Leo Harrison
TutorialsWhat Is an Agent OS? Architecture, Components, and Use Cases
Agent OS traces to a real design: the AIOS kernel with scheduling, context, memory, storage, tool and access management. What exists today, and how to read the claim.
Leo Harrison
October 13, 2026
ToolsOpenAI dots, explained: always-on agents, and what a team still needs around them
OpenAI dots are always-on ChatGPT agents on GPT-6 Astra with their own cloud computer. What they do well, and what a team running several agents still needs.
Leo Harrison
October 13, 2026
TutorialsCheap Browser Agents with Haiku 5.5: The Review Queue Is the New Cost
Haiku 5.5 scores 72.4% partial credit but 37.1% strict on OSWorld. How teams design browser-agent Tasks, review and submit gates around that gap.
Leo Harrison
October 8, 2026
Human-Agent CollaborationOpenAI's Agents API Runs the Loop for You. Who Runs the Agents?
OpenAI's Agents API hosts the Codex harness and, since DevDay, computer use with approval events. What stays with your team: assignment, review, acceptance.
Leo Harrison
October 2, 2026
TutorialsModel Handoffs Lose the Reasoning: What Claude Sonnet 5.5 Changes for Multi-Agent Work
Claude Sonnet 5.5 binds thinking blocks to the model, conversation and account, so reasoning never survives an agent handoff. What the Task must carry instead.
Leo Harrison
September 29, 2026
ToolsClaude Opus 5.5 vs GPT-6 Sol: Choosing Your Crew's Default Runtime
Claude Opus 5.5 and GPT-6 Sol shipped the same day. Compare them on cost per completed task, how long they hold a task, and what they do when stuck, and note OpenAI benchmarked Sol against Opus 5.
Leo Harrison
September 23, 2026
TutorialsGPT-6 Sol as a Crew Leader: When the Leader Seat Stops Being the Expensive One
GPT-6 Sol at $2 / $10 makes the Crew leader seat affordable on every Crew. What leader-first execution actually costs, when Astra or Opus 5.5 still earns the seat, and the review queue a price cut does not fix.
Leo Harrison
September 23, 2026
Ai And Agent Trends
Why Your Coding Agent Looks Stuck When It Is Thinking: Time to First Token
At max reasoning, GPT-6 Sol was clocked at 102 seconds before its first token. A thinking agent and a hung agent look identical in a terminal. What a run state has to separate, and the per-Task timeout that quietly kills long thoughts.
Leo Harrison
September 23, 2026
ToolsIs Kie.ai Legit? A Developer's Read on the One-Key Aggregator
What Kie.ai is, who runs it, what the reseller model does and does not guarantee, and the specific risks worth pricing in before you build on it.
Leo Harrison
September 22, 2026
Human-Agent CollaborationCoding Agent Permissions: What an Agent Should and Shouldn't Be Allowed to Do
Coding agent permissions in three layers: runtime tools, the machine and its credentials, and task scope. The failure mode at each step, plus a checklist for what an agent should and shouldn't do.
Leo Harrison
September 20, 2026
Human-Agent CollaborationOperations Teams and AI Agents: Triage, Automate, Close the Loop
Operations work arrives unpredictably. Here's how Sharkly splits it: Triage for the unpredictable third, Automations for the recurring rest, and results that return to the Task for human review.
Leo Harrison
September 15, 2026
ToolsAgentic Project Management: Is It Worth the Review Cost?
The honest case for agentic project management: throughput and traceability that hold up, the review bottleneck and automation blast radius nobody mentions, and the limits.
Leo Harrison
September 10, 2026
TutorialsHow to Manage Multiple Codex Agents Without Losing Track
Running three or more Codex agents means losing track of what finished, failed, or needs review. Here's the manual worktree setup, where it stops scaling, and a checklist you can run today.
Leo Harrison
September 10, 2026
Human-Agent CollaborationRunning coding agents remotely: your server, your laptop, or both
Run coding agents on a remote server, your laptop, or both. The DIY SSH-and-tmux setup with its failure modes, and how a connected Computer keeps every run visible.
Leo Harrison
September 9, 2026
TutorialsGPT-6 Astra as Your Crew Leader: A Mixed-Model Agent Team in Sharkly
Put GPT-6 Astra in the leader seat of a Sharkly Crew and let cheaper models implement in parallel worktrees. How leader-first execution, mixed Runtimes, and one Task record fit together.
Leo Harrison
September 5, 2026