Skip to main content

Chatbot - Sprint 02 Goals (2026-Q3)

Delivery / Program Management - team-level, cross-initiative. A sprint is a derived render of the roadmap Priority table x per-discipline implementors: x each initiative's rfcs/*.task-breakdown.md readiness x capacity.md, plus the sprint envelope above. It is not a source of truth for priority, effort, or staffing - those stay canonical in the roadmap / README / capacity. See CLAUDE.md "Sprint layer".

Sprint window: 2026-07-27 -> 2026-08-10 (2 weeks)

Sprint goal: convert Sprint 1's 28-item in-flight tail into shipped, code-evidenced work; give P2 AI Agent Impact Report its first implementation queue by starting the unblocked mock-first FE phase; and close the two gaps that stop the board telling the truth — P5's missing task breakdown and the unmodeled 26Q3 Fast-track lane.

Source (derived from): the ## Priority table and ## Adhoc / Interrupt lane in ../roadmap.md, the staffing mirror in ../capacity.md, recorded initiative implementors: / qa:, the available task breakdowns, and — for carry-forward — sprint 01's findings ledger and pass 16.

Carry-forward caveat. This sprint opens on the day sprint 01 closes (2026-07-27), matching sprint 01's Monday-to-Monday boundary convention. Sprint 01 closed at pass 16 with 12 findings open or escalated. No item is carried by default: every row below was re-derived from current roadmap rank and its breakdown's remaining work.

Completion updates — 2026-07-27

CRM Deals Custom Fields (P10 / BOT-4662) reached General Availability effective 2026-07-27 — see status 2026-07-27.

Sprint 2 outcome targets

  1. The in-flight tail lands. 28 items entered this window mid-flight — 6 Ready to Merged, 5 In Deployment, 9 In Testing, 4 In Progress, 2 In Review. Landing them, with merged-code evidence, is the first call on every queue and outranks starting new slices.
  2. P2 AI Agent Impact Report gets implementation tickets and starts. The rank-2 initiative has carried an escalated coverage-gap (F-2/F-42) since sprint 1: 1 board issue, 0 BE/FE, against a complete 33.5d / 19-task breakdown. Phase 1 (FE, mocked, 13d) has no external gate — file it and start it.
  3. P3's progress number becomes true. A 7% floor is now recorded (status 2026-07-27); resolve the F-14 descope and the F-52 breakdown join that cap it.
  4. P5's planning gap closes. Unified Agent Quality Scorecard Phase 1 still has no task breakdown and no implementors: — the one sprint-1 outcome target that did not move. (P4's breakdown did land, closing the other half.)
  5. The 26Q3 Fast-track lane gets modelled or explicitly rejected. Epic BOT-4554 holds live production-bug work and maps to no initiative and no bucket in this repo. It is consuming real capacity invisibly.

Initiative index

All paths are relative to this file. Priority ranks come from the roadmap ## Priority table.

PriInitiative (Epic)Sprint 2 roleReady basis / effort contextBuild gate
P1Autonomous AI Agent - Phase 2: AI-Assisted Refinement (BOT-4602)Finish + close — all 6 BE/FE issues Done; QA is the only remainderbreakdown: 24.5d (FE 12, BE 8, QA 4.5); README progress_pct: 51None remaining for BE/FE; QA evidence outstanding
P2AI Agent Impact Report - Phase 1: Live Impact Report (BOT-4568)Primary new-build track — file implementation tickets, start mock-first FEbreakdown: 33.5d (FE 15.5, BE 10, QA 8); Phase 1 FE 13d ungatedData/BI datamart + OQ-1/OQ-2/REV-2 gate Phase 2 only
P3AI Agent Testing - Phase 1: Historical Validation (BOT-3351)Land the tail + make progress truebreakdown: 37.5d (FE 13, BE 16.5, QA 8); 7 items in flightAI platform batch capacity; F-14 descope decision
P4AI Agent Knowledge - Phase 1: Conversation History (BOT-3992)Start mocked FE + unblocked BE contract slicesbreakdown now exists (new since sprint 1): 30.5d (FE 7, BE 16, QA 7.5) — conflicts with README 13d, see blockersData & AI masking/training/webhook contracts gate Tasks 2.4–2.9
P5Unified Agent Quality Scorecard - Phase 1: Settings & Rubric (BOT-4573)Planning gap — close itstill no task breakdown; no implementors: recordedFE design frames; DSAI rubric wording (non-blocking for Phase 1 per its PRD)
P6Unified Agent Quality Scorecard - Phase 2: Auto-Scoring & In-Room (BOT-4573)Visible backlog onlynot projected into Sprint 02Not pulled unless PM/TPM explicitly reprioritizes
P7Unified Agent Quality Scorecard - Phase 3: Analytics Report (BOT-4573)Visible backlog onlynot projected into Sprint 02Not pulled unless PM/TPM explicitly reprioritizes
P8Mekari Action (AI Agent) (BOT-4571)Visible backlog onlynot projected into Sprint 02Not pulled unless PM/TPM explicitly reprioritizes
P9Native Integration — Google Calendar Phase 1 (BOT-4482; umbrella BOT-4600)Land + evidence the sprint-1 Goal-1 sliceeffort still TBD; 3 items in flight (1 FE merge-ready, 2 QA in testing)Google app verification is external — never represent as approved
P10Qontak Action — CRM Deals Custom Fields (BOT-4662; umbrella BOT-4597)Landed / GA — close evidence reconciliation onlyGA effective 2026-07-27 — see status 2026-07-27F-26: BOT-4668 descoped; BOT-4669/4673 remain board-Done without code trace
P11AI Agent Idle Action (BOT-4567)Visible backlog onlynot projected into Sprint 02Not pulled unless PM/TPM explicitly reprioritizes
P12AI Agent Monitoring (BOT-4569)Visible backlog onlynot projected into Sprint 02Not pulled unless PM/TPM explicitly reprioritizes
P13Bot-vs-Human Traffic Split (A/B) (BOT-4570)Visible backlog onlynot projected into Sprint 02Not pulled unless PM/TPM explicitly reprioritizes
P14Multi-WABA Impact (BOT-4572)Visible backlog onlynot projected into Sprint 02Not pulled unless PM/TPM explicitly reprioritizes
Chatbot 26Q3 Engineering Initiatives (bucket, Epic BOT-4620)Per-story pull — A1 in flight, A4 closing, A6 newly pulledtracker; classed not rankedPer-story; A1 Expedite authorization still TBD
Chatbot Q3 2026 Code Readiness (bucket, Epic BOT-4500)Per-story pull — A5 in progresstracker; classed not rankedPer-story; open-contributor

Per-engineer queues

Each person works their list top-to-bottom. A genuinely blocked item is skipped and revisited once the blocker clears. Queue order mirrors roadmap priority except where listed in ## Deviations from roadmap priority.

Sprint-wide rule this window: an item you already have in flight outranks a new slice. Land it, merge it, and make sure the Jira key appears in a commit on the default branch — pass 16 found 12 of 30 done issues unmerged and 9 with no trace at all, and unevidenced "done" buys the team nothing.

Farras Herdicho (BE)

Agent brief: start at row 1 and get all four P3 branches merged to origin/master before touching anything else — Linggar's two In-Deployment FE items and Tito's QA sit behind them. Skip row 3 if the board has already been corrected. You unblock P3's entire close-out.

#Epic (Jira)TaskStatus
1AI Agent Testing (BOT-3351, P3)Land the 4 in-flight BE items: BOT-4619 (Task 2.3 gap — rating PATCH + transactional recompute), BOT-4723 (stale-question daily cron), BOT-4728 / BOT-4729 (list/detail adjust)Ready — all 4 Ready to Merged; BOT-4619 is the oldest WIP on the board (~13d)
2Code Readiness (BOT-4500, bucket story A5 🟡)CR-S05 Internal Tooling Improvement (BOT-4742) — in-progress per trackerReady · Pull rule verdict PULL-TO-CLOSE — 🟡 Standard, single-discipline BE, closeable in-window
3Code Readiness (BOT-4500, P—)Board correction for F-27: BOT-4655, BOT-4656, BOT-4658 are done/Won't Fix with zero code trace on any ref — merge the work or correct the boardReady — a board fix, never a repo edit
4AI Agent Testing (BOT-3351, P3)Next unstarted BE tasks from the breakdown once rows 1–3 clearReady from breakdown; execution gated on AI platform batch capacity

Samudra Linggar (FE)

Agent brief: clear the remaining in-flight FE items in row 1 first. Row 1 is verification and release, not new code; the CRM Custom Fields slice (BOT-4725) is already done in FE 3.8.4. Then close engineering bucket story A4. Only then start P4's mocked FE — it is genuinely unblocked because Phase 1 mocks the Data & AI contracts.

#Epic (Jira)TaskStatus
1aAI Agent Testing (BOT-3351, P3)Land BOT-4697 + BOT-4732 (Testing FE port + modal/flag gate)ReadyIn Deployment; verification + release
1bCode Readiness (BOT-4500, bucket)Land BOT-4657ReadyIn Deployment
1cEngineering Initiatives (BOT-4620, bucket)Land BOT-4736 (reasoning value not saved)ReadyIn Deployment
1dQontak Action — CRM Deals Custom Fields (BOT-4662, P10)BOT-4725 (QACF TypeError) — completed in FE 3.8.4Done — passed QA and released
2Engineering Initiatives (BOT-4620, bucket story A4 🟡)Subscription feature-flag gating reliability — BOT-4696, in-reviewReady · Pull rule verdict PULL-TO-CLOSE — 🟡 Standard, FE-only, already in review. Note: A4 is a placeholder story with no RFC, reconstructed from BOT-4650 code changes
3AI Agent Knowledge (BOT-3992, P4)Start Phase 1 mocked FE from the new breakdown: Task 1.1 (entry + service contract), 1.2 (configuration form), 1.3 (training-status states + source detail) — FE 6dReady — Phase 1 mocks the Data & AI contracts, so it is not gated
4Autonomous AI Agent Phase 2 (BOT-4602, P1)FE complete — support Tito's QA close-out onlyNo remaining FE build

Baghiz Firdausa (FE)

Agent brief: clear the remaining merge-ready items in row 1 first; the CRM Custom Fields slice is already closed (BOT-4724 done, BOT-4727 re-parented/done). Row 2 is the sprint's headline new build and the tickets do not exist yet: do not start until sprint-jira-sync has filed them, or the work will be invisible to the board exactly as it was in sprint 1. Tasks 1.1–1.8 are mock-first; never fabricate a datamart field.

#Epic (Jira)TaskStatus
1aAI Agent Knowledge (BOT-3992, P4)Land BOT-4734 (save — page unresponsive). Epic cited is where the docs say it belongs; on the board it is unparented — parenting is a PM/TPM fix, see Deviations + blocker 6ReadyReady to Merged
1bNative Integration — Google Calendar (BOT-4482, P9)Land BOT-4735 (Google Calendar icon)ReadyReady to Merged
1cQontak Action — CRM Deals Custom Fields (BOT-4662, P10)BOT-4724 (QACF field picker) — completed; BOT-4727 completed under Get Ticket & Get Deals Epic BOT-4790Done — QACF slice closed
2AI Agent Impact Report (BOT-4568, P2)Sprint 2 headline: Phase 1 mock-first FE, breakdown order — Task 1.1 (Pinia store/service/fixtures), 1.2 (feature+role gate middleware), 1.3 (report shell + baseline-forming + error/retry), 1.4 (value-delivered tiles), then 1.5–1.8 (quality tiles, BlendedJourneyBar, ForecastPanel, CostAssumptionModal). FE 13dBlocked on ticket creation, not on engineering — F-2/F-42: zero BE/FE issues exist under BOT-4568. Run sprint-jira-sync first. Use placeholder constants for OQ-1/OQ-2
3AI Agent Testing (BOT-3351, P3)FE work from the breakdown (13d) behind Linggar's port landingReady from breakdown
4Code Readiness (BOT-4500, bucket A2/A4 🟡)CR-S02 testing improvement, CR-S04 agent skills (with Linggar)Pull rule verdict BE-advance → defer — 🟡 Standard, both backlog, not closeable alongside rows 1–3

Eko Aprianto (BE)

Agent brief: row 1 is live production work you already hold — finish it, but see Deviations: it sits under an Epic this repo does not model, so log what you spend. Row 2 is a board-truth fix, not a feature. Row 3 is your real build lane, and only Tasks 2.1–2.3 are startable. Row 5 is a new bucket story pulled 2026-07-27, adjacent to your row-1 ticket.

#Epic (Jira)TaskStatus
126Q3 Fast track (BOT-4554)BOT-4746 — AI Resource Upload stuck in Processing statusIn progress — unmodeled lane; see ## Deviations and blocker 6
2Qontak Action — CRM Deals Custom Fields (BOT-4662, P10)F-26 evidence reconciliation: BOT-4668 is Won't Fix/descoped; BOT-4669 and BOT-4673 remain board-done without merged-code evidence — correct the board truthDone for rollout — corroborated in status 2026-07-27; board-truth follow-up remains
3AI Agent Knowledge (BOT-3992, P4)Start the ungated BE contract slices from the new breakdown: Task 2.1 (division scope on sources), 2.2 (divisions+agents directory endpoint), 2.3 (extend create contract)Ready — Tasks 2.4–2.9 (delete guard, sync worker, webhook consumer, masking/quality gate, room-link contract) stay blocked on Data & AI
4AI Agent Testing (BOT-3351, P3)BE support; note capacity records him as Jira-assigned here (BOT-3942) without a sized allocationReady; sequencing behind Farras
5Engineering Initiatives (BOT-4620, bucket story A6 🟡)Alert detecting an AI knowledge upload stuck in Processing >5h — BOT-4796, To Docontext noteReady · Pull rule verdict PULL-TO-CLOSE — 🟡 Standard, single-discipline BE, closeable in-window

Agus Suparman (BE)

Agent brief: your two Ruby/Rails items are the oldest in-progress work on the board (~13 days) — land them first. Row 2 is P2's backend, and only Task 2.1 is genuinely startable; do not invent datamart fields to unblock 2.2/2.4/2.6.

#Epic (Jira)TaskStatus
1Engineering Initiatives (BOT-4620, bucket story A1 🔴 Expedite)Land Ruby 3.4 / Rails 7.2 Phase 1 tail: BOT-4635 (Task 13 — aegis boot spike, In Review) and BOT-4636 (Task 14 — staging bake + dual-cloud canary, In Progress) — per Phase 1 RFC + breakdownReady · PULL-TO-CLOSE — 🔴 Expedite, WIP limit 1 respected; oldest WIP on the board (~13d). Tracker and Expedite-authorization drift: see Deviations
2AI Agent Impact Report (BOT-4568, P2)BE Phase 2 from the breakdown — start Task 2.1 only (migrations for ai_activity_logs + ai_cost_assumptions + rooms indexes, incl. the REV-4 scan-cost risk)Partially ready — Task 2.2 (aggregator correctness core, REV-1/REV-3), 2.4 (entitlement, OQ-1/OQ-2), 2.6 (forecast, REV-2/OQ-10) all blocked on Data/BI + open RFC questions
3Native Integration — Google Calendar (BOT-4482, P9)Assemble the Google OAuth app-verification technical inputs: scope justification, demo environment, consent/branding evidenceReady for prep — submission and approval are external and owned by Google
4Engineering Initiatives (BOT-4620, bucket story A2 🟡)Phase 2 Rails 8.1 discoveryPull rule verdict defer — 🟡 Standard, backlog, not closeable while A1 is in flight

Tito Amanda (QA)

Agent brief: you are the sprint's bottleneck — 9 items sit In Testing across 6 Epics. Work rows 1a–1c in that order (P9 Goal-1 carry-over, then P1 close-out, then BAU), and raise it early if the queue cannot clear: several initiatives cannot be called done without your evidence.

#Epic (Jira)TaskStatus
1aNative Integration — Google Calendar (BOT-4482, P9)BOT-4642 — test the Google OAuth app-registration flow and record the verification evidence; Dimas Fauzi Hidayat helps with the actual registration/submission after the evidence is readyTo Do — Tito owns QA testing and recording; Google's verification decision is not a sprint test result
1bAutonomous AI Agent Phase 2 (BOT-4602, P1)BOT-4644 — Refiner test-case creation. This is the only thing between P1 and doneIn testing
1cAI Agent Testing (BOT-3351, P3)BOT-4646 — Historical Validation test-case creationIn testing
2AI Agent Impact Report (BOT-4568, P2)BOT-4645 — test-case creation; then QA the mocked FE flow as Baghiz lands Tasks 1.1–1.8In testing — data correctness waits on the BE/Data contract
3QA BAU (BOT-4609; ../qa-bau.md)Land the 4 in-flight BAU items: BOT-4638 (test-case mgmt migration Pt.2), BOT-4639 (delivery maintenance automation), BOT-4647 (Sprint 1 deployment verification), BOT-4648 (P0 coverage web+API)In testing — scheduled deployment verification takes precedence
4Qontak Action (BOT-4662, P10)BOT-4688 — QACF QA across S01–S04Done — coverage completed; no QA man-days estimated

Izzul (QA, backup)

Agent brief: short queue by design — you are backup QA. Row 2 is a five-minute board fix that clears a standing reconciliation finding, so do it even though it isn't test work.

#Epic (Jira)TaskStatus
1QA BAU (BOT-4609)BOT-4637 — Sprint 1 automation-coverage increase, In ReviewReady
2QA BAU (BOT-4609)Reparent BOT-4461 ("Increase Coverage P0 Automation Test WEB") from BOT-3972 to BOT-4609. BOT-3972 is the 26Q2 BAU Epic and is itself Done — this is the quarter rollover qa-bau.md describesReady — closes the last live part of F-41
3Product initiativesBackup QA for Tito's queue on request; capacity records 0 committed daysReady on demand

How to point your AI agent at this plan

Copy-paste brief:

You are working the Chatbot 26Q3 Sprint 2 plan at chatbot/delivery/sprints/2026-07-27-sprint-02.md. Find your engineer's ### <Name> section, read the Agent brief, and work the numbered queue top-to-bottom. For each row, the task breakdown linked from the Initiative index is the work contract — read the task's What to build, Implementation steps, Acceptance criteria, and Run to verify before writing code. Stop and ask if a row's gate is still closed.

Ground rules:

  • The task breakdown is the contract. Do not re-derive scope from the PRD or from this sprint doc; this doc only says which tasks and in what order.
  • Man-days are sequencing context, never a cap or a deadline. Capacity lives in ../capacity.md.
  • A Jira key must reach the default branch. Put the key in your branch name and commit messages. Reconciliation counts merged code, not board status: 21 of sprint 1's 30 done issues were uncorroborated and contributed 0 man-days.
  • Never touch the jira-sync / main-full automation. Board items for this sprint are filed by sprint-jira-sync (manual, per-user-story). The per-AC pipeline is a different convention and pointing it at these Epics would create duplicate Stories.
  • Do not invent contract fields. Where a row says blocked on Data/BI, Data & AI, or an OQ, use a clearly-labelled placeholder and leave the gate visible.
  • Do not renumber a published AC or story id. It re-keys its Jira ticket.

Deviations from roadmap priority

  • Landing the in-flight tail outranks rank this window. 28 items entered the sprint mid-flight and every queue starts with them, so lower-ranked work (P9, P10, buckets) appears above higher-ranked new build in several queues. This is sequencing on state, not a reprioritization — the roadmap order is unchanged. It should not persist beyond this window; if it does, the tail is structural and belongs in a decisions/ ADR.

    The seven specific rank inversions this produces (each raises an advisory SPRINT_QUEUE_ORDER in lint:delivery — all intentional, enumerated here so none is silent):

    QueueInversionWhy
    Samudra Linggar (FE)P1 below P4P1 has no remaining FE build — all 6 BE/FE issues are Done. Its row is a support note for Tito's QA close-out, so it sits last by nature, not by deprioritization.
    Eko Aprianto (BE)P4 below P10The P10 row is now rollout-complete evidence reconciliation (F-26: BOT-4668 is descoped; BOT-4669/4673 remain board-done without code), not feature work. The GA record is complete; only board truth remains.
    Eko Aprianto (BE)P3 below P4P3 BE is sequenced behind Farras's four merges (his row 1). P4 Tasks 2.1–2.3 are startable now; P3 support is not.
    Tito Amanda (QA)P1 below P9Sprint-Goal-1 Google Calendar QA is carried over from sprint 1 and was already a documented deviation there. Same reason, one window later.
    Tito Amanda (QA)P2 below P3Both are In Testing; P3's test-case creation gates P3's close-out this window, while P2's gates a build that has not started (no tickets exist).
    Baghiz Firdausa (FE)P2 below P102026-07-27, from the row-1 split: rows 1a–1c were tail-first completion work; the QACF slice is now closed (BOT-4724 done, BOT-4727 re-parented/done) while P2 has no tickets to start (F-2/F-42).
    Samudra Linggar (FE)P4 below P102026-07-27, from the row-1 split: rows 1b–1d were tail-first release verification; QACF BOT-4725 is now done in FE 3.8.4, not new build.
  • 2026-07-27: the two Epic = Multiple tail rows were split one-row-per-Epic. Samudra Linggar's and Baghiz Firdausa's row 1 each bundled a mid-flight batch spanning four Epics under a Multiple Epic cell. Multiple joins to nothing in the ## Initiative index, so sprint-drift-audit could not attribute those 9 items to an initiative or rank, lint:delivery Group I could not check their queue order, and their merged-code evidence had no initiative to roll up to at close — precisely the items most likely to drift. They are now rows 1a–1d / 1a–1c, each carrying its real Epic key, kept contiguous at the top so the tail-first rule still reads. The two inversions this made visible are in the table above. No scope, staffing, or rank changed.

  • Three SPRINT_EPIC_DRIFT advisories are expected under the umbrella model. Autonomous (BOT-4602 ≠ README BOT-4601), Native Integration (BOT-4482 ≠ BOT-4600), and Qontak Action (BOT-4662 ≠ BOT-4597) each cite the child Epic where Stories/Tasks actually hang. Do not resolve these by collapsing rows onto an umbrella anchor — see the ## Initiative index comment.

  • P2 (rank 2) starts implementation only now, in sprint 2. It has been ranked since 2026-07-02 with a complete 33.5d breakdown and has never had a single BE/FE ticket. This is the standing escalated finding F-2/F-42, not a planning choice — recorded here so the delay is visible rather than re-absorbed. Ticket creation is a prerequisite for Baghiz's row 2.

  • P4 gets a real build queue while P5 does not. P4's task breakdown landed during sprint 1; P5's still does not exist and P5 has no implementors:. This follows readiness, not rank — P5 remains limited to breakdown generation and staffing.

  • The 26Q3 Fast-track lane is NOT rendered as queue rows. Epic BOT-4554 carries three live In-Progress production bugs (BOT-4739 — Dhita Putri Pratama, BOT-4746 — Eko, BOT-4741unassigned). It maps to no initiative and no bucket in this repo, so per doctrine it is flagged to the PM for scoping, not invented into a plan. Eko's row 1 records only the item he already holds. Two knock-ons: Dhita Putri Pratama is not in roster.json or capacity.md at all (F-46), and BOT-4734 (Baghiz) has no parent Epic — both need a PM/TPM decision this window.

  • Engineering bucket A1's tracker status contradicts the board. The tracker rows A1 as backlog while Tasks 13–14 are In Review/In Progress and Tasks 1–12 are already board-done. The bucket's ## Expedite log also still shows Authorized by: TBD for a 🔴 Expedite that has been executing for two weeks. Both are tracker-truth fixes owed by the DRI; this sprint renders the verdict and does not re-class the story.

  • P6–P14 remain backlog-visible only. Pulling lower-ranked work into Sprint 2 requires an explicit PM/TPM decision and should be reflected in the roadmap or a delivery decision if it persists.

Adhoc interrupts (this sprint)

One 🔴 Expedite is in flight, rendered from the bucket tracker — not originated here:

Bucket (Epic)StoryClassOwnerStateVerdict (from Pull rule)
Chatbot 26Q3 Engineering Initiatives (BOT-4620)A1 — Ruby 3.4 / Rails 7.2 Phase 1🔴 ExpediteAgus (BE), Tito+Izzul (QA)Tasks 13–14 in flight, ~13d oldPULL-TO-CLOSE — finish in this window; WIP limit 1 respected, no second Expedite may enter

The 26Q3 Fast-track lane (BOT-4554) is deliberately absent from this table: it is not a declared bucket, so it has no Class-of-Service, no Pull rule, and no WIP limit to render. See Deviations.

The real blockers (not capacity)

  1. P2 AI Agent Impact Report — no implementation tickets exist. The nearest-term blocker is internal, not external: 0 BE/FE issues under BOT-4568 (F-2/F-42, escalated since sprint 1). Ask: TPM/DRI to run jira-alignment-check then sprint-jira-sync before Baghiz's row 2 starts.
  2. P2 Phase 2 — Data/BI ai_activity_logs datamart + open RFC constants. Ask: confirm fields, per-org aggregation rules, freshness, access path, owner, AI add-on feature code (OQ-1), canonical FE role strings (OQ-2), after-hours source, and the forecast algorithm (REV-2/OQ-10).
  3. P3 AI Agent Testing — AI platform batch shadow-inference capacity, plus the F-14 descope decision: BOT-3949/4616/4617/4618 are Won't Fix with no code, and three of them are the only breakdown-joinable done tasks besides Task 2.4. Until the descope is confirmed, P3's progress is capped at the recorded 7% floor. Ask: assignee/DRI + Dimas to confirm descope and cut the 37.5d baseline, or correct Jira.
  4. P4 AI Agent Knowledge — Data & AI contracts (PII masking, training/vector service, training-status webhook) gate Tasks 2.4–2.9. Ask: confirm payloads, retry semantics, 90-day rolling sync, source lifecycle, and masking SLA.
  5. P4 effort figures now conflict. The new breakdown totals 30.5d (FE 7 / BE 16 / QA 7.5) while the README and capacity.md still carry 13d (FE 1.5 / BE 8 / QA 3.5) — a 17.5-day gap on a ranked initiative. Ask: DRI to reconcile the canonical README effort_*_days against the breakdown and re-mirror to roadmap + capacity, then lint:delivery.
  6. 26Q3 Fast-track lane (BOT-4554) is unmodeled. Three live production bugs, one unassigned, one held by a person absent from the roster. It consumes capacity invisibly and cannot be reconciled. Ask: PM/TPM to declare it (a bucket initiative with a Class-of-Service tracker, or an explicit out-of-scope support lane), add Dhita Putri Pratama to roster.json + capacity.md, assign BOT-4741, and give BOT-4734 a parent.
  7. P5 Unified Agent Quality Scorecard — no task breakdown, no implementors. The one sprint-1 outcome target that did not move. DSAI rubric wording is explicitly non-blocking for Phase 1 per its own PRD, so this gap is internal. Ask: DRI/TPM to generate the breakdown and name implementors.
  8. P9 Google Calendar — external app verification. Production Cloud project, exact redirect URI, authorized-domain ownership, consent-screen links/contacts, least-privilege scope justification, and an English demo video. Sensitive-scope verification is Google's decision; testing proves readiness and cannot guarantee approval.

Sprint review

  • Shipped / done: TBD at close.
  • Carried over: TBD.
  • Added mid-sprint: TBD.
  • Fed into status: the status/<date>.md reports updated with this sprint's progress (link them).
  • Decisions taken: link any delivery/decisions/ ADR + roadmap Priority reorder this sprint triggered.