Overview
Current progress, meaningful changes, data quality, and the areas most worth inspecting next.
Private applicationDesktop firstRead only source access
NutriTrackInsights
The daily log becomes evidence.
NutriTrack Insights is the desktop analytical companion to NutriTrack. It reads the same history without changing it, then organizes the evidence into reports, long-range patterns, comparisons, wellness relationships, and projections with explicit limits.
Built as a private analytical workspace. It is not a medical or diagnostic system.
InsightsProgressTrendCoverageQualityRecent paceWindow
Evidence brief
Calculations first. Interpretation second. Confidence and caveats stay visible.
One product familyNutriTrack records the day.
NutriTrack is the mobile-first operational application. Insights is the separate read-only workspace for reviewing what the accumulated data actually supports.
The workspace
Six views, each answering a different question.
The application separates summary, investigation, comparison, and forecasting so one dashboard does not pretend to answer everything at once.
Reports
Weekly, monthly, custom, and since-first-food-log reporting with privacy controls and saved snapshots.
Trends
Long-range weight and nutrition patterns, adherence bands, weekday effects, meal-slot contribution, and behavioral phases.
Compare
Goal-normalized comparisons that focus on each person's own targets and habits rather than raw body-weight rankings.
Wellness
Relationships among mood, movement, body state, nutrition, logging behavior, notes, and later scale outcomes.
Projections
Goal timing, maintenance estimates, calorie scenarios, plateau detection, and confidence-aware forecast ranges.
Analysis model
Evidence first. Narrative second.
AI is not asked to invent the analysis. Deterministic calculations establish the evidence, then AI can translate selected results into a concise brief without changing the underlying numbers.
- 01Read the source
Insights loads user, food, weight, and check-in data from the NutriTrack spreadsheet.
- 02Calculate the evidence
Application code computes trends, coverage, phases, threshold bands, correlations, plateaus, and forecast inputs.
- 03Interpret selectively
AI briefs summarize supplied evidence and are instructed not to recompute metrics, diagnose, prescribe, or imply causation.
- 04Keep limits visible
Confidence, sample size, data completeness, and caveats remain part of the result rather than being hidden in fine print.
System boundary
Read-only where it matters.
Insights is deployed separately and does not write to the NutriTrack source database. Its own spreadsheet holds only the analytical application's configuration, logs, AI usage, and saved report snapshots.
Source systemNutriTrack spreadsheetFood, weight, users, check-ins
Read only→
AnalyticsNutriTrack InsightsCalculations, reports, narratives
Writes only here→
Insights storageSeparate spreadsheetConfig, logs, usage, snapshots
Reports
Designed to leave the screen.
Reports can be generated for standard periods or custom ranges, including the full period since the first food log. The reader can control how much sensitive detail is included before exporting or saving the result.
- Summary-only mode
- Optional written notes
- Optional food names
- Optional AI interpretation
- CSV export by evidence type
- Print and PDF-ready layout
- Saved analytical snapshots
NutriTrack InsightsProgress reportSelected range · Privacy controls applied
01
SummaryMeaningful changes and evidence quality
02
Nutrition and weightTrends, adherence, and pace windows
03
Wellness contextIncluded only when selected
Confidence and caveats remain visible in the exported report.
Person AAgainst their own goal
Adherence · trend · coverage
Person BAgainst their own goal
Adherence · trend · coverage
Normalized comparison, not a body-weight scoreboard.
Comparison
Comparable does not have to mean identical.
Raw weight and calorie totals are poor comparison tools across different people. Insights compares users against their own configured goals, logging coverage, adherence patterns, and changes over time.
This keeps the analytical question focused on behavior and progress rather than turning personal health data into a competition.
Guardrails
Uncertainty is part of the interface.
Projections are withheld when the evidence is too thin. Wellness language avoids clinical interpretation. AI prompts explicitly prohibit diagnosis, prescription, shame, invented numbers, and unsupported causation.
Coverage and weigh-in density affect whether a conclusion is labeled high, medium, or low confidence.
Arrival dates remain ranges, and insufficient evidence produces a withheld projection rather than false precision.
Structured relationships and recurring note themes guide inspection. They are not medical conclusions.
Observation cards preserve the numerical basis and the place in the workspace where the underlying pattern can be inspected.