chengtao. Connect on LinkedIn

Cheng2

A second brain. Notes, research and decisions kept as plain files that any model can pick up, so the context carries across models instead of being re-explained every time.

The workspace itself stays private. Here is the shape of a week in it.

Private workspace

One week, two models

example

Mon Claude Code
"Plan the October trip around the two weeks I'm off." Dates go to context/, the plan to travel/.
Wed Codex
"Which nights did we settle on for Kyoto?" Reads Monday's file. Answers. Nothing re-explained.
Fri either
"Book it." The decision and the why land in decisions/, dated, for next time.

Rack Runner

A personal training app. The plan is updated automatically by AI in the back end. The front end is only what I need at the gym: the next session, the sets, and what I lifted last time.

Live

Open Rack Runner
A white porcelain kettlebell on a plaster surface, lit from the upper left.
Rack Runner home screen: the week's plan, Lower A back squat on Monday through Zone 2 on Saturday, with how long ago each was done.
Rack Runner mid-session: back squat, set one of four, with weight, reps done and an RPE selector above a Log Set button.

The week's plan, and the session it opens.

The one I use most

FinancePlanner

Ask it a question in plain English and it runs the projection against the real numbers.

Its own first suggestion is "How long until we can retire?", answered from the same thirty-year projection the dashboard draws.

Live, sign-in required

Open FinancePlanner View full size
finance.chendy.org
FinancePlanner at full size. The advisor panel offers questions such as how long until we can retire and whether to salary sacrifice more to super, beside net worth, buffer and goals tiles and a thirty-year projection.

Built with React, Fastify and SQLite. The advisor runs against the same database the dashboard reads.

Every name and figure shown here is scrubbed demo data.

Chengtao Liang.

chengtao.

AI and data specialist, eight years across neuroscience research and commercial analytics. I've worked as a consultant across AI, analytics engineering, data analysis and research, and these days I help teams actually use AI and build the systems that make it stick.

The research years are where the habit of testing things properly comes from. Most of what's on this page started as something I wanted for myself and couldn't find.

If any of this is useful to you, I'd like to hear about it.

Connect on LinkedIn