einstein — a grade assistant that explains what changed

01 — core capabilities

Three things that change how grades feel.

Built on the principles in docs/einstein-app-overview.md — transparency over opacity, files over servers, explanation over anxiety.

The Delta Engine

Every new score triggers an instant autopsy: which category shifted, how much leverage it had, and where your trajectory is pointing. Read how it works in docs/architecture.md.

delta calculation

What-If Lab

Enter hypothetical scores, set a target grade, and see exactly what you need on the final. Dynamic drop-lowest rules update in real time.

simulation

File-First Sovereignty

All course data lives in transparent JSON files in your workspace — no proprietary database, no remote server, no vendor lock-in. Inspect, back up, or edit directly. Schema defined in docs/data-schema.md.

local storage

02 — app comparison

The same grade, four very different tools.

A deeper cut of the README comparison: what each kind of tool actually does, where it stops being useful, and which one answers the question you're asking at 11pm.

standard lms

Canvas, Blackboard, PowerSchool

Holds the official gradebook. It is the source of truth because your teacher types the scores there.

Wins at: Being authoritative. Every other tool is a copy of it.
Breaks down at: Explaining itself. A zero for something you haven't turned in lands in the total like a real score, and category weights stay opaque.

Trust its numbers, not its narrative.

ad-hoc calculators

RogerHub, Excel sheets

One formula for one question: what do I need on the final to keep my A?

Wins at: Speed. No account, no sync, no permission — type numbers, get an answer.
Breaks down at: Anything past a single course and a single term: no history, no drop rules, no GPA picture.

Great at the one question, blind to the semester around it.

other grade apps

GradeWay, Gradiate

Mobile apps that pull grades from a supported portal and layer GPA goals and alerts on top.

Wins at: Polished mobile UI, notifications, and zero manual entry once a portal is linked.
Breaks down at: Portal dependence (HAC, PowerSchool), closed data, and what-ifs that inherit the portal's own category math.

Convenient if your school is on the list. Einstein doesn't need a list.

einstein

Explain first, then simulate

Reads course files you own, decomposes every grade change, and models the semester split explicitly — categories weighted inside the coursework share, the final weighted on its own.

Wins at: Saying why a number moved, and exactly what the next one has to be.
Breaks down at: It's early: there's no portal sync yet, so a course starts as a file you write or import.

Different trade: less automation, more explanation, and the data stays yours.

Capability	Standard LMS Canvas, Blackboard, PowerSchool	Ad-hoc calculators RogerHub, Excel sheets	Other grade apps GradeWay, Gradiate	Einstein
Delta explanation	❌ Raw number changes only	❌ No history to compare	⚠️ Trend lines, no causal breakdown	✅ Plain-English impact breakdown
What-if simulation	⚠️ Clunky or read-only	⚠️ One isolated formula	⚠️ Score mocking tied to portal sync	✅ Multi-course sandbox
Category weight math	❌ Opaque category grouping	⚠️ You build the formula yourself	⚠️ Mirrors LMS categories	✅ Automatic weight & point resolution
Coursework vs final split	❌ Baked in, never shown	⚠️ You model it by hand	⚠️ Whatever the portal assumes	✅ Explicit: categories × 85% + final × 15%
Drop-lowest rules	⚠️ Frequently miscalculated	❌ Manual spreadsheet logic	⚠️ Mirrors LMS, skipped in what-ifs	✅ Dynamic drop recalculations
Curve & extra credit	⚠️ Applied silently, if at all	⚠️ Manual offsets	⚠️ Limited custom curve math	✅ Modeled and explained
Assignment leverage	❌ Not shown	❌ Not shown	⚠️ Basic priority hints	✅ Ranked by impact on the final
Cross-course GPA planning	❌ Term GPA only	❌ One course at a time	⚠️ Per-course GPA goals	✅ Trade-offs across courses
Data privacy	❌ School-managed	⚠️ Ad-supported sites	⚠️ Portal credentials in a third party	✅ 100% local, offline
Open file formats	❌ Institutional database	⚠️ Proprietary spreadsheets	❌ Closed app data	✅ Human-readable JSON/Markdown
Platform freedom	❌ Portal lock-in	✅ Any browser	❌ Supported-LMS shortlist	✅ Portal-agnostic
Setup cost	✅ None, it's already there	⚠️ Build the formula first	⚠️ Account, link portal, grant access	⚠️ Write or import a course file
Price	✅ Paid by tuition	✅ Free	⚠️ Free tier or subscription	✅ Free & open source

Reflects public feature sets as of October 2026 — corrections welcome via a GitHub issue. Einstein's honest losses are the last two rows: no portal sync yet, and a course file to set up.

03 — interactive preview

The What-If Lab

Before you turn anything in, run the numbers. Change the final exam projection and watch the grade, letter, and required score recalculate instantly.

Organic Chemistry II — Semester Projection preview

Category weights sit inside the coursework share. The final exam is weighted separately, on top of it — the same way most syllabi actually read.

Category	Avg	Weight in coursework
Major avg
Minor avg
Other grade avg
Coursework	--	90
Comprehensive Final	simulating	15%

Semester coursework weight

% of the semester

The categories above are weighted inside 85%. The final exam takes the remaining 15%.

--

Project final score

84%

Projected overall

--

Required for 90%

--

Term GPA (4.0)

--

Five other courses average 3.60 — this is the sixth. Preview only: the full app runs this math across every course, assignment, and term.

04 — documentation

Built with evidence, not guesswork.

The architecture, data schema, and vision are documented in the repo itself — no hidden specs, no external wikis.

vision

Run the numbers before the next grade lands.

Try the lab, read the vision, or clone the repo and keep your data in files you own.

Try the lab Read the vision Why I built this

A grade assistant that explains what to do next — not just the number.