PI Calendar¶
FCC's Program Increments run on a 10-week cadence aligned to the Research Center (RC) quarterly rhythm. Each PI starts on the first day of a calendar quarter (Q3 = Jul 1, Q4 = Oct 1, Q1 = Jan 1, Q2 = Apr 1) and closes on the last day of that quarter. Inside each PI: five 2-week iterations.
Rolling 4-PI horizon¶
| PI | Start | End | FCC release | Focus | RC-aligned milestones |
|---|---|---|---|---|---|
| PI-2026-Q3 (current) | 2026-07-01 | 2026-09-30 | v1.4.1 → v1.4.2 | ice_ext absorption + SAFe instance bootstrap + vocabulary-parity hardening | RC PI-2026-Q3 ecosystem-integration-tests CI gate maturation; RC NEXUS v2.1.0 registry drift audit |
| PI-2026-Q4 | 2026-10-01 | 2026-12-31 | v1.4.3 → v1.4.4 | Real seed data foundation prep + GraphRAG scaffolding + event-split (PE-008) refactor | RC portfolio review; JV partner alignment audit |
| PI-2027-Q1 | 2027-01-01 | 2027-03-31 | v1.5.0-rc → v1.5.0 | v1.5.0 pillar convergence (real seed data, GraphRAG, patent delegation, POLARIS/LYRA) | RC v2.2.0 cross-ecosystem contract freeze |
| PI-2027-Q2 | 2027-04-01 | 2027-06-30 | v1.5.1 → v1.5.2 | v1.5.0 hardening (i18n, multi-user collab, advanced federation, perf optimization) | RC portfolio-review quarterly sync |
Current PI objectives (PI-2026-Q3)¶
Committed (3-5 per ART):
- Framework ART:
- Land the SAFe instance scaffold (this
docs/safe/tree +make safe-report+ 5 new personas SMC/SFBI/TLDR/VDS/ZCA). - Absorb ice_ext classification patterns into
src/fcc/classification/. -
Register POLARIS + LYRA v1.5-preview plugins for PE-009 readiness.
-
Ecosystem ART:
- Integrate
ice_ext_element_mapping.yamlinto packaged vocabulary; add ice_ext provider to VDS parity audit. - Re-baseline SHA-256 manifests across all 13+ VocabularyProviderPlugins.
-
Publish the v1.4.1 bridge report under SFBI ownership.
-
Docs ART:
- Publish
docs/safe/into MkDocs nav at v1.4.1 integration. - Re-run TLDR reconciliation against all 10 constellations + ice_ext + research_center_ext.
- Regenerate 173 → 178 model cards (5 new personas landing).
Uncommitted (stretch):
- SAFe audience tutorial under
docs/for-teams/(or a v1.3.10-era successor tree). - SAFe-instance Streamlit walkthrough demo.
Predictability target: 80-100% of committed objectives hit. Below 60% triggers PI-end Inspect & Adapt.
Iterations inside PI-2026-Q3¶
| Iteration | Dates | Theme | Type |
|---|---|---|---|
| I-1 | 2026-07-01 → 2026-07-14 | v1.4.1 scaffolding (this scaffold lands here) | Delivery |
| I-2 | 2026-07-15 → 2026-07-28 | ice_ext absorption hardening + vocabulary-parity CI | Delivery |
| I-3 | 2026-07-29 → 2026-08-11 | POLARIS/LYRA preview graduation prep + PE-009 backlog refinement | Delivery |
| I-4 | 2026-08-12 → 2026-08-25 | v1.4.2 release (real seed data foundation prep) | Delivery |
| I-5 (IP) | 2026-08-26 → 2026-09-30 | Innovation & Planning: PI-2026-Q3 close, system demo, Inspect & Adapt retrospective, backlog-refinement for PI-2026-Q4 | Innovation & Planning |
Note: I-5 extends to PI-end rather than a clean 2-week slice because FCC uses the final iteration as a dedicated Innovation & Planning (IP) iteration. During an IP iteration the ARTs reduce feature velocity (target ≤50% of a normal iteration), absorb feedback from the system demo, close out PI ceremonies, and refine the next PI's backlog. This is aligned with SAFe 6.0 IP-iteration guidance and kept on a 4-to-5-week slot because FCC's release-oriented PI has more ceremonial overhead than a pure SAFe PI.
IP iteration pattern (per PI)¶
Every FCC PI ends with an IP iteration that absorbs:
- System demo across all three ARTs (SMC publishes the PI scorecard).
- Inspect & Adapt retrospective (mandatory if predictability fell below 60%).
- Backlog refinement for the next PI (BC/JVC/EC-led sessions).
- Enabler catch-up work: CI hygiene, vocabulary parity re-baselining, three-layer reconciliation sweeps.
- Any pattern-absorption proposals PHV surfaced during the PI.
Feature delivery is explicitly down-weighted during IP iterations to protect this capacity; SMC's flow-efficiency metric excludes the IP iteration from normal-iteration trend lines.
Holiday skip-weeks¶
Holiday skip-weeks are intentionally absorbed into the adjacent IP iteration rather than extending the PI boundary. Known holiday windows on FCC's calendar:
| Window | Dates (annual) | Absorbed into |
|---|---|---|
| US Independence Day | Jul 04 (± 2 days) | PI-Qn Q3, I-1 |
| US Labor Day | first Monday of September | PI-Qn Q3, I-5 (IP) |
| US Thanksgiving | fourth Thursday of November (+ Friday) | PI-Qn Q4, I-5 (IP) |
| Winter holidays | Dec 20 – Jan 3 | PI-Qn Q4, I-5 (IP) — PI closes earlier, IP compressed |
| Easter + Passover | variable (late March / mid-April) | PI-Qn Q2, I-1 |
When a holiday window falls into a normal Delivery iteration, velocity expectations for that iteration are re-baselined (commonly by one story-point-equivalent); SMC excludes the impacted week from iteration-level trend lines so flow metrics stay comparable.
PI ceremonies¶
- PI planning — ~1 week before PI start (scheduled: 2026-06-23 through 2026-06-30 for PI-2026-Q3).
- System demo — Last Friday of each PI. SMC publishes PI scorecard; each ART demos committed objective completion.
- Inspect & Adapt — First week after PI close. Root-cause retrospective if predictability fell below 60%.
- Portfolio sync — Once per PI, late in the iteration cycle. SFBI aligns with RC's portfolio tier to confirm vocabulary + cadence stay synchronized.
RC-alignment notes¶
FCC's PI boundaries align with RC's but FCC's scope is narrower (one authority repo + 10 constellation verticals versus RC's 20+ projects). When RC's PI planning or I&A retrospective identifies a cross-project dependency affecting FCC, SFBI authors a bridge report so the dependency lands in FCC's backlog with source + line references.
Key RC-alignment signals tracked here:
- RC NEXUS v2.1.0 registry drift audit (quarterly).
- RC portfolio review outputs affecting FCC's Framework or Ecosystem ART.
- Cross-project ADR references where RC and FCC co-own a decision.
- JV partner alignment audits RC runs quarterly and FCC consumes at PI start.
Historical PIs (informational)¶
For continuity with the v1.3.x series predating SAFe adoption, historical release cadence was not PI-structured. v1.4.0 closed the "ecosystem pivot" pillar; v1.4.1 bootstraps the SAFe operating model and begins the PI-2026-Q3 PI proper. Earlier release cadence (v1.3.5 → v1.3.11 → v1.4.0) is preserved in CHANGELOG.md and the per-release memory files.