Back to The Forge

Knowledge Forge

Guides and news on AI agents for non-developers — release notes, workflow checklists, permissions, and review routines for small businesses and schools.

Start here

Choose a first workflow

A guide to a first safe AI workflow — without touching customer data.

Read the guide

Check your AI readiness

Seven quick questions that show where you stand and what comes next.

Take the test

Use a worksheet

Practical templates and checklists to download and use right away.

To the resources

Articles

Miro AI for beginners: turn workshop ideas into three priorities
Prompt EngineeringTool Forge

Miro AI for beginners: turn workshop ideas into three priorities

Have Miro AI propose themes, correct the grouping, and vote for three priorities with a clear prioritization receipt.

Read more
When AI can write in Microsoft 365: draw the permission map first
Agentic AITool Forge

When AI can write in Microsoft 365: draw the permission map first

Map the action, object, identity, data flow, approval and recovery path before an AI add-in can modify Excel or triage Outlook.

Read more
OpenAI Codex release notes: 0.145.0 imports Cursor and Claude Code work
Tool Forge

OpenAI Codex release notes: 0.145.0 imports Cursor and Claude Code work

OpenAI Codex release notes for 0.145.0: Cursor and Claude Code users can import supported configuration, while the app server gains searchable paginated history.

Read more
Claude Code release notes: 2.1.217–2.1.218 bounds the agent queue
News

Claude Code release notes: 2.1.217–2.1.218 bounds the agent queue

Claude Code 2.1.217–2.1.218 caps concurrent subagents, stops work at the budget limit, and moves code review into the background for teams running agent workflows.

Read more
Your AI support replies fast. Measure which conversations stay stuck.
Agentic AITool Forge

Your AI support replies fast. Measure which conversations stay stuck.

Intercom can now measure snoozed time in support. Use it to find waiting that fast first replies and strong AI numbers otherwise hide.

Read more
Copilot in Excel for beginners: compare budget to actuals
Prompt EngineeringTool Forge

Copilot in Excel for beginners: compare budget to actuals

Compare budget to actuals in a workbook copy. Have Copilot propose the plan, add one variance column, and check three rows.

Read more
Background AI jobs: a safety card for schedules, stop controls and fallback
Agentic AIMindset Forge

Background AI jobs: a safety card for schedules, stop controls and fallback

A practical safety card for unattended AI jobs: owner, trigger, duplicate protection, budgets, stop controls, audit trail and manual fallback.

Read more
Google Antigravity release notes: CLI 1.1.5 controls models and reasoning effort

Google Antigravity release notes: CLI 1.1.5 controls models and reasoning effort

Google Antigravity release notes for CLI 1.1.5: choose reasoning effort with /effort, pin stable model slugs, and assign the right model to subagents.

Read more
OpenAI Codex release notes: alpha adds audio and classifies multi-agent v2 as stable
Tool Forge

OpenAI Codex release notes: alpha adds audio and classifies multi-agent v2 as stable

OpenAI Codex release notes for 0.145.0-alpha.28: tools can pass audio to the model and multi-agent v2 is classified as stable, though it remains off by default.

Read more
Claude Code release notes: 2.1.216 resumes the right agent
News

Claude Code release notes: 2.1.216 resumes the right agent

Claude Code release notes for 2.1.216 restore the right agent role, tool boundaries, and worktree after a pause for teams running parallel AI workflows.

Read more
Gemini in Docs now speaks Swedish. Start with your best old proposal.
Prompt EngineeringTool Forge

Gemini in Docs now speaks Swedish. Start with your best old proposal.

Google has expanded Gemini in Docs with Swedish. Use an approved proposal for structure and a current fact sheet as the source for the next reviewable draft.

Read more
AI Masterclass Podcast Week 29: when AI agents start finishing the job

AI Masterclass Podcast Week 29: when AI agents start finishing the job

Week 29 shifted the focus from smarter answers to agents that run code, use credentials, and deliver finished files. That makes permissions, sandboxes, and approvals part of the feature itself.

Read more

The Forge newsletter

Get new articles in your inbox

Pick the topics you care about. No noise, at most one email a week.

Get new articles in your inbox

We follow GDPR. Unsubscribe anytime.

Need a hand turning reading into action?

Mindset Forge, Tool Forge, Skill Forge — pick the forge that matches where you are right now.

See the service