Skills That Compound

Same AI.
100x the output.

Everyone has the same AI. The 100x people taught theirs how they work — with skills. Distill captures your process and turns it into reusable skills automatically.

Join the Waitlist See how it works
Good evening, Sumant
Last session: investigating onboarding drop-offs · 12m ago
12
This Week
277
Knowledge
7
Streak
23h
Time Saved
Recent Workflows View All
Investigating drop-offs in the Onboarding funnel
Apr 5, 2026 · 8 steps · 14 min
Adding new events to track for new features
Apr 4, 2026 · 5 steps · 9 min
Creating a brand for my app
Mar 28, 2026 · 12 steps · 22 min

Your AI doesn't know you

You spend all day solving hard problems — debugging, designing, configuring, researching. By evening, the work is done but the knowledge is gone.

Tomorrow, you'll solve a similar problem from scratch. So will your AI — because it has no memory of the process you just perfected.

The bottleneck is never the model's intelligence. Models know how to reason. They fail because they don't understand your data, your conventions, your particular shape of problem.

Knowledge retained
~5%
of workflow knowledge captured by most knowledge workers
The productivity gap
100x
between people using the same AI — with vs. without skills
The fix
Skills
Reusable procedures that teach YOUR AI how YOU work
"
Every skill you write is a permanent upgrade to your system. It never degrades. It never forgets. And when the next model drops, every skill instantly gets better.
— Garry Tan, President of Y Combinator

Simple app. Rich skills.

The secret isn't a smarter model. It's giving your AI the right context, at the right time, without drowning it in noise.

Skills · where the value compounds
90%

AI capabilities that learn your patterns

Real-time observation, session synthesis, drift detection, pattern recognition, and self-improving rules. Each skill is reusable and gets smarter with every session you record.

The App · lightweight by design
7%

Captures screen, routes to skills, stores results

The macOS app does four things: captures your screen, pipes data to the right skill, stores results, and enforces privacy. It stays simple so the skills can be rich.

The Reliable Base
3%

Screenshots, timestamps, storage, export

Same input, same output. Every time. Screenshot capture, local database, markdown export. The foundation everything runs on.

From screen to skill in three steps

Record your work. AI extracts the knowledge. Skills compound over time.

Step 01 · Capture

Hit record. Work normally.

Start a session. Distill captures your screen at optimized intervals while AI watches in real-time — extracting observations every 10-15 seconds. Zero performance impact, zero friction.

Recording workflow 14:23
Live Observations
User navigated to Amplitude dashboard and selected the retention chart. 14:21
Switched to VS Code — editing analytics event configuration file. 14:19
Opened terminal, ran test suite for event validation. 14:17
Step 02 · Synthesize

AI turns hours into a brief.

When you stop, the model reads every observation and produces a structured report: phases, steps, screenshots, key decisions. One-page intelligence from hours of work — the documentation you'd never write manually.

Investigating Onboarding Drop-offs
3 Phases · 8 Steps · 14 min · 72% confidence
Phase 1: Setup & Navigation Complete
Opened Amplitude retention dashboard
Configured date range to last 30 days
Phase 2: Investigation Complete
Traced user event streams for drop-off cohort
Identified phantom open signal in session data
Cross-referenced with push notification timing
Phase 3: Documentation 3 Steps
Step 03 · Compound

Skills compound over time.

Workflows become reusable skills. After 10+ sessions, the system sees patterns you can't — which approaches work, where you waste time, when your focus fades. It proposes new rules. You approve. The system gets smarter.

Skills
5
Skills
12
Gotchas
74%
Confidence
3
Agent Runs
Investigate Onboarding Funnel Drop-offs
analyticsintermediate20min
Set Up Amplitude Retention Chart
analyticsbeginner8min
Debug CSS Layout in Chrome DevTools
frontendadvanced15min

Five capabilities that get smarter

Each capability learns from your work. The more you use Distill, the better it understands how you think.

Real-Time Watching

AI observes your screen every 10-15 seconds. Extracts what you're doing, which tools are active, what changed.

● Live

Session Synthesis

Reads all observations and produces a structured brief: phases, steps, tools, decisions. One-page intelligence.

● Live

Drift Detection

Detects when you're stuck or off-task. Nudges with context-aware suggestions — not generic productivity alerts.

○ Coming Soon

Pattern Recognition

"You debug 2x faster when you reproduce first." Insights that only emerge after watching you work over weeks.

○ Coming Soon

Self-Improvement

After every 10 sessions, the system proposes new rules from your patterns. You approve. Skills get smarter. Permanently.

○ Coming Soon

Build Your Own

Skills are just procedures. Write one, and your AI learns a new capability tailored to your specific workflow.

The system that rewrites itself

Other tools analyze the same way on day 1 and day 1,000. Distill gets better at understanding YOUR work by watching YOU work.

Week 01

Capture your work sessions

Hit record, work normally. AI captures observations in real-time, synthesizes structured steps when you stop.

Week 02

Patterns start emerging

After 10+ sessions, the system sees what you can't: where you waste time, which approaches work, when focus fades.

"Debugging sessions that start with reproduction are 2x faster than ones starting with log reading."
Week 03

System proposes new rules

The self-improvement skill reads your sessions and proposes rules. You approve. They get written into active skills.

"When app-switching exceeds 3 apps in 60s, user is likely stuck. Suggest: 'What are you trying to figure out?'"
Week 04+

Your AI knows how you work

Every session feeds the loop. Skills get smarter. Nudges get more relevant. Your personal AI isn't generic anymore — it understands your craft.

Knowledge workers who build

Developers & Engineers

Turn debugging sessions, architecture decisions, and deployment workflows into reusable skills your AI can reference next time.

Product Managers

Capture feature explorations, analytics investigations, and stakeholder workflows. Transform tacit knowledge into teachable skills.

Designers & Creators

Record your design process — explorations, iterations, decisions. Build a library that makes AI understand your taste.

Anyone Learning a Craft

Like Strava for runners, Distill shows you how you work. See patterns, track improvement, get nudges that matter.

Stop losing
what you learn.

Every workflow you capture is a permanent upgrade. It never degrades. It never forgets. The system compounds.

Join the Waitlist
macOS · Free beta · Shipping 2026