Changelog

How Faaast is finding its shape.

This is a public product journal, not only a release log. It shows how the work moved from social training, to coach listening, to a human-first AI training platform.

Commit trail

The log should show trust boundaries, not just shipping volume.

Representative hashes sit beside the story where they add trust, but the public shape matters more than listing every internal change.

  • Started with people making, joining, and sharing real workouts.
  • Moved through coach listening and coach-owned planning leverage.
  • Became a reviewable AI weekly planning workspace for athletes.

Stance

Human first, AI-native, and you approve changes.

The changelog records whether the product is moving toward that line or drifting away from it.

  • Public artifacts make the direction inspectable before signup.
  • Assistant evaluation and observability make AI claims more concrete.
  • Native and calendar work move the weekly loop closer to daily training.

2026-07-01

Product

Phase 5

Athletes own their data. Coaches own their work.

The public surface now makes the market stance explicit: self-led athletes should use their training data without being owned by it, and coaches should own the coaching relationship while AI supports the work.

Added

Public product artifacts

Roadmap, changelog, community, and manifesto routes now make the product direction inspectable before someone starts a trial.

Human-led AI manifesto

533e56e8

The public story names the boundary: AI drafts and explains, humans approve, coach, and train.

Changed

Faaast and HorizonCoach became one platform story

94aa7fab
f1dd3318

The athlete and coach lanes now point back to each other: Faaast for self-led training, HorizonCoach for coaches who want leverage without surrendering the relationship.

The changelog became phased

The log now shows how the positioning was found over time: social training, coach listening, self-guided planning, reviewable AI, and trust by design.

Improved

Reviewable AI claims

f27c8809
d8143f14
16a1644e

Assistant evaluation, observability, and no-reviewable-action handling make the AI story more concrete before broader launch.

2026-06

Product

Phase 4

AI became reviewable, not autonomous.

The assistant work moved toward proposals, context, validation, diagnostics, and visible approval instead of silent plan rewrites.

Added

Week assistant proposals

9751c599
b216f5ab
85570f66
382a9b08

The assistant gained tools for memory, context, planner change-set proposals, and semantic planner commands.

AI settings and permissions

cd2a981f
828653d8

The product added an AI settings surface so control can become visible instead of hidden in prompts.

Changed

Trust moved into architecture

e4f398d2
928954bb
9361646a
11c2ec9c

Undo, snapshots, realtime presence, storage readiness, and workspace membership made trust more than a tagline.

Training data stayed bounded

e03c15d4
6ace3e60
d8143f14

Privacy and legal routes, storage boundaries, and observable assistant behavior became part of the launch surface.

Improved

Mobile moved closer to daily training

e4ecdc45
ac391411
64b531b3
fa13a790
2a2258bb

Today, native week views, Health Connect, week repair, proposal previews, and streamed assistant reasoning brought the weekly loop closer to where training happens.

2026-05

Product

Phase 3

Faaast became the self-guided athlete product.

Faaast turned into the athlete-side planning surface: useful when someone trains without a coach, when the week changes, or when structure needs to feel less pressuring.

Added

Weekly planning loop

cd7a6966
53d2fd8f
c599d9e6
4d86d433
1193d2b9

Weekly rhythm templates, planner flows, text editing, and autosave gave self-guided athletes a concrete place to work.

Real-world signals

062d8273
d3271927
63f67cfd

Garmin and Google Calendar work made training interact with devices, schedules, and actual life constraints.

Changed

From perfect plan to changed week

66f595af
9fcd5f72
2a79cfdc
079dabe2

Readiness, re-entry, recovery, and work-life planning made the product less about ideal weeks and more about the week someone really has.

2025-12-2026-04

Coach

Phase 2

HorizonCoach tested the coach side.

HorizonCoach began as the coach-facing exploration: how can software save time, preserve context, and support better coaching without replacing the coach?

Added

Coach-facing product surface

6a75ff9
317aa92
9ffbc53
9a71d80

The first HorizonCoach app shipped hero, insight, tools, philosophy, pricing, and application surfaces.

Plan Your Week

a0762d2
2db8120

The coach website added a planning section focused on weekly workflow and time efficiency.

Changed

From AI coaching to coach leverage

9d26ce4
e9082d2
a386e70
4146834
52db706

Copy moved toward human-centric coaching, judgment, trust, time-saving, athlete management, and product shaping.

Demand became measurable

f573c47
7beb5c6
0c4e358
2a805e8

Application tracking, consent, referrals, and coach data made the listening loop more concrete.

2022-10-2025-11

Platform

Phase 1

Social Workouts proved the human starting point.

The work started with social training: people making sessions, joining others, seeing who is coming, sharing maps, and learning that training software is social before it is intelligent.

Added

Sessions, joining, maps, and avatars

31c06f53
170ebc05
717a4920
5825e5d8
9a855d2d
84c6a642
b1124f7e

Social Workouts began with events, joined-by-name context, places, maps, and readable people.

Spaces, clubs, and invitations

94e52b21
bd1afb17
0b9338e6
edbae998
1b1c7c87

The social layer became spaces, club training, join/leave flows, invitations, notifications, and activity feeds.

Changed

Activity data became a trust problem

0084b857
2f3d1a5
b2709bad
1026402
a704150

Garmin, Komoot, GPX/FIT, token masking, request sanitization, privacy policy work, and session-summary storage made privacy concrete.

FAAAST slows into SLOOOW, then returns.

Try the adaptation

Perfection is the enemy of progress.

Pick up from where you are. FAAST helps you adapt before pressure turns into burnout or injury: protect what still matters, reduce what needs reducing, and leave the rest behind. The training journal is there when reflection helps: capture what got in the way, spot the bottlenecks, and keep the context useful with or without AI. Reflect or skip it. Up to you.