Experiments
Things we are trying before pretending we know the answer.
A running notebook of what works, what fails and what we would do differently next time.
AI
Can one good AI hire remove five software subscriptions?
Status: worth testing.
OPERATIONS
How much back-office work can an SME automate without becoming brittle?
Status: learning by doing.
PEOPLE
What does an AI-native employee actually look like?
Status: role definitions are moving quickly.
DATA
Can internal knowledge become genuinely queryable?
Status: yes, provided the source data is not a swamp.
VOICE
Can AI follow up leads quickly without sounding dreadful?
Status: commercially useful when designed carefully.
CHARITY
Can a services company route 70% of revenue directly to impact?
Status: that is this experiment.
A useful rule
Cheap experiments are underrated. Expensive certainty is usually available from consultants.
What belongs here.
- Actual tests.
- Clear hypotheses.
- What happened.
- What it cost.
- What changed our mind.
- Enough honesty to publish failures.
Bring us something worth thinking about
Got an experiment worth running?
If it teaches us something useful even when it fails, it may qualify.
Tell us what is missing