Fitnnovation is a native iPhone app for personal trainers. It holds a persistent memory file for every client, reads their Apple Health data in real time, and surfaces the patterns worth your attention, so you walk into every session knowing exactly where you left off.
Trainer home screen
When you open a client profile, the app already knows who they are. A persistent memory store called the client file holds structured facts about each person: movement history, injuries, test results, programming responses, anything the AI has learned or you have written. Facts are append-only with a timestamped audit trail, so you can see how a squat max or body composition has changed over time, not just where it stands today.
The client file anchors every AI interaction in the app. When the Intelligence Bar answers a question, it reads from this file. When auto-progression evaluates a weight increment, it reads from this file. The memory persists across sessions, across weeks, across blocks.
You walk into every session with full context. You never start by re-reading a stale intake note.
The intake session captures four structured assessments: overhead squat, mobility, body composition, and fitness tests. Each has its own editor and saves directly to the client's record.
The overhead squat assessment is wired to program generation. When you complete the screen, the AI uses the findings to draft a mobility program, which you review and apply. Assessment results seed the client file, so the AI has structured facts to reason from, not free text to interpret. Re-test reminders auto-generate as tasks, so you are prompted to reassess at the right intervals without tracking it yourself.
The first session produces a structured client file, not just a note. Every program and every suggestion after that draws on the same foundation.
Assessments workspace
You can build a program from a Notion-imported template, from scratch, or by asking the AI to draft one based on the client file. However it starts, the program is yours to customize: exercise selection, sets, reps, target weights, session structure, focus, duration.
Notion templates import directly from your workspace. The importer reads your program pages and writes them into the template library. From there you apply a template to a specific client, which materializes it as their individual workout plan.
Each program has a volume dashboard that shows training load across the block: week-by-week volume distribution, set counts by muscle group, and session density. The numbers that used to live in a spreadsheet.
Your existing Notion library becomes the foundation. Add a client, pick a template, customize, and the program is live.
Program and volume
When you enable auto-progression on a plan, the app evaluates each session after the client logs it. It reads the set log, the client's Apple Health recovery data, and the relevant facts in the client file, then decides whether to increment, hold, or flag for your review.
Every decision is written to an audit log with the previous value, the new value, the full input snapshot, and the AI's reasoning. A summary task lands in your list, for example "3 adjustments applied to Brian's hypertrophy program: squat up, bench held, deadlift up. Tap to review." Each row in the Recent auto-progressions section has a one-tap revert.
Two things are never automatic, even when auto-progression is on: deload recommendations and low-recovery holds. Both always surface as tasks for you. The AI does not bump weight on a day the client's HRV is down and their sleep was poor. You can disable auto-progression on any plan at any time without breaking it, and existing audit rows stay queryable.
The increments that took twenty minutes per client per week happen on their own, with a full audit trail and your hand on the override at every point.
Every client's profile has a recovery and activity section that pulls from their Apple Health data. It shows a composite recovery score, HRV, resting heart rate, sleep duration, daily steps, active kilocalories, and workout count, each with a 7-day sparkline and a delta chip versus their personal baseline.
The data syncs in the background via a scheduled task, so by the time you open a profile the numbers are current. Color coding follows a consistent zone system: green for healthy range, yellow for caution, red for significantly outside baseline. When auto-progression is evaluating an increment, this snapshot is one of the three inputs alongside the set log and the client file. If the client is under-recovered, the AI holds. Whoop is also wired for clients who wear one.
You see the full picture of how a client is recovering between sessions. You know before they walk in whether today calls for a full session or a modified one.
During onboarding, a 5-step interview captures your training style, specialties, certifications, coaching philosophy, programming defaults, and communication tone. This profile is injected into the system prompt on every AI interaction: the Intelligence Bar, Research Chat, session note distillation, follow-up drafts.
The result is that AI responses sound like you, not like a generic assistant. When the bar drafts a follow-up for a client, it draws on the communication tone you described. When it evaluates a progression, it draws on the programming philosophy you articulated. The interview is in Settings, and any update takes effect on the next interaction.
The Intelligence Bar is a persistent command input at the bottom of every trainer screen. It accepts natural-language queries, for example "where is Brian in his block," and structured commands. It reads the trainer profile, the current screen, and the date. It can look up clients, surface facts from the client file, create tasks, and navigate to any profile.
The AI behaves like a colleague who has read your intake notes and knows your philosophy, not a chatbot you brief from scratch every time.
Brian is in week 4 of a 6-week strength block. Squat is up two increments since the start, bench has held twice. His recovery has been clear all week. Week 5 is the planned intensity step, so this is the session to confirm the next jump on squat.
Messaging is bidirectional, trainer to client and client to trainer, with 15-second polling so conversations stay current. Voice dictation is supported: tap the microphone and speak rather than type.
The home screen shows the five most recent threads. A Signal Blue dot marks threads where the last message is from the client, so you see at a glance who is waiting. The full inbox is one tap away.
Follow-up automation runs on configurable rules: prospect at 7 days, intro scheduled at 5 days, intro completed at 3 days. When a rule triggers, a local push notification fires and a task is created. You review and send. Nothing goes to a client without your review. Group messaging is supported, and email sequences run on the same pipeline.
You do not miss a follow-up, and you do not have messages spread across texts, email, and a separate CRM. Everything sits in the client record where it belongs.
Messages
Research Chat is a trainer-facing assistant whose domain is health and human performance: nutrition, programming, supplementation, recovery protocols, dietary patterns, and the literature behind them. It is reachable from the home screen quick-access card and from the Intelligence Bar. Responses are calibrated for someone on the gym floor between sessions: direct, cited where it matters, without the hedging of a generic search result.
Workout Chat is a dual-mode surface. When a client is logging a session, it extracts structured set data from conversational input and structures it into the workout log. You can also use it to review and annotate session data. And session notes have a distillation tool: after a session, the AI reads the raw notes and produces a structured summary for the client file, which you review and apply.
You can answer a client's question about protein timing or periodization in the time it takes to walk to the water fountain. The research sits in the same app as the client file.
Research chat
Clients log in to their own account, which you create and invite via a secure invite code. Their home screen shows their current workout, a log view for entering sets in real time, and the message thread with you.
When a client logs a session, the data flows back into the trainer view immediately. Set logs become the input for progression analysis. HealthKit data syncs in the background via a scheduled task, so recovery metrics are current without the client doing anything.
The client does the logging. You see the data. The coordination that used to happen over chat happens inside a structured record.
Client side
The schedule view reads from iCloud Calendar and matches events to clients in the app by name. The pre-session brief on the home screen surfaces the next matched event when it is within roughly two hours, with a one-tap link to that client's profile. The next-session widget shows the soonest upcoming session as a persistent entry point.
Adding a session to the schedule writes to iCloud by default, and sync status is visible in the app.
You open the app and the first thing you see is who is next and a direct link to their file. No manual navigation.
Schedule
Every AI output surfaces as a task, a draft, or a suggestion with a one-tap apply, modify, or dismiss. The app handles the math and the memory. You handle the human. These are the lines it does not cross.
Follow-ups and replies are drafted for you. The draft waits in a task. You read it, edit it if you want, and you hit send.
A weight increment, an exercise swap, a deload: these are suggestions surfaced as tasks. The one exception is auto-progression, which you enable per plan, which logs every change with full reasoning, and which you can revert in one tap.
Conversions, completed assessments, the end of a block: those are coaching decisions. The app surfaces the moment and the data. The call is yours.
When the recovery signals say a client is under-recovered, the AI holds or flags a deload as a task. It does not progress through a day the body is asking to rest.
Fitnnovation is a native iPhone app. It runs on iOS 17 and later. It connects to Apple Health, your Notion workspace, and iCloud Calendar. All client data lives in a private Supabase database: your data, not a shared platform. The app is in active development, and new trainers are joining by invite during the current launch period.