For iOS and Android

Honest AI that helps you see yourself clearly.

Darshn is a food, hydration, weight, and training logger built around estimates you can audit. Use it without an account, correct what the AI got wrong, and save your progress later when you are ready.

Request beta access Audit an estimate yourself ↓

Guest-first logging AI estimates shown as ranges Food, water, weight, and training

The loop

No fake precision. No mystery numbers.

Darshn starts with common whole foods, packaged-food reference values, and AI when it needs judgment. Every uncertain estimate stays inspectable.

Scan

Search common whole foods, scan a barcode, take a food photo, speak a log, or scan a menu for the best choice right now.

Explain

Whole foods and packaged items use database or label reference values. Portions and preparation still matter. AI food estimates show a range and a confidence label.

Correct

Tap smaller, bigger, cooked in oil or sauce, did not finish, or edit grams. Review the updated estimate before logging it.

Save

Keep food, water, sets, and weight together in your daily log. AI analysis and plan generation need an internet connection.

Adapt

Calorie targets adjust from your weight trend after a weekly check-in, with roughly four weigh-ins a week for a useful trend.

Built carefully

Pre-launch, built around auditability.

Darshn is for people who want nutrition and training guidance that admits uncertainty instead of pretending to be exact.

StatusBuilt for iOS and Android. Request early access below.
DailyCalorie ring, macros, hydration, daily summary, and kg/lb plus cm/in units.
TrainingSmart Intake turns one goal sentence into setup; AI plans, Quick Workout, live sessions, history, and coach notes follow.

Demo 01 — Food scan

Audit the estimate. Right now.

This is the review surface Darshn puts around a photo estimate: a calorie range, confidence label, visible uncertainty, and one-tap corrections that update the log.

Preset scenarios are scripted examples. Ask sends your question about the selected example to live Gemma 4. If unavailable, an example answer is labelled.

Evidence overlay · demo photo

Explain this range

AI estimate, not fake precision.

≈450kcal (380–520)

≈35gprotein

≈46gcarbs

≈13gfat

Lowconfidence

Detected

Grilled chicken
White rice
Broccoli

Possible

Teriyaki sauce
Sesame oil

Unknown

Exact portion size

Reasoning summary

Chicken, rice, and broccoli are visible, but bowl depth and oil or sauce are uncertain. Darshn keeps the meal as a range until you correct the portion.

Visual cluesChicken, rice, broccoli, sauce sheen.

Macro assumptionsGrilled chicken serving plus rice base.

Uncertainty factorsSauce/oil and exact bowl depth.

Confidence levelLow confidence until portions are confirmed.

Demo only, not medical advice. Database and label values are references; photo estimates remain estimates after correction.

Edited chicken to 170g.

Updated
≈520 kcal (470-590) · Protein 52g · Confidence Medium

Demo 02 — Smart Intake

Say the goal. Darshn fills the setup.

Describe your goal in a sentence, review the extracted setup, and build a training plan. Generation time varies with the plan and connection.

Preset goals are scripted examples. Ask uses live Gemma 4 to explain the selected setup. Website answers use cloud AI; on-device features vary by phone.

Goal sentence

Build muscle, 4 days a week, dumbbells at home, beginner.

Smart Intake pulls out the setup instead of making you fill every field by hand:

Goal: build muscle Schedule: 4 days/week Equipment: dumbbells at home Level: beginner

Generated setup

4-day beginner hypertrophy plan using dumbbells at home. First session: full-body strength with locked future sessions available to preview.

AI-generatedfirst session

8 weeksplan length

Auto-progressedprogression

Live sessionSticky Complete Set / Rest bar, last-time numbers, weight steps or typed weight, exercise swaps, pause and resume.

Honest finishFinish early counts only the sets you did, then shows duration, sets, volume, and coach notes.

Next timeWorkout history auto-progresses targets for the next session without hiding what changed.

Demo only, not medical advice. Quick Workout creates a single session; full plans remain editable as you train.

Under the hood

What runs where, plainly.

Some intelligence runs on the phone. Some uses cloud AI. Darshn says which is which because privacy claims should be inspectable too.

Food photosSingle or multi-angle photo analysis uses cloud AI and returns a range with confidence.

Plan generationSmart Intake and full multi-week AI plans use cloud AI.

On-device nudgesSelected coaching features can run locally on supported Android phones after a model download.

Fallback phonesSome features use cloud fallbacks; others may be unavailable when the local model cannot run.

Health integrationApple HealthKit on iOS and Health Connect on Android provide supported health signals only with your permission.

Daily loggingKeep food, hydration, weight, and completed workout sets in one place.

No hand-waving

Not everything stays on your phone. Food photo analysis and plan generation use cloud AI. On supported Android phones, private coaching nudges run locally; availability and fallback behavior vary by feature and device. Cloud AI features need an internet connection.

Private beta

Explore Darshn for iOS and Android.

Darshn supports iOS and Android. Request an invitation by email; availability and pricing will be confirmed before launch.