling · 12 July 2026
OpenAI launched ChatGPT Work and Anthropic pushed Claude Cowork to iOS and Android on the same day — 9 July 2026. Two agentic tools, one head-to-head. Julian Goldie ran a full breakdown comparing what each does, where the differences matter, and which one holds up under real business use.
Read the summary →AI news · AI agents & tooling · 10 July 2026
Running Claude in a persistent loop with a builder, a judge, and a stop condition changes what single-person teams can produce — you set the brief and check the output, Claude handles every step in between. Julian Goldie walks through five specific loop patterns, from the builder/judge cycle to an Oracle monitoring loop that runs on a schedule.
Read the summary →AI news · AI agents & tooling · 9 July 2026
SpaceXAI released Grok 4.5 publicly on 9 July 2026 — a 1.5T V9 model trained on Cursor interaction data, priced at $2 per million input tokens. It sits just below Claude Opus 4.8 on the intelligence index. Julian Goldie ran a full benchmark test. Here is the honest read on where it holds up.
Read the summary →AI news · AI agents & tooling · 4 July 2026
After an 18-day suspension caused by US export controls, Claude Fable 5 access was restored on 1 July 2026. Julian Goldie covers how to get the most out of it — including the Olympus Protocol for Hermes Agent integration and a usage management approach for keeping costs in check.
Read the summary →AI news · AI agents & tooling · 2 July 2026
Hermes Agent v0.18 ships a /learn command that saves any successful task sequence as a reusable skill mid-session, adds a Judgment verification layer that checks agent output before it is returned, and improves reliability on long-running multi-step tasks. Julian Goldie walked through all three additions.
Read the summary →AI news · AI agents & tooling · 29 June 2026
With Claude Fable 5 and GPT-5.6 locked behind closed access programmes, Mixture of Agents inside Hermes offers a practical route to frontier-quality output from models you already have. Julian Goldie tested whether MoA genuinely closes the gap — and what the benchmark numbers actually show.
Read the summary →AI news · AI agents & tooling · 28 June 2026
Hermes has shipped an update to its Agent OS that brings Mixture of Agents, Oracle automated news monitoring, and Obsidian-based memory into a single dashboard. Each component was already available as a separate feature; the integration is what is new. Julian Goldie walked through the combined setup live.
Read the summary →AI news · AI agents & tooling · 28 June 2026
A panel of Opus 4.8 and GPT-5.5 running inside Hermes's Mixture of Agents scored 0.82 on Hermes Bench versus 0.76 for Opus alone. Julian Goldie ran the comparison. The headline number is real; the caveats are worth knowing before you rebuild your agent stack around it.
Read the summary →AI news · AI agents & tooling · 27 June 2026
MiniMax M3, an open-source Chinese agentic model with a 1-million-token context window, connects to Hermes Agent via Ollama to create a free local AI worker that handles keyword research, content creation, meta tagging and internal linking from a single prompt. Julian Goldie tested the setup live. Here is the honest take on what it actually produces.
Read the summary →AI news · AI agents & tooling · 25 June 2026
Hermes shipped a notable update in late June 2026 with three new capabilities: computer use extended from Mac to Windows and Linux, a Loop Engine that reruns tasks that fall short of a quality threshold, and Grok integration giving the agent real-time access to X's search index. Julian Goldie demonstrated all three live.
Read the summary →AI news · AI agents & tooling · 24 June 2026
Sakana AI's Fugu Ultra routes prompts across a panel of models, fuses the results, and scores 73.7 on SWE-Bench Pro — ahead of Opus 4.8 at 69.2 and GPT-5.5 at 58.6 — at roughly 25% of OpenRouter Fusion's cost. Julian Goldie ran the full test. Here is the plain-English read on whether those numbers hold up for UK businesses.
Read the summary →AI news · AI agents & tooling · 23 June 2026
Hermes Agent v0.17.0 — The Reach Release — landed on 19 June 2026 with background subagents, iMessage and WhatsApp integrations, and a redesigned desktop app. Julian Goldie's review covers what each change actually means. Here is the plain-English read for UK businesses.
Read the summary →AI news · AI agents & tooling · 22 June 2026
SpaceX announced a $60 billion all-stock acquisition of Cursor while Z.ai's GLM 5.2 topped the open-weight model rankings. Julian Goldie's weekly roundup covers both stories. Here is the plain-English read on what each actually changes for teams building with AI agents.
Read the summary →AI news · AI agents & tooling · 18 June 2026
Julian Goldie demonstrated Hermes Agent's Oracle voice system — a keyword-activated AI that builds apps in real time and saves outputs to a local memory graph. It runs locally on any model, with no cloud API required.
Read the summary →AI news · AI agents & tooling · 16 June 2026
OpenRouter's Fusion API sends one prompt to up to eight models simultaneously — Opus, Gemini, Grok and others run in parallel with live web search — then a judge model synthesises the best reply. Julian Goldie covered this alongside the GLM 5.2 and Kimi K2.7 releases and the continuing fallout from the Claude Fable 5 export ban.
Read the summary →AI news · AI agents & tooling · 14 June 2026
Three days after Anthropic launched Claude Fable 5 and Mythos 5, an emergency US export control order cut off access for all foreign nationals — UK businesses included. Julian Goldie covered what the directive says, which Claude models are still available, and what to do now.
Read the summary →AI news · AI agents & tooling · 12 June 2026
Claude Fable 5 landed on 9 June 2026 — Anthropic's most capable publicly available model, with a 1-million-token context window and multi-day autonomous task execution. Julian Goldie ran through five practical use cases and what the reasoning improvement actually changes for AI agent workflows.
Read the summary →AI news · AI agents & tooling · 10 June 2026
Most AI agents sit in a terminal or browser tab, waiting to be asked something. Hermes Agent's new email integration changes that — it polls your inbox every 15 seconds, reads incoming messages, and replies in-thread using the same persistent memory it carries across every other session.
Read the summary →AI news · AI agents & tooling · 8 June 2026
Typing commands into an AI agent is still the default for most setups. Julian Goldie's latest video changes that — a Hermes Agent configured with Minimax M3 as the reasoning layer and voice mode enabled produces a working voice-first agent that takes on tasks, not just questions.
Read the summary →AI news · AI agents & tooling · 8 June 2026
Nous Research shipped Hermes v0.16 — The Surface Release — on 5 June 2026, with a one-click native desktop application at the centre. The update also adds a remote gateway, a browser-based admin panel, and a fuzzy model picker across every interface.
Read the summary →AI news · AI agents & tooling · 8 June 2026
Julian Goldie built a Jarvis-style personal AI assistant using Hermes Agent in a single session and shared the full walkthrough. The build does not require any coding beyond basic configuration — goals, voice mode, memory, the right model.
Read the summary →AI news · AI agents & tooling · 8 June 2026
Julian Goldie demonstrated Hermes Agent configured as a Jarvis-style assistant — voice input, spoken replies, persistent memory across the session, and genuine task execution rather than simple answers. Here is what the demo shows.
Read the summary →AI news · AI agents & tooling · 3 June 2026
Hermes Workspace V2 ships with direct browser chat, multi-profile management, a Kanban task board and mobile PWA support — the kind of interface layer that was missing when Hermes was a purely terminal tool. Here is what the upgrade changes in practice.
Read the summary →AI news · AI agents & tooling · 1 June 2026
Julian Goldie used Claude Opus 4.8's Ultracode mode to build a functional agentic operating system in about 15 minutes — the kind of locally hosted dashboard that used to take a full day. Here is what the build produced and what you'd need more time for.
Read the summary →AI news · AI agents & tooling · 30 May 2026
Julian Goldie connected Claude, Hermes Agent and OpenClaw into a single locally hosted dashboard — shared memory, shared goals, one screen to manage all three. It took roughly an hour to build using Claude Desktop, Next.js and Tailwind. Here is what the Agent OS setup looks like and what genuinely changes when your AI tools stop running in isolation.
Read the summary →AI news · AI agents & tooling · 29 May 2026
Anthropic shipped Claude Opus 4.8 on 28 May 2026 with Ultracode — a high-effort orchestration mode that spawns parallel sub-agents from a single goal. Julian Goldie reviewed the launch the same day. Here is the honest read on what changes and what doesn't.
Read the summary →AI news · AI agents & tooling · 27 May 2026
Julian Goldie ran Hermes Agent against the leading commercial AI agent on identical real-world tasks and argued the result isn't close any more. Here's what's actually new in the update and what it means if you're considering an AI agent for business work.
Read the summary →AI news · AI agents & tooling · 27 May 2026
Hermes Agent has added a built-in video generation capability that runs inside the same workflow you already use for SEO and content tasks — no separate tool, no extra login. Julian Goldie walked through the new skill live.
Read the summary →AI news · AI agents & tooling · 26 May 2026
Julian Goldie ran the updated Hermes Agent against Claude CoWork and OpenClaw on real business tasks. Here's the honest read on what each tool does well and where Hermes has genuinely moved ahead.
Read the summary →AI news · AI agents & tooling · 25 May 2026
Running Hermes, OpenClaw and Claude separately is useful. Connecting all three as a coordinated system — with shared goals, persistent memory, and one dashboard — changes what's possible. Julian Goldie walked through the setup live.
Read the summary →AI news · AI agents & tooling · 24 May 2026
Julian Goldie rebuilt a functional version of Hermes Agent using Claude Code in a single session. The exercise reveals something more interesting than the end result — what it now means for businesses to build and customise their own AI agents.
Read the summary →AI news · AI agents & tooling · 23 May 2026
Hermes Agent OS 3.0 adds an Obsidian memory layer, Grok integration, ChatGPT data syncing, and a redesigned dashboard. Julian Goldie walked through the full update. Here's what each piece adds and whether it's worth the upgrade effort.
Read the summary →AI news · AI agents & tooling · 22 May 2026
The 14th major release of Hermes Agent shipped on 16 May with 808 commits from 215 contributors. Key additions: xAI Grok as a first-class OAuth provider, native X (Twitter) search, a local proxy that makes Claude Pro or ChatGPT Pro work inside any OpenAI-compatible tool, and browser operations 180 times faster via persistent connections.
Read the summary →AI news · AI agents & tooling · 22 May 2026
Connecting Claude to Hermes through a shared Agent OS dashboard gives each tool its proper role — Claude handles reasoning, Hermes handles execution and persistent memory. Julian Goldie walked through the four-layer Goldie Mission Stack. Here is what the configuration involves and what you get from running them as a coordinated system.
Read the summary →AI news · AI agents & tooling · 22 May 2026
The latest Hermes update adds computer use on Mac — your AI agent clicks, types, scrolls, and navigates the desktop while you use another application. Combined with swarm mode and a new free model on OpenRouter, here is what the capability looks like and where it is ready to use now.
Read the summary →AI news · AI agents & tooling · 18 May 2026
Julian Goldie walked through a setup that runs Claude Code, OpenClaw and Hermes Agent simultaneously without paying for API access — by routing all three through free models on OpenRouter. Here's what the configuration actually requires and what you give up by skipping the paid tiers.
Read the summary →AI news · AI agents & tooling · 18 May 2026
Most coverage of Hermes Agent focuses on SEO workflows. Julian Goldie's latest walkthrough goes further — showing the tool handling inboxes, app builds, website deployments, and recurring task management. Here's what actually holds up in practice.
Read the summary →AI news · AI agents & tooling · 18 May 2026
Hermes Agent has always been open-source, but you still needed a paid model to get useful results — until Owl Alpha appeared on OpenRouter with a 1-million-token context window and no cost. Julian Goldie showed how to connect the two.
Read the summary →AI news · AI agents & tooling · 15 May 2026
Julian Goldie ran OpenHuman, Hermes Agent and OpenClaw against the same brief. The comparison cuts through the YouTube hype to show where each tool earns its keep in a real workflow — and where Hermes has pulled into a clear lead on open-source AI agent usage.
Read the summary →AI news · AI agents & tooling · 15 May 2026
The new Hermes Agent UI replaces terminal commands with a browser-based dashboard showing chats, files, swarms, tasks, and missions in one place. Here's what actually changes when your AI agent has a proper interface — and who it genuinely helps.
Read the summary →AI news · AI agents & tooling · 15 May 2026
Julian Goldie walked through a Hermes Agent session handling real business tasks — content drafting, prospect scraping, site auditing — all from goal-based prompts. Here's an honest look at what the tool handles well and where it still needs a human in the loop.
Read the summary →AI news · AI agents & tooling · 14 May 2026
The Hermes HUD UI is a free browser dashboard that shows memory state, scheduled jobs, skill usage, API health, and token costs in real time — no terminal needed. Here's what changes when you can finally see what your AI agent is actually doing.
Read the summary →AI news · AI agents & tooling · 13 May 2026
Connect Claude to Hermes Agent through the Model Context Protocol and you give your AI a way to stop just answering questions and start doing things. Julian Goldie walked through the setup — here's how the delegation loop works and what tasks it handles well.
Read the summary →AI news · AI agents & tooling · 13 May 2026
Julian Goldie published a complete Hermes Agent setup walkthrough for business owners starting from scratch — install, model, memory, skills, and first goal. Here's what you actually need and what to configure before your agent starts doing anything useful.
Read the summary →AI news · AI agents & tooling · 12 May 2026
Julian Goldie demonstrated Hermes Swarm mode running a coordinated team of planner, builder, reviewer and router agents against a single SEO goal simultaneously. The entire setup runs on free open-source models through OpenRouter — no paid API subscription required.
Read the summary →AI news · AI agents & tooling · 12 May 2026
Julian Goldie ran a live session combining Hermes Agent, Aion UI, Obsidian and Omi — a four-tool stack that gives your AI agent continuous memory across sessions. Here's what each piece does and what the full stack looks like in practice.
Read the summary →AI news · AI agents & tooling · 11 May 2026
Julian Goldie tested Hermes Agent with Claude powering a coordinated swarm of specialist AI agents on a single SEO brief — planner, builder, and reviewer all working in parallel. Here's what the workflow actually produces.
Read the summary →AI news · AI agents & tooling · 10 May 2026
Google is quietly building a Gemini-powered AI agent codenamed Remy, designed to manage tasks across Gmail, Calendar, Drive and Android apps. Julian Goldie covered what we know — here's the honest take on whether it poses a real challenge to OpenClaw.
Read the summary →AI news · AI agents & tooling · 10 May 2026
Hermes Workspace brings your AI agent's task list, memory, skill library and swarm controls into one graphical dashboard — no terminal needed. Here's what it actually adds day to day.
Read the summary →AI news · AI agents & tooling · 9 May 2026
The latest Hermes Agent update adds Codex browser use: your AI agent can now click, search, fill forms, and navigate websites automatically. Here's what it actually changes for automation workflows.
Read the summary →AI news · AI agents & tooling · 9 May 2026
Most coverage of Hermes v0.13 focused on /goal and Kanban. The changes that matter most for teams building with it long-term are quieter: swappable AI providers, eight security closures, proper state persistence, and support for seven languages.
Read the summary →AI news · AI agents & tooling · 9 May 2026
Hermes Agent v0.13 — the Tenacity release — shipped on 7 May 2026 with 864 commits and 295 contributors. Here's the practical first-look: what shifted in day-to-day behaviour, and what to review before pulling the update.
Read the summary →AI news · AI agents & tooling · 8 May 2026
Hermes v0.13 adds a /goal command that locks an AI agent onto a target and keeps it working until the job is genuinely done — no re-prompting, no drift. Here's what actually changes in practice.
Read the summary →AI news · AI agents & tooling · 8 May 2026
NVIDIA's Nemotron 3 Nano Omni is an open-source model that can see images, transcribe audio, analyse video, and read documents in one pipeline. Julian Goldie showed how it connects to Hermes Agent to run the whole stack locally, without paying for cloud APIs.
Read the summary →AI news · AI agents & tooling · 8 May 2026
Hermes Swarm Mode lets you configure distinct agent profiles — researcher, planner, writer, reviewer — and launch them all simultaneously against a shared brief. Here's what changes when you move from sequential to parallel agents.
Read the summary →AI news · AI agents & tooling · 8 May 2026
PaperClip coordinates multiple AI agents autonomously. Stack it with Hermes and Gemma 4 and you get a free, open-source multi-agent system that handles complex workflows end-to-end — no paid API required.
Read the summary →AI news · AI agents & tooling · 7 May 2026
Hermes Agent's Kanban update gives multiple AI agents a shared task board to draw from, work in parallel, and hand off when stuck — the biggest change to how Hermes manages work since launch.
Read the summary →AI news · AI agents & tooling · 7 May 2026
Hermes V0.12 adds a background Curator that automatically reviews and tidies your skill library every seven days, native Google Meet and Spotify integrations, a 57% faster startup, and 360 bug fixes.
Read the summary →AI news · AI agents & tooling · 5 May 2026
OpenClaw 2026.5.3 is a repair release — over 200 bug fixes, a new bundled file transfer plugin with path-based security, and cleaner plugin management across Discord, Telegram, WhatsApp and Slack. Julian Goldie walked through what shipped. Here's what matters.
Read the summary →AI news · AI agents & tooling · 4 May 2026
OpenClaw 5.2 adds proper multi-agent mode — sub-agents that run in parallel with isolated sessions and automatic recovery when a task stalls. Julian Goldie walked through the release. Here's what changed and what it means for anyone running AI agents in production.
Read the summary →AI news · AI agents & tooling · 4 May 2026
Anthropic accidentally shipped 512,000 lines of Claude Code source in an npm package, naming Sonnet 4.8 and revealing KAIROS — a built-but-unshipped persistent background agent mode. Julian Goldie covered the findings. Here's what developers actually need to know.
Read the summary →AI news · AI agents & tooling · 3 May 2026
Hermes takes an SEO goal — keyword cluster, content calendar, landing page build — and works through every step in a loop until the job is done, no prompting at each stage. Julian Goldie tested it on real SEO targets. Here's what it produced and where it still needs human oversight.
Read the summary →AI news · AI agents & tooling · 1 May 2026
Comfy UI now connects to Hermes Agent, giving it a free, local layer for generating images, video, and audio on demand. Julian Goldie walked through the integration — here's what the combination can actually do for a business and where the setup still bites.
Read the summary →AI news · AI agents & tooling · 30 Apr 2026
Aion is a free multi-agent dashboard that wraps a proper graphical interface around Hermes Agent — chat, session history, model switching, and task management in one place. Julian Goldie walked through the setup; here's what it changes for day-to-day use.
Read the summary →AI news · AI agents & tooling · 30 Apr 2026
Hermes Labyrinth v0.1.0 is a read-only observability plugin that maps every prompt, tool call, failure, and model switch your Hermes Agent makes. Released 29 April 2026, it's the first practical answer to the AI agent black-box problem.
Read the summary →AI news · AI agents & tooling · 29 Apr 2026
Hermes Agent v0.11.0 ships a free desktop app that replaces terminal access with a GUI for sessions, memory, model switching, and tools. Here's what actually changed and whether it matters for running a local AI agent.
Read the summary →AI news · AI agents & tooling · 28 Apr 2026
Google merged Vertex AI into a new Gemini Enterprise platform at Cloud Next '26, adding Agent Studio and Agent Gateway for building multi-agent workflows. Julian Goldie walked through it — here's what matters for UK businesses right now.
Read the summary →AI news · AI agents & tooling · 26 Apr 2026
On 23 April 2026, Anthropic added 13 personal app connectors to Claude Cowork — Spotify, Uber, Uber Eats and more. The point is the pattern, not the app list: business-grade connectors are the obvious next step.
Read the summary →AI news · AI agents & tooling · 24 Apr 2026
By default Claude starts fresh every session — it doesn't know your clients, conventions, or history. Connecting it to an Obsidian vault through MCP gives it a persistent knowledge base it reads before every task.
Read the summary →AI news · AI agents & tooling · 23 Apr 2026
Moonshot AI's open-weight model coordinates up to 300 specialist agents in parallel and runs autonomously for 12-plus hours. Here's what's real, what's noise, and what it means for a UK business.
Read the summary →AI news · AI agents & tooling · 21 Apr 2026
Anthropic's Skills feature packages reusable workflows into short files Claude Code loads on demand — cutting token costs and keeping agent runs consistent.
Read the summary →Category
LLM ranking, AI Overviews, Google SGE, getting cited in ChatGPT and Perplexity answers. Related service: SEO & AI Ranking.
AI news · SEO & AI ranking · 21 August 2026
Julian Goldie's mid-August 2026 video covers the latest Hermes Agent OS additions — Claude, Hermes, and Fugu gaining a self-building loop, expanded SEO automation via Oracle and Firecrawl, and wake-word voice control. Here is what changed and where the limits still are.
Read the summary →AI news · SEO & AI ranking · 13 August 2026
Julian Goldie's August video shows Hermes Agent handling keyword research, content drafting, and publishing in a single uninterrupted loop — no manual prompt at each step. Here is what the automation actually covers and where a human still needs to step in.
Read the summary →AI news · SEO & AI ranking · 20 July 2026
Google's Gemini 3.5 Pro has missed its third public deadline — Bloomberg and CNBC confirmed the July slip on 16 July 2026. Alphabet shares fell around 5%. Four senior DeepMind researchers have left and Google is weighing a Gemini 3.6 Flash interim release. Julian Goldie covered what it means for UK businesses using Google and AI search.
Read the summary →AI news · SEO & AI ranking · 5 July 2026
Leaked details of Google's Gemini 3.5 Pro (internal codename Cappuccino) point to a 2-million-token context window, a Deep Think reasoning mode, and an always-on Spark agent that runs tasks across Gmail, Docs, Drive and Search without prompting. Julian Goldie covered it as the leaks broke.
Read the summary →AI news · SEO & AI ranking · 1 July 2026
Google added Short Video Overviews to NotebookLM on 30 June 2026, powered by Nano Banana 2 Lite. The feature generates 60-second vertical videos from notebook research automatically — script, images, voiceover, and editing included. Julian Goldie tested the output and compared it against other AI video tools.
Read the summary →AI news · SEO & AI ranking · 27 June 2026
Hermes Agent Swarms coordinate planner, builder, reviewer and router agents in a self-correcting quality loop. Julian Goldie documents two real sites going from zero organic traffic to hundreds of clicks per day, and explains the loop engineering that makes the results reproducible rather than lucky.
Read the summary →AI news · SEO & AI ranking · 27 June 2026
Google's second spam update of 2026 started rolling out on 24 June — targeting scaled content and, for the first time, manipulation of AI Overviews and AI Mode in Search. Julian Goldie covered the update as it landed. Here is the plain-English read on what actually changed and what UK site owners should do now.
Read the summary →AI news · SEO & AI ranking · 22 June 2026
Perplexity launched Brain on 18 June 2026 — a memory layer that records every task its Computer agent runs and updates itself overnight, improving accuracy by 25%. Julian Goldie covers what this means for businesses whose content appears in Perplexity answers.
Read the summary →AI news · SEO & AI ranking · 19 June 2026
Z.ai's GLM 5.2 is a 744-billion-parameter open-weights model that matches Claude Opus 4.8 on long-horizon tasks and carries an MIT licence. Julian Goldie wired it into Hermes Agent to run an Infinite Video Engine and an Everywhere Engine that publishes across Google, ChatGPT and Perplexity simultaneously — reaching 200 organic clicks a day from a fresh domain without paid API access.
Read the summary →AI news · SEO & AI ranking · 16 June 2026
Two new open-source models arrived in the same week — GLM 5.2 from Zhipu AI and Kimi K2.7 from Moonshot — and Julian Goldie put all three through the same content and SEO tasks alongside Claude Opus 4.8. Here is which model performed best for each type of SEO work, and why benchmark tables alone miss the point.
Read the summary →AI news · SEO & AI ranking · 9 June 2026
Google's AI Overviews pull from video as well as text, and Hermes Agent's HyperFrames skill produced a video that hit first position in both standard results and AI Overviews within 19 hours of publishing. Julian Goldie walked through the setup and the results.
Read the summary →AI news · SEO & AI ranking · 5 June 2026
Julian Goldie configured Hermes Agent with a goal-based loop and a judge-style quality check that runs keyword research, content calendars, blog drafts, and internal link plans end to end from one prompt — working through the entire job without manual prompting at each stage.
Read the summary →AI news · SEO & AI ranking · 3 June 2026
Gemini 3.5 Flash landed at Google I/O 2026 as a speed-first frontier model. Julian Goldie tested it inside Google AI Studio for SEO tooling — keyword clustering, meta descriptions, and content planning at scale.
Read the summary →AI news · SEO & AI ranking · 31 May 2026
Google's NotebookLM added automatic Google Drive syncing this week; Anthropic shipped Opus 4.8 with stronger reasoning the week before. Julian Goldie shows what happens when you wire the two into a content pipeline.
Read the summary →AI news · SEO & AI ranking · 29 May 2026
Julian Goldie applied Claude Opus 4.8's Ultracode mode to a live SEO workflow — keyword research, content planning and page drafts running in parallel from one prompt. Here is what it produced and where the limits still sit.
Read the summary →AI news · SEO & AI ranking · 27 May 2026
Google's May core update started rolling out on 21 May, at the same moment Google I/O was overhauling AI Mode. Julian Goldie covered the combined impact. Here's the plain-English version of what shifted and what to do about it now.
Read the summary →AI news · SEO & AI ranking · 26 May 2026
Hermes Agent has built a quiet reputation as one of the more capable free AI agents available, but most coverage is about installing it. Julian Goldie put it through a full search workflow — keyword research, content planning, page generation, and swarm-mode production. Here is what held up.
Read the summary →AI news · SEO & AI ranking · 25 May 2026
Julian Goldie walks through the Hermes Agent OS SEO playbook: Claude selects the keywords, HyperFrames builds the video, an optimised publish gets it indexed, then a multi-format spread multiplies the reach. Here is what the workflow looks like in practice and what you can realistically expect from it.
Read the summary →AI news · SEO & AI ranking · 25 May 2026
Minimax 2.1 has been benchmarking well against Claude and Gemini. Julian Goldie tested all three on identical SEO briefs — keyword clustering, content outlines, meta copy — and the results weren't what the benchmark tables would suggest.
Read the summary →AI news · SEO & AI ranking · 24 May 2026
Google I/O 2026 introduced Gemini Spark — a personal AI agent that runs continuously in the background across Gmail, Drive, Docs and Calendar. Julian Goldie covered the update; here's the plain-English take on what it changes for SEO and how businesses should think about it now.
Read the summary →AI news · SEO & AI ranking · 22 May 2026
Hermes Agent's Kanban swarm mode takes a single SEO goal and distributes the work across specialist agents on a shared task board — planner, builder, reviewer, all running at the same time. Julian Goldie demonstrated the workflow on a real brief. Here is what the output looks like and where the approach holds up.
Read the summary →AI news · SEO & AI ranking · 14 May 2026
HyperFrames is a new open-source Hermes skill that generates SEO-ready MP4 videos from written descriptions — no timeline editor needed. Julian Goldie tested one on an SEO brief and reported it ranking first on Google and inside AI Overviews within 19 hours of publishing.
Read the summary →AI news · SEO & AI ranking · 13 May 2026
Connect Claude Code to Google's NotebookLM via MCP and your AI can interrogate a curated research library before writing anything — from keyword groupings to topical authority maps. Here's what the workflow actually produces.
Read the summary →AI news · SEO & AI ranking · 10 May 2026
Julian Goldie tested Claude Opus 4.7 as a full SEO workflow — from keyword brief through to indexed page — in under 24 hours. Here's what the process actually looks like and where it still needs a human in the loop.
Read the summary →AI news · SEO & AI ranking · 10 May 2026
A UI string inside the Gemini app referenced both Gemini 3.2 and a model called Omni ahead of Google I/O on 19 May 2026. Julian Goldie covered what the leak implies for AI-first search and what site owners should expect.
Read the summary →AI news · SEO & AI ranking · 7 May 2026
The Kanban update to Hermes Agent changes more for SEO teams than almost any other use case — specialist agents can now run simultaneously across keyword research, content briefs, and on-page build tasks from one shared board.
Read the summary →AI news · SEO & AI ranking · 5 May 2026
Julian Goldie handed Hermes Agent a 50-page SEO site brief and let it work through the project in a loop — keyword clusters, URL structure, internal links, content outlines. The output is more capable than expected and also more limited than the demos imply.
Read the summary →AI news · SEO & AI ranking · 5 May 2026
Google's Gemini Gems and NotebookLM combine into a free SEO pipeline — NotebookLM for competitive research, Gemini for drafting, with trigger-based automation to scale content output. Julian Goldie demonstrated the workflow. Here's the honest take on where it works and where it doesn't.
Read the summary →AI news · SEO & AI ranking · 2 May 2026
A model called Owl Alpha appeared on OpenRouter with no lab name attached and a 1 million token context window. Julian Goldie tested it on SEO keyword planning and content briefs. Here's what the output looked like and what you should know before using it with client data.
Read the summary →AI news · SEO & AI ranking · 2 May 2026
The latest Gemini update adds structured planning capability — Julian Goldie used it to generate a complete 90-day SEO content calendar from a single prompt. Here's what the output actually contains, why prompt structure is the deciding factor, and where it fits in a real content workflow.
Read the summary →Category
AI copy, AI design, landing pages, conversion. Related service: Web Design.
AI news · Web, content & design · 13 August 2026
Julian Goldie used Claude Code to write the site structure and Seedance 2.5 to generate motion video for the visuals — building a finished website in one session. Here is what the two-tool workflow involves and what it cannot yet do alone.
Read the summary →AI news · Web, content & design · 18 July 2026
Google AI Studio shipped Design Variations in mid-July 2026 — generate five different design versions of a landing page from a single click, with no designer or Figma account needed. The update also added free custom URLs on the ai.studio domain, an annotation editing mode, and GitHub sync. Julian Goldie walked through all four features live.
Read the summary →AI news · Web, content & design · 13 June 2026
Higgsfield's MCP server connects Claude Fable 5 to its video and image generation engine — describe a website, and Claude plans the layout while Higgsfield produces the motion graphics. Julian Goldie walked through what the finished result looks like and where you still need human input.
Read the summary →AI news · Web, content & design · 13 June 2026
Julian Goldie demonstrated Claude Fable 5 taking a single advertising brief and using Higgsfield's MCP server to produce a finished video advertisement. Claude acts as the planning layer — writing the script, defining shots, and calling Higgsfield for each visual in turn.
Read the summary →AI news · Web, content & design · 24 May 2026
Gemini Omni can accept video, audio, images and text and output new video directly. Veo 3.1 is now inside Google Vids. And Ask YouTube lets subscribers search video through natural conversation. Julian Goldie covered what each one does — here's the practical read for businesses using video content.
Read the summary →AI news · Web, content & design · 27 Apr 2026
Google's April 2026 update adds a custom report builder to NotebookLM with tone and structure controls. It shifts the tool from a research dumping ground into a first-draft production system — here's what actually changed.
Read the summary →Category
Coding with AI, scraping, APIs, iOS and Android builds, automations. Related service: Programming.
AI news · Programming & automation · 21 August 2026
Julian Goldie's comparison video tests Prime Agent against Hermes on coding tasks. The result depends on task type: Prime Agent's self-editing notebook gives it the edge on isolated repeated coding work; Hermes wins when shared memory across a wider agent stack matters more. Here is why most comparisons frame this wrong.
Read the summary →AI news · Programming & automation · 13 July 2026
Google Gemma 4 now runs fully offline inside Android and Mac apps — no internet needed, no API costs, nothing leaving the device. The Google AI Edge Gallery brings Gemma 4N to Android, Gemma Chat builds small apps on Mac, and Ollama 0.31 makes it 90% faster on Apple Silicon. Julian Goldie walked through the full setup.
Read the summary →AI news · Programming & automation · 3 July 2026
Poolside's Laguna XS 2.1 is a 33-billion-parameter mixture-of-experts coding model that runs locally and outscores North Mini Code and Claude Haiku 4.5 on SWE-bench Pro. With only 3 billion parameters active per inference, it stays fast on consumer hardware. Julian Goldie tested it across nine programming languages.
Read the summary →AI news · Programming & automation · 28 June 2026
OpenMontage is a free, open-source agentic video production system with 12 pipelines, 52 tools and over 500 agent skills. Connect it to Claude Code, Cursor or Codex and a one-sentence brief becomes a finished video — research, scripting, asset generation, timeline editing and final render all run without further input. Julian Goldie walked through the full setup live.
Read the summary →AI news · Programming & automation · 18 June 2026
Moonshot AI's Kimi K2.7 Code is an open-source agentic coding model that costs a fraction of frontier alternatives. Julian Goldie ran a full benchmark against Claude Opus 4.8 and GPT-5.5 on real coding tasks. The results are competitive on structured tasks and narrower than the headline suggests on open-ended development work.
Read the summary →AI news · Programming & automation · 16 June 2026
Zhipu AI's GLM 5.2 launched on 13 June 2026 with a 1 million token context window and MIT open weights. Julian Goldie ran five identical coding prompts through GLM 5.2, Claude Opus 4.8, and Kimi K2.7. The result is a practical comparison for teams deciding which model to use in their build stack.
Read the summary →AI news · Programming & automation · 3 June 2026
DeepSeek V4 Flash became available free on the Nous Portal, making it possible to run a full Hermes Agent automation stack without paying for tokens. Julian Goldie tested the combination on real business tasks.
Read the summary →AI news · Programming & automation · 25 May 2026
Google Antigravity 2.0 is Google's answer to Claude Code and OpenAI Codex — an AI-first development environment that plans, builds and tests applications autonomously. Julian Goldie ran a full application build from scratch. Here's what it produced.
Read the summary →AI news · Programming & automation · 8 May 2026
Moonshot AI's Kimi K2.6 is a free open-source model built for coding. Julian Goldie paired it with Hermes Agent to build a dashboard frontend from scratch — here's what the session actually produced.
Read the summary →AI news · Programming & automation · 7 May 2026
OpenAI Codex can take a keyword brief and produce a fully structured, publish-ready SEO page — headings, copy, schema, meta tags and HTML — in a single pass. Julian Goldie tested it on a real brief; here's what the output looked like.
Read the summary →AI news · Programming & automation · 6 May 2026
OpenAI's Symphony is a free open-source spec that gives Codex-based AI agents a proper operating procedure — solving the problem of agents that start well but drift off-task halfway through complex jobs. Julian Goldie covered the latest update. Here's what changed.
Read the summary →AI news · Programming & automation · 2 May 2026
OpenAI's latest Codex update takes GPT-5.5 well beyond coding — it now controls apps, drafts and sends email, builds web pages, generates slides, and runs recurring scheduled tasks. Julian Goldie tested it. Here's what's genuinely new.
Read the summary →AI news · Programming & automation · 29 Apr 2026
Xiaomi's MiMo V2.5 Pro scores close to paid models on coding benchmarks, runs free on the Nous Portal, and is already integrated into OpenClaw and Hermes Agent. Julian Goldie tested it against DeepSeek V4 — here's what held up.
Read the summary →AI news · Programming & automation · 25 Apr 2026
OpenAI's new model scored 82.7% on Terminal Bench and can code autonomously for 20 hours — but the real capability only shows inside Codex, not ChatGPT. Here's what matters for a UK developer or small team.
Read the summary →AI news · Programming & automation · 25 Apr 2026
DeepSeek V4 launches with open-source weights, free API access, and a 1M token context window. Julian Goldie tested it against GPT-5.5 and Claude Opus — strong on reasoning, less polished on design tasks.
Read the summary →If a story sparks an idea for your business, send it over. We'll tell you honestly whether it's worth doing — and what it'd cost.