Trust broken by lying or sneaking out is rebuilt gradually, not restored in one sentence — a realistic step-by-step protocol.

Before any reply, summarize your speaker's argument and verify with them you got it before disagreeing.

When provoked, don't reply for 90 seconds — the neural wave scientifically subsides.

Place every stakeholder in one of four boxes: high-power/high-interest gets close management, high-power/low-interest gets kept satisfied.

Don't lie — state real scarcity in your time or offer.

As mentor: organize a 60-min GROW group session for 4 people.

Move attention from feet to head, noticing each area without changing it.

Tense then relax each muscle group — one by one.

Don't cling to position — find interest beneath. Flexibility without yielding.

Notice topics he avoids — usually contain gold.

Listen to understand not prep your reply — shifts every conversation.

Ask "why?" five times on a recurring fight's root — discover the real cause.

Your small leader words exclude or include. 10 simple swaps shift your team's culture in 60 days.

Ask your child with curiosity about his apps and digital friends — before discovering by surprise.

Before applying to an internal role, schedule a 20-min exploration chat with the target team's manager.

After hearing your partner, reflect back three sentences: content, feeling, and underlying value.

Sign a civility pact with your team: no interrupting, no sarcasm, no replying angry.

When stakeholder A wants X, B wants opposite: don't pick — surface underlying interests.

Before relying on AI in specialist field, show to human expert — 30% of time uncovers core error.

After a Socratic learning session, ask AI to generate retrieval questions at increasing intervals (1 day, 3 days, a week) instead of relying on random memory.

Whenever a prompt works exceptionally well, save it immediately with a simple tagging system instead of relying on memory or digging through old chats.

For critical information, ask two different models (Claude and GPT, say) the same question independently, and compare before relying on either.

Imagine a friend in the same situation — what would you tell them? Tell yourself.

Within a week of any project delivery, gather to document what worked and what failed — before memory evaporates.

Before presenting your proposal in the formal meeting, win 2-3 influencers individually — don't surprise them with a public vote.

Before any open forum, write the 10 hardest expected questions and answer them in writing — never improvise live.

Track for three consecutive meetings: who spoke, for how long, and how often they were interrupted — hidden patterns show in numbers, not impressions.

Deliberately (not accidentally) diversify the interview panel and define written evaluation criteria before seeing any candidate — consistency protects against individual bias.

Neutrally review your family routine: is rule flexibility the same for every child, or does it silently vary by who bothers you more?

Learn 4 signals that reveal the model invented info instead of citing documented fact.

The model does not know what happened after its training cutoff. Know each model's cutoff.

After each output, ask: "What are 3 weaknesses in this answer? How to improve?" — use its intelligence against itself.

Any question on recent events/prices/releases: check model training date — ignoring = confident wrong answer.

Use AI for a bad-fast initial draft, then YOU rewrite completely. Blank page is the killer.

Before using any specific AI-provided fact (number, name, date, quote), apply a 5-question checklist in under a minute.

Don't try to stop scrolling through willpower alone — replace the trigger itself (boredom, anxiety) with a predefined alternative action within 3 seconds.

Review every account you follow with one strict criterion: does it add real value, or does it drain your mood? Neutrality isn't reason enough to stay.

Set a pre-defined protocol for sorting comments: constructive criticism deserving a reply, or passing hate deserving only ignoring and blocking.

Try a full week using a browser (Brave/Firefox) and search engine (DuckDuckGo) that reduce tracking by default, and compare the experience.