# BrainDrive Phase 2 V.1 Sequence

> The ordered plan for building the Phase 2 V.1 cut. Future active phases should get their own sequence docs.

**Status:** Accepted — read through item-by-item and confirmed by DW + DJ on the 2026-06-12 dev call ([[braindrive-repo]] D258); GitHub holds the always-latest version, the forum thread is the historical record
**Current phase:** Phase 2 - Benefit
**Spec:** [`spec.md`](spec.md)

## Purpose

`braindrive/spec.md` defines what BrainDrive is building and what done means. This sequence defines the order we are building the Phase 2 V.1 cut in.

This doc ranks build items, not every implementation task. Pulse owns task-level execution; item-level `build-plan.md` files own coding and delivery detail.

This doc is product-only. Operations and GTM work do not appear here.

## Sequencing Rules

1. Build in row order.
2. Each row's `Done for V.1` is the acceptance gate for that item.
3. Before completing a row, confirm the V.1 harness can express, run, and evaluate it. If it cannot, improve the harness first as part of that row.

---

## Completed

These are Phase 2 V.1 items already true enough to build on.

| Order | Build item | Done for V.1 | AI/build-system readiness | Sources | Why/order |
|---:|---|---|---|---|---|
| C1 | Core desktop/web access baseline | Owners can open BrainDrive through the supported desktop/web path and reach the main product surfaces reliably. Mobile/phone access is not included in this completed baseline. | Usable baseline; remaining access work is Tailscale/phone bridge. | [`foundation/spec.md`](../projects/production/braindrive-repo/foundation/spec.md), [`interface/spec.md`](../projects/production/braindrive-repo/interface/spec.md) | Required before any page or Your Agent flow matters. |
| C2 | BrainDrive Models baseline hosted access | BrainDrive can run V.1 flows through the current hosted model/credits path, including Anthropic through the production credits service. Model choice, routing, and evaluation are separate V.1 items. | Usable billing/access baseline; model evaluation, routing, and default-choice process remain active. | [`braindrive-models/spec.md`](../projects/production/braindrive-repo/foundation/braindrive-models/spec.md) | Gives V.1 flows a hosted model path while the real model-selection factory is built. |

---

## In Progress

These are the active V.1 build items, ranked in proposed order.

| Order | Build item | Done for V.1 | AI/build-system readiness | Sources | Why/order |
|---:|---|---|---|---|---|
| 1 | Spec-anchored V.1 test/build harness operating loop | The harness can turn V.1 specs into Katie-style scenarios, run them, and produce acceptance evidence for each sequence item. Done requires deterministic functionality/code checks plus subjective user-story judgment against the universal and page-surface stories. Diagnose/fix orchestration is explicitly outside this acceptance gate. | This is the first factory capability: before completing any item below, confirm the harness can express and evaluate that item. | [`braindrive-test-harness/spec.md`](../projects/active/foundation/braindrive-test-harness/spec.md) | Make the harness the first move, then use item 2, item 3, and each later item to harden it. |
| 2 | Shared V.1 page/runtime baseline across default surfaces | Finance, Fitness, Career, Relationships, New Project, and Your Agent all pass the base BrainDrive page loop: navigate/open chat; read known context first; interview for goals/current state; update the correct `spec.md`, `plan.md`, task, journal, or child-app artifact; propagate required local/index/rollup changes; and tell the owner where to review the result. | This is what the harness is for after item 1: prove the main page function across the default surfaces before deeper product work scales. If the shared runtime cannot do this, add the needed Tier 2 enabler before separate page or child-app builds. | [`foundation/spec.md`](../projects/production/braindrive-repo/foundation/spec.md), [`zero-to-high-quality-spec-and-plan/spec.md`](../projects/production/braindrive-repo/foundation/zero-to-high-quality-spec-and-plan/spec.md), [`interface/spec.md`](../projects/production/braindrive-repo/interface/spec.md), [`finance/spec.md`](../projects/production/braindrive-repo/finance/spec.md), [`fitness/spec.md`](../projects/production/braindrive-repo/fitness/spec.md), [`career/spec.md`](../projects/production/braindrive-repo/career/spec.md), [`relationships/spec.md`](../projects/production/braindrive-repo/relationships/spec.md), [`new-project/spec.md`](../projects/production/braindrive-repo/new-project/spec.md), [`your-agent/spec.md`](../projects/production/braindrive-repo/your-agent/spec.md) | This is the core product function: every default surface must work at a base level before beta/model decisions, child-app deepenings, or additive Foundation capability can be trusted. |
| 3 | Set beta BrainDrive Model path | BrainDrive has an owner-usable beta model policy selected from current evidence, applied to BrainDrive Models/default selection, verified on a live install, and paired with rollback and re-evaluation triggers. | Starts with harness preflight for the initial model path; this can be a structured/manual beta decision once the baseline page loop passes, not the full repeatable Model Evaluation factory. | [`braindrive-models/spec.md`](../projects/production/braindrive-repo/foundation/braindrive-models/spec.md) | Beta users need a good-enough owned model path after the baseline product loop is proven. Keep this explicitly interim so it does not become permanent by inertia. |
| 4 | Model evaluation/routing factory | BrainDrive has a repeatable process to narrow candidate models, run BrainDrive-specific evaluations, compare quality/cost/tool-use tradeoffs, and choose default, fallback, escalation, or routing policy with evidence. | Starts with harness preflight for model-eval scenarios after the beta model path is live; this makes future BrainDrive Model choices less manual. | [`model-evaluation/spec.md`](../projects/production/braindrive-repo/foundation/model-evaluation/spec.md) | Every V.1 flow inherits the model quality/cost/tool-use floor, but beta should not wait for the full factory if an acceptable model path can be verified now. |
| 5 | Todo List app task follow-through | Todo List supports create, view, update, complete, and remove behavior; preserves source/page context; follows the memory architecture; and is visible at `Settings -> Your Todos`. | Starts with harness preflight for the shipped task loop; needs runtime confirmation and likely repair before broader Your Agent work scales. | [`your-agent/spec.md`](../projects/production/braindrive-repo/your-agent/spec.md), [`braindrive-todo-list/spec.md`](../projects/production/braindrive-repo/your-agent/braindrive-todo-list/spec.md) | Shipped trust comes next. If the harness cannot evaluate the shipped task anchor, improve the harness, then complete the repair/hardening. |
| 6 | Page follow-up journaling loop | BrainDrive supports the post-plan repeat loop: owner-entered `journal.md` entries under page context; pattern reflection as owner-calibrated hypotheses; owner-approved updates to page `spec.md`/`plan.md`; and Your Agent routing to the owning page journal or pending route. | Starts with harness preflight for journal create/read/update/list/summarize, pattern-reflection quality, parent-summary handoff, Your Agent journal routing, pending/ambiguous placement, and memory-architecture compliance. | [`page-journal/spec.md`](../projects/production/braindrive-repo/foundation/page-journal/spec.md), [`fitness-journal/spec.md`](../projects/production/braindrive-repo/fitness/fitness-journal/spec.md), [`relationship-journal/spec.md`](../projects/production/braindrive-repo/relationships/relationship-journal/spec.md), [`finance-journal/spec.md`](../projects/production/braindrive-repo/finance/finance-journal/spec.md), [`career-journal/spec.md`](../projects/production/braindrive-repo/career/career-journal/spec.md), [`new-project-journal/spec.md`](../projects/production/braindrive-repo/new-project/new-project-journal/spec.md), [`general-journal/spec.md`](../projects/production/braindrive-repo/your-agent/general-journal/spec.md) | This is the core "Repeat" behavior after interview -> spec -> plan. It is lighter and more foundational than Budgeting because every page depends on current owner context and follow-up coaching before deeper execution apps scale. |
| 7 | Document processing | Owners can manually upload supported documents/files from the relevant page/app or Your Agent, parse them through the approved BrainDrive Model path, preserve source traceability, route extracted context to the correct memory/artifact, and avoid silent current-truth updates. | Starts with harness preflight for source/evidence preservation and document-backed flows. | [`foundation/spec.md`](../projects/production/braindrive-repo/foundation/spec.md), [`document-processing/spec.md`](../projects/active/foundation/document-processing/spec.md) | Manual intake supports future document-backed deepenings and can feed page/journal context, but Budgeting no longer makes it a V.1 Finance gate. |
| 8 | Tailscale Mobile Access | Owners can reach their desktop BrainDrive from a phone/browser through Tailscale or an equivalent bridge, with in-interface access links, status, and owner-facing expectations. | Starts with harness preflight for access/link behavior if confirmed in V.1. | [`braindrive-tailscale/spec.md`](../projects/active/foundation/braindrive-tailscale/spec.md) | Still behind the Tailscale V.1 gate. Treat as additive access expansion unless DW/DJ confirm it is a current V.1 blocker. |

---

## Source Of Truth

| Need | Go to |
|---|---|
| Product definition / V.1 rollup | [`spec.md`](spec.md) |
| Current V.1 sequence | this doc |
| Active execution tasks | [`pulse/index.md`](../pulse/index.md) |
| Human focus contract | [`operations/agendas.md`](../operations/agendas.md) |
| Prioritization rules | [`operations/planning/prioritization-rules.md`](../operations/planning/prioritization-rules.md) |
| Current prioritization run | [`operations/planning/v1-prioritization-pass-2026-06-03.md`](../operations/planning/v1-prioritization-pass-2026-06-03.md) |
| Surface specs | `projects/production/braindrive-repo/<surface>/spec.md` |
| Child app specs | relevant `projects/active/...` or `projects/production/...` folder |
| Item-level test/build plans | created by the software-factory build workflow when an item enters build (see [`system/workflows/build-workflow.md`](../system/workflows/build-workflow.md)); Foundation substrates with existing build plans keep them |

---

## When This Sequence Changes

Update this doc when:

- a V.1 build item ships;
- the build order changes;
- a V.1 boundary changes in `spec.md`;
- a Tier 1 issue preempts the sequence;
- Phase 2 V.1 is done.

Do not update this doc for ordinary task progress that only belongs in Pulse, agenda, or an item-level build plan.

---

## Changelog

| Date | Change | Details |
|---|---|---|
| 2026-06-12 | Tightened `Done for V.1` acceptance gates. | Rewrote the Done column to be more testable and less mixed with rationale, caveats, or implementation notes. |
| 2026-06-12 | **Status: Proposed → Accepted** ([[braindrive-repo]] D258). | DW read the sequence through on the dev call and DJ confirmed each item; DW pushed the doc to GitHub + posted the forum thread live on the call. To-Do List (item 5) carries an explicit conditional: if it fights back, journaling (item 6) takes the slot. Triggered by `/capture 2026-06-12`. |
| 2026-06-12 | Added shared Page Journal spec to item 6 sources. | Journal specs now use one shared pattern spec plus page-specific delta specs instead of repeating the universal contract in every page journal. |
| 2026-06-12 | Moved Budgeting out of Phase 2 V.1 and promoted page follow-up journaling as the V.1 repeat loop. | Budgeting becomes Finance V.1.1 execution work; V.1 now focuses on the shared page loop plus journal follow-up/pattern-reflection/Your Agent routing before heavier child apps. |
| 2026-06-12 | Removed the separate Active Gates section. | The sequence itself is the ordered gate: each row's `Done for V.1` is the acceptance gate for that item. |
| 2026-06-12 | Noted broader journal spec coverage across page surfaces. | Finance, Career, New Project, and Your Agent General Journal Router now have first-draft specs to support the shared follow-up journal pattern; Fitness Journal and Relationship Journal remain the first concrete page journals to prove the loop. |
| 2026-06-11 | Added the V.1 launch gate ([[braindrive-repo]] D255). | "Start giving BrainDrive to people" requires the top three: harness done-done (item 1) + interview/spec/build working across surfaces (item 2) with continuing build-harness progress + beta model path set (item 3). Auto-key provisioning (`T-689`) is not part of the gate. Triggered by `/sync braindrive-repo` after the 2026-06-11 dev-call capture. |
| 2026-06-11 | Sharpened item 1's done definition with the harness done-done boundary ([[braindrive-test-harness]] D9). | Done-done = deterministic functionality checks + subjective user-story judgment against the Universal + page-surface user stories; the diagnose→fix/execution loop is a separate bucket (bonus, not required). |
| 2026-06-11 | Budgeting keep/cut gate updated to the 2026-06-05 resolution. | Kept; residual check moves to the item-3 beta model path. `T-678` (GPT-5.5 escalation) closed superseded — model selection owned by Model Evaluation. |
| 2026-06-11 | Added lightweight journaling apps to the V.1 sequence. | Fitness Journal and Relationship Journal replace internet search's V.1 slot because owner-provided journaling is a lighter, higher-value way to deepen page memory without expanding outside BrainDrive. |
| 2026-06-11 | Removed staging/release validation from the V.1 sequence. | DW confirmed item 7 is not needed for V.1; downstream sequence items were renumbered. |
| 2026-06-11 | Moved internet search to V.1.1 unless it blocks the V.1 baseline. | Search is outside-BrainDrive current-fact lookup. Keep it out of V.1 unless a default-surface baseline flow cannot pass without current external facts. |
| 2026-06-11 | Moved shared page/runtime baseline directly after the harness. | The harness exists to prove the main function of every default page surface: known-context-first interview, right-artifact updates, owner review, and base runtime usability. Interface V.1, Finance V.1, and Your Agent routing/updating are now included in this baseline instead of separate later rows. |
| 2026-06-11 | Split beta model path from Model Evaluation factory. | After the initial harness checks pass, the sequence now sets an interim beta BrainDrive Model path first so beta users can start, then builds the repeatable Model Evaluation factory to make future default-model choices less manual. |
| 2026-06-10 | Split Model Evaluation out from BrainDrive Models. | Model Evaluation now points to the standalone Model Evaluation spec/roadmap; BrainDrive Models remains the operational credits/LiteLLM deployment path for the approved policy. |
| 2026-06-09 | Removed links to deleted planning-generated build plans. | Page/surface `build-plan.md` files were removed per D245/D246/D248 — test/build plans are now created by the software-factory build workflow when an item enters build. Spec links unchanged; Foundation substrate build plans (BrainDrive Models, document processing) still exist and stay linked. |
| 2026-06-03 | Promoted the V.1 harness operating loop to item 1. | Each sequence item now starts with harness preflight: confirm the harness can express/run/evaluate the item, improve the harness if not, then complete the item through that validation path. |
| 2026-06-03 | Promoted BrainDrive Models evaluation/routing factory to top product substrate. | Added the model factory as the first shared product blocker after harness preflight because every flow inherits the model quality/cost/tool-use floor; later superseded by the 2026-06-11 ordering that proves the shared page/runtime baseline before model selection. |
| 2026-06-03 | Added minimum-factory-before-product sequencing rule. | Pulled the shared V.1 page runtime pattern into the active sequence ahead of separate page builds and framed it as the factory piece for repeatable page slices. |
| 2026-06-03 | Reordered active V.1 items around shipped/live trust. | Moved Your Agent task follow-through, Interface V.1, Finance V.1, and Your Agent routing above additive Foundation expansion work. Search and Tailscale no longer outrank live-app hardening unless confirmed as blockers; search was later moved to V.1.1 by the 2026-06-11 sequencing decision. |
| 2026-06-03 | Drafted Phase 2 V.1 sequence. | Replaces the overloaded master `build-plan.md` concept with a product-only build sequence: completed / in-progress / not-started V.1 items, AI/build-system readiness, and source links. |