Each one costs energy. By evening you're empty. The important decisions get the worst version of you.
73%
of night decisions are regretted
3 min
average freeze time on simple choices
0
apps that know when you're tired
A week with Baddle
It thinks while you sleep. It feels when you're tired. It decides when you can't.
Monday · 3:47 AM · You're sleeping. Baddle isn't.
☾
Baddle Scout
autonomous · cycle 3 · 34 nodes explored
While you slept, Scout explored your graph and found:
🔗BRIDGE FOUNDquality: 72%
Your project: auth module
⟶
Article: distributed systems
Hidden axis: "scalability as shared constraint" — your auth bottleneck mirrors the article's point about single points of failure.
🔗BRIDGE FOUNDquality: 58%
Last week: "burnout prevention"
⟶
HRV data: Tuesday dips
Hidden axis: "recovery debt accumulation" — your Tuesday HRV dips correlate with Monday overwork.
Connected your personal data with your research. You didn't ask for this.
🤔QUESTION GENERATED
"If your auth module is a single point of failure, and your energy on Tuesdays is a single point of failure — what other single points of failure exist that you haven't noticed?"
META question · waiting for morning
3:47 AM · 7 nodes added · 2 bridges · 1 question · graph growing while you sleep
Monday · 7:30 AM · Morning
B
Baddle
Polar H10 synced · coherence 0.78
88/100
daily energy
Good morning. You slept 7h 20m, recovery 82%. You're well-rested today.
Overnight Scout found 2 bridges — one connects your auth module to that distributed systems article. Worth 10 min?
Later. What should I eat?
From your fridge, 3 options:
1
Oatmeal + berries
320 kcal · 5 min · light
2
Omelette + avocado ★
450 kcal · 12 min · recommended
3
Yogurt + granola
280 kcal · 2 min · quick
decision cost: 2 energy · or I choose for you
Go with #2
Starting with code review — your coherence is 0.78, best for complex work. Lighter tasks saved for afternoon.
energy 86/100 remaining
↑
Monday · 2:15 PM · Energy drop
B
Baddle
coherence dropping · 0.41
31/100
daily energy
⚠ Coherence dropped to 0.41. Energy 31/100. You've made 14 decisions today.
Heavy decision "change tech stack?" — move to tomorrow morning?
Yeah, tomorrow
Moved. Tonight Scout will think about it in background. Meanwhile — those 2 bridges from last night are still unexplored. Low energy cost, high insight potential.
recovery mode · no pressure
↑
Wednesday · 11:00 AM · Meeting prep
B
Baddle
analyzing both sides
meeting in 3h
I need arguments for the architecture meeting. Microservices vs monolith.
FOR MICROSERVICES
Independent scaling, team autonomy, fault isolation. Your team of 12 can own services.
confidence: 74%
AGAINST
Operational complexity, distributed debugging, premature at your scale.
confidence: 68%
SYNTHESIS
Modular monolith now, with clear boundaries. Extract services only when a module needs independent scaling. Best of both.
balanced · confidence: 81%
3 perspectives verified · 15 min instead of 2 hours
↑
Wednesday · 11:47 PM · Baddle saves you
!
Baddle
proactive alert
9/100
energy critical
⚠ You have an angry email draft open. Energy 9/100, coherence 0.28. Decisions at this level have a 73% regret rate in your history.
Save draft for morning
Send anyway
...fine. Morning.
Saved. Reminder at 9 AM when your coherence is usually 0.75+.
you'll thank me tomorrow
↑
Thursday · 7:15 AM · Before you wake up
B
Baddle
morning briefing ready
☀️ Today's briefing
🌦️
14°C, rain after 3pm
Take umbrella. Move outdoor run to morning
👕
Client meeting at 11 + gym at 7pm
Navy blazer + white shirt. Pack gym bag
🍳
Eggs + toast + coffee
You skip breakfast on Thursdays 60% of the time → headache by 2pm
💡
Pattern detected
Last 3 Thursdays: skipped breakfast → energy crash at 2pm → postponed tasks → Friday overload. Breaking the cycle today.
↑
Thursday · 11:30 PM · Impulse control
$
Baddle
spending guard
3 items in cart. Total: $380. It's 11:30 PM, energy 14/100.
Your impulse purchase history:
Night purchases (energy <20)68% returned or regretted
Morning purchases (energy >70)91% kept and satisfied
Save cart for morning
Buy anyway
saved you $380 last month with this feature
↑
Friday · 10:00 AM · Big decision, no guessing
B
Baddle
bayesian mode
71%
P(switch stack)
Should we switch tech stack?
Let me ask a few things first. Team size? Current stack age?
12 people, 3 know the new framework, current stack is 4 years old
Initial estimate: P = 30%. Now adding observations:
"3 people already know the new framework"supports → 58%
"Migration takes 3 months"contradicts → 38%
"Current stack EOL in 1 year"supports → 71%
3 observations · decision on numbers, not feelings
↑
Sunday · 9:00 PM · What changed this week
B
Baddle
weekly summary
Your week
5/7
exercise streak
6/7
meals on time
62
decisions made
+8%
HRV trend (month)
Mornings strong (avg 85 energy), afternoon dips (avg 35). Tuesday was hardest — 3 complex decisions drained you by noon.
Scout found 7 bridges this week. 2 became active ideas. Graph grew by 34 nodes.
Recommendation: move hard decisions to before 11am. Your data shows 2.3x better outcomes in morning sessions.
↑
Tuesday · 1:15 PM · "Finished lunch"
B
Baddle
🟢 healthy_load · workspace: personal
64/100
daily energy
Just had lunch — pasta with salad
♻✓
«eat 3 meals/day» — 2/3 today
no pasta question, no mode switch — router detected "fact/instance" and matched to your recurring habit.
Going to the gym
🎬
Tracker started: Gym (health)
↳ also +1 to «exercise 3× per week» (1/3 this week)
↑
Thursday · 8:45 PM · System noticed a pattern
💡
Baddle · Observation
noticed 3 Thursdays in a row
Checking your patterns, I noticed something:
♻ Create a habit?
💡 Because: last 3 Thursdays you skipped breakfast (usually you eat mornings 86% of days).
Breakfast every morning at 8:00 · 1/day · food
pattern detector · 1 suggestion today · max 2/day, no spam
Saturday · 9:00 AM · Plan for the day
B
Baddle
📋 plan · 2/5 · workspace: personal
95/100
rested
09:00Morning walk♻health
14:00Call with parentssocial
— HABITS —
♻Drink water 4×/day2/4 /day
♻Exercise 3×/week1/3 /week
plans, habits, constraints — one unified view
How it works
Under the hood: a graph of thoughts, adaptive controller, and 14 modes built from 5 primitives.
baddleResearch mode EXECUTIONP = 0.72
Click any node to inspect
Daily Energy
67/100
12 decisions today~5 left
HRV Coherence
0.73 RMSSD 42ms
Verified thoughts
Open-source enables local AI92%
Privacy drives adoption87%
Enterprise resistance61%
Regulation slows progress32%
Every situation has a mode
14 modes, one engine. You describe the situation — Baddle picks the right mode.
"BMW or Tesla?" → verifies both, picks winner with reasons
"Pros and cons of moving" → forced counter-argument on each side
"What to eat?" → picks from your fridge, considers energy
"Will it rain?" → Bayesian update with each observation
"Work / family / health" → balances all three, tracks drift
"Exercise every day" → streak + HRV correlation + nudges
"Understand crypto" → explores until novelty exhaustion
"Just wander" → finds bridges between your ideas at 3 AM
⚡
Forced counter-argument
Every thought tested from both sides. Not opinion — a weighted position with numerical confidence.
🔗
Hidden connections
Two unrelated ideas in, hidden axes out. Insights you can't see yourself. Works while you sleep.
❤️
Body as input
HRV sensor syncs with your state. High coherence — explore. Dropping — focus or rest.
🎯
Energy economy
Every decision costs energy. Baddle tracks your budget and protects you from bad choices when you're empty.
🔒
Self-hosted
Your device. Your data. 8B model, 4GB RAM. No cloud, no subscription. Works offline.
🧠
Neuroscience-backed
Active Inference, Bayesian brain, dialectics. Independently discovered, then found in 30-year-old research.
How your brain works. How Baddle works.
Your brain constantly predicts what will happen next. When reality doesn't match — you pay attention, learn, adapt. Baddle does the same.
Too chaotic? → Baddle narrows focus, gets precise
Too predictable? → Baddle opens up, searches wider
Just right? → Flow. The best work happens here