On your device
The model downloads once and runs in your browser. The file you give it is not uploaded. It is limited by your own hardware, so a long recording takes minutes, not seconds.
Meeting TranscriberBackground RemoverBrowse the guides
Filters
Tools & GuidesWhat an AI-Native Issue Tracker Adds to Jira
A practical guide to executable AI work items, human approval boundaries, evidence-led status changes, and a low-risk pilot beside Jira.
Tools & GuidesMove Off Trello Into One AI-Readable Project
A practical guide to moving from scattered Trello boards into one AI-readable project with a canonical queue, owner files, decision history, and a safe archive.
Tools & GuidesRun Your House as an AI Project
Build a private, AI-readable home system for documents, maintenance dates and decisions, with clear rules for scanning, backups and human approval.
Tools & GuidesRun Your Professional Network as an AI Project
Build a private AI-readable system for contacts, events and follow-up while keeping personal data and every outbound action under human control.
Tools & GuidesLocal AI or Cloud AI? Choose Per Job
A practical guide to choosing local, cloud, or hybrid AI per workload, using data boundaries, demand, quality tests, and total operating cost.
Tools & GuidesHow to Choose Between AI Context Methods
A practical comparison of broad context, hard curation, and the hybrid routing pattern, with a controlled test for choosing between them.
Tools & GuidesHow to Structure AI Projects With PARA
Adapt PARA for Claude Code and Codex with clear folder roles, owner files, permissions, decision logs, archives, and a local-data boundary.
Tools & GuidesBuild a Personal Budget With an AI Agent
A practical guide for building a personal or household budget from raw bank exports using a local AI coding agent, with a free empty Excel workbook and copy-paste prompts.
AI ExplainedAI Orchestration Is Like a Dispatch Desk
A practical analogy for understanding AI orchestration as the coordination layer that routes work between tools, agents, data, and people.
Local AI
Where the model runs is a decision
Every AI feature runs a model somewhere. Where that is decides what leaves your computer, what it costs to run and what it can handle.
On a server
Your text goes to a hosted model. It is more capable and needs nothing installed, but what you paste leaves your device. Our SQL, Python, Terminal and Power BI assistants work this way.
Free updates
New guides and tools by email
One email when we publish a new guide or tool. Free, no account.
Who writes this
Nikita Goncharenko — AI Integration
Nikita Goncharenko is a Belgium-based digital delivery professional who connects analytics, execution, websites, and AI-assisted workflows. He builds his own AI-assisted tools and writes up what works: the workflow setups, the prompts that hold up, and the ones that quietly waste time.
Read the full profileFAQ
Quick answers on AI Integration
Short answers about this topic. Open only what you need.
Where should a small team start with AI?
With one repeated task, not a strategy. Pick something the team does weekly, write down the steps it already follows, and automate that — the value shows up in the second week, not in the pitch.
Which AI tools do you actually use?
The ones that fit an existing workflow: assistants for drafting and code, small local tools for file and image work, and a handful of scripts. The tools on this site are the ones that survived daily use.
When does AI not help?
When the process underneath is unclear. If nobody can describe the steps, an assistant will produce confident output that nobody can check, and the review costs more than the task did.
Do you use AI to write these articles?
AI helps with structure and drafting. The work described, the numbers and the conclusions are from real projects, and anything that cannot be backed up does not get published.