Building the AI Coach Workspace: What We Shipped Over the Last Few Weeks

The Theme: Less Dashboard, More Assistant

The last few weeks of Matter work have been about one product bet: coaches should not have to hunt through a dashboard to figure out what matters today.

The software should do more of the first pass. It should read the roster overnight, understand each athlete's context, draft the next useful action, and let the coach approve, revise, or ignore it. The coach stays in charge. The AI does the scanning, synthesis, and repetitive setup work.

That idea shaped almost everything we shipped recently.

The Coach Dashboard Became a Daily Work Surface

We tightened the coach dashboard around the way a coach actually starts their day:

• Today-first navigation for the work that needs attention now
• Roster signals that separate urgent issues from normal training noise
• Inbox and alert surfaces that make pending decisions easier to triage
• Cleaner workspace chrome so the app feels like an operating surface, not a marketing page

The goal is not to make the dashboard louder. It is to make it easier to trust. Coaches should be able to open Matter before practice and know where to look first.

The AI Assistant Got Closer to the Work

The biggest product change is that AI is now threaded through the coach workflow instead of living as a separate chat box.

You can still ask questions directly, but the more important shift is contextual AI:

• Ask about an athlete from the athlete page
• Ask about a workout while reviewing the session
• Ask for a program revision while editing the plan
• Ask about a trend without manually pasting the data
• Open the assistant from the global coach shell with the page context already attached

That matters because generic AI advice is cheap. Contextual coaching intelligence is the hard part. Matter is moving toward AI that understands the current team, current athlete, current training block, and current decision.

Program Generation Became a Real Builder, Not a Prompt Box

We spent a lot of time on the generate-program flow because coaches do not just need a model to write workouts. They need a workflow for getting from intent to something they would actually assign.

Recent work improved:

• Editable server-generated drafts instead of throwaway text output
• Refinement loops so a coach can critique the first version and keep structure intact
• Sport-aware defaults for endurance, strength, team, and class contexts
• Better preservation of workout details while revising
• Review surfaces that make the generated plan easier to inspect before assigning

The product principle is draft-and-approve. Matter can move fast, but the coach should always have the final word.

Calendar, Classes, and Team Management Got More Practical

The less glamorous work matters just as much. A coach workspace only works if the everyday operational surfaces are solid.

We improved the calendar, class sessions, staff/team management, and adjacent workflows so the system can support real teams rather than only demo paths. That includes better focus controls, cleaner class and session flows, stronger route coverage, and fewer dead-end states.

This is the difference between a cool AI demo and a product a coach can open every day.

Athlete Experience Is Part of the AI System

AI is only as useful as the context it can understand. That means the athlete side matters.

Recent athlete work focused on making workouts, classes, wellness, completion, and session details easier to consume and log. The athlete experience should feel simple: know what is assigned, complete it, give useful feedback, and move on.

Every clean athlete interaction improves the coach's morning briefing later. Better input creates better AI output.

Design Direction: Matter, With AI at the Core

We also clarified the brand system. The product name is Matter. The AI is still central to the product, but it should not make the brand feel noisy or generic.

That means the interface is moving toward:

• Simple wordmark: Matter
• One clear signal color instead of competing AI gradients
• Dense, calm coach surfaces built for repeated use
• Marketing that still explains the AI advantage directly
• Less decorative UI where the user is trying to make decisions

The point is to make Matter feel like serious coaching software with AI at the core, not a chatbot product wearing a sports skin.

What This Unlocks Next

The work from the last few weeks sets up the next phase: making the assistant more agentic without making it reckless.

The near-term direction is clear:

1. AI should surface the right issue before the coach asks.
2. AI should draft the safest useful action.
3. The coach should approve, edit, or reject with minimal friction.
4. Matter should learn from that decision and get better next time.

That loop is the product. Not a dashboard. Not a generic chatbot. A coaching workspace where AI does the overnight analysis and the coach makes the call.

That is what we have been building.