AI portfolio operating system

Track every AI initiative from intake to evidence-backed production status.

HPS needs a durable portfolio view beyond periodic email updates. The portfolio layer connects PMO scope, business objective, current status, blockers, decision owners, delivery evidence, technical readiness, and release posture in one controlled operating view.

Review AI portfolio register Review operating model SprintLoop product brief Download register CSV

July 1 AI portfolio cockpit 16 initiatives tracked

Active delivery 9

AWS proposal alignment, ACA pathway, GroupLink, claims, SQL Talk, ReconLink, voice assist, ET logs, and benefits/enrollment work.

Decision dependent 4

AWS proposal, VOC QA, transcription audit, and Group contact center AI require proposal, lead, platform, or proceed/defer decisions.

Queued 3

MHC IVR, self-service/BPaaS call center, and Phase 3 analytics/sentiment remain sequenced future-state lanes.

Production / complete 2

Recon Buddy phase 1 is live and image-to-text conversion is operational.

Current AI initiative register | July 1, 2026

A single status surface for work in progress, queued work, held work, and delivered capability.

The register converts project-update narrative into a repeatable portfolio object. Each initiative carries PMO identity, domain, current phase, health, blocker, next action, and evidence need.

Total tracked 16

AI and modernization initiatives across strategy, ACA, contact center, GroupLink, ReconLink, IVR, data access, and document processing.

Today’s strategic focus AWS proposal

Formal proposal remains the next decision input after the June 24, 2026 preview; use it to align modernization scope, sequencing, and economics.

Key leadership decisions 4

AWS proposal decision path, VOC discovery ownership, proceed/defer decision, and Group Contact Center AI SME / BA alignment.

Operational proof 2 live

Recon Buddy phase 1 and image-to-text conversion provide production-operating evidence.

Portfolio source files Structured register available for reporting, import, or future portfolio-system buildout.

CSV JSON

Initiative	Domain	Status	Current update	Next action / decision
AI Transformation & Platform ModernizationPROGRAM	Enterprise modernization strategy	AWS proposal pending	AWS proposal preview was reviewed June 24, 2026. Formal proposal remains the next decision input to confirm modernization approach, sequencing, staffing model, and delivery economics.	Use the strategy slides to align on transformation approach, modernization factory pattern, AI SDLC delivery model, and decision path once the formal AWS proposal is received.
Broker / Agent PathwaysACA / EDE	ACA platform and regulated enrollment	Demo / evidence ready	Broker and agent pathway demo, security architecture, CMS/EDE evidence walkthrough, traceability, investment estimate, and documentation are available for today’s run-down.	Use as the regulated delivery proof point: access, enrollment workflow, evidence, security posture, and auditor-ready operating model.
GroupLink Portal ModernizationGROUP PLATFORM	Group administration modernization	Modernization proof	Modern GroupLink experience demonstrates the pattern for moving beyond UI refresh into guided workflows, data/service modernization, and reusable platform components.	Position as a dual-track modernization lane: improve the front-end experience while moving database, service, and integration components forward.
CCAI ClaimsPMO 3684	Claims contact center	Risk: complexity	Priority 1 Claims Contact Center AI covers voice IVR for eligibility/benefits, claims status, and claims payment. BRD is in progress with daily meetings to define requirements.	Review Ops Portal APIs and data exchange; testing follows. Target is Aug 2026.
SQL Talk for KPPMO 3418	Membership data access	Business UAT	Conversational AI for internal teams to query Kaiser Permanente membership data using natural language.	Confirm July 1, 2026 production readiness, capture deployment decision, and update release evidence.
ReconLink Platform ModernizationPMO 2764	Reconciliation operations	UAT remediation	Two open UAT issues remain: EDI policy ID mismatch and overly verbose recommendation output flagged by Ops analysts.	Close defects, complete Round 2 business working session, and obtain signoff.
Recon BuddyPMO 2762	Recon inquiry assistant	Live phase 1	Phase 1 has been live since Dec 2025. Phase 2 NERD Next integration accuracy improved from 35% to 42% across three test rounds.	Resolve content-quality gaps before expanding the integration scope.
CCAI Voice Assist - Benefits & EnrollmentPMO 2202	Benefits and enrollment contact center	Release in progress	Spanish-language IVR release planned this week. Call recording vendor replacement is underway following the Verint rollback.	Deploy Spanish call flows and stabilize recording dependency for the next release path.
CCAI Agent Assist ET Log CreationPriority 3 · PMO 2022	After-call work automation	Risk: Verint dependency	Uses agent/member voice transcripts to create ET logs and reduce after-call work by 30 seconds. Design discussions are underway.	Resolve the 48-hour SLP posting KPI, Verint dependency, and cases where 100% log creation cannot be guaranteed. Target is Oct 2026.
Contact Center VOC QA AI AgentPMO 3636 / 3938	Quality automation	Discovery	Priority 2 quality-scoring automation for Voice IVR and agents, aligned to call scoring worksheets and procedures for 100% call audits.	Design discussions have started with AI team and business SME. Phase 1 focuses on CCAI Voice IVR checklist guidelines; Phase 2 adds voice call with screen capture. Target is Sep 2026.
CCAI Transcription Audit AutomationPMO 3636	Transcription QA	On hold	Cost-benefit analysis is complete and ready for leadership review.	Leadership decision required to proceed or formally defer.
MHC IVR Billing & Enrollment MigrationPMO 3912	IVR modernization	Queued	Migration of MHC Billing and Enrollment IVR call flows into the CCAI platform to consolidate the voice experience.	Prioritize after Claims MHC delivery.
CCAI - Group / Benefits & EnrollmentPMO 4014	Group platform voice AI	Risk: timeline / integration	Priority 4 voice assist AI agent for current/future Group platform covering eligibility, enrollment, tier structure, updates, billing, payments, and renewals.	Align SME and BA ownership, clarify new platform integration and GroupLink ownership, then determine cost/benefit on 4 FTE headcount.
Self-Service / BPaaS Call CenterIndividual and Group	Future-state contact center strategy	Yet to start	Discovery with Strategy Office for WHPS; researching future web-based contact center model via Cisco.	Document future-state direction and use cases for POC using internal and external sources.
Contact Center AI Phase 3Analytics and sentiment	Service intelligence expansion	Yet to start	Analytics and reporting, sentiment analysis, and additional agent-assist features for predictive help systems.	Hold strategic design discussions on platform changes required to support these functions.
AI Seed Image-to-Text ConverterPMO 3013	Document processing	Operational	Tool converts seed-data images to structured text and reduces manual data-entry effort for document workflows.	Continue production support and resolve defects as reported by the business.

Market signal

Current tools solve pieces of the problem. The agentic delivery gap is the controlled visibility layer.

Jira Product Discovery, Productboard, Aha!, Linear, GitHub Projects, GitLab, and Azure Boards all validate the same market need: leaders want roadmaps, requirements, source-of-truth status, and delivery traceability. The missing layer is agent-aware execution visibility with evidence and approval gates.

Discovery and roadmap Jira Product Discovery / Productboard / Aha!

Strong at idea intake, prioritization, stakeholder roadmaps, and syncing product intent into delivery tools.

Jira discovery signal

Team execution Linear / GitHub Projects / Azure Boards

Strong at issue views, custom fields, project updates, delivery plans, pull request linkage, and team workflow.

GitHub Projects signal

Flow analytics GitLab value streams

Strong at measuring idea-to-production cycle time, bottlenecks, merge request flow, and SDLC throughput.

GitLab value stream signal

Agentic coding GitHub Copilot coding agent

Agent work can now start from issues and produce pull requests, increasing the need for non-developer visibility.

Coding agent signal

Operating model

The portfolio platform reconciles what was requested, what is blocked, and what evidence proves readiness.

The core product promise is simple: every stakeholder request stays connected to the requirement, the agentic work plan, the code change, the evidence packet, and the release gate.

01 Intent capture

PM, BA, operations, security, or leadership submits a request with value, risk, deadline, and acceptance intent.

02 Requirement reconciliation

Agents decompose the request into requirements, acceptance criteria, dependencies, and testable scope.

03 Agentic execution plan

Work is assigned to specialized agents and developers with risk tier, model/tool policy, and human-review gates.

04 Source and delivery signals

GitHub or GitLab branches, commits, PRs/MRs, checks, reviews, and release notes update status automatically.

05 Evidence package

Screenshots, JSON manifests, QA results, SAST, vulnerability scans, architecture notes, and audit mappings are captured.

06 Approval and promotion

Low-risk items can move to lower environments for product review; production promotion remains governed.

Portfolio control model

HPS manages AI work through a real portfolio object model, not manually rewritten updates.

A mature deployment inside HPS lets every initiative be reviewed from executive summary down to requirements, work items, release evidence, dependency health, and decision history.

01 Intake and PMO identity

PMO ID, sponsor, business owner, domain, objective, expected outcome, requested date, and delivery lane.

02 Status and health

Phase, RAG health, confidence, blocker age, planned date, target release, last update, and next checkpoint.

03 Decision queue

Leadership decisions, owner, requested-by date, impact if delayed, and proceed / defer / redirect outcome.

04 Evidence and release gates

UAT results, screenshots, test runs, architecture notes, security review, defects, signoff, and production proof.

05 Dependency map

APIs, schemas, data warehouses, vendors, business SMEs, call flows, recording systems, and platform decisions.

06 Portfolio rollup

Auto-generated executive summary, domain health, blockers, upcoming releases, completed outcomes, and aging risks.

Capability blueprint

The site becomes a visibility product, not just a status page.

This is the natural next step for the WHPS command center and the missing product shape for SprintLoop: a leadership-safe visibility platform purpose-built for agentic software delivery.

Portfolio command center

Executive portfolio posture, delivery confidence, decisions needed, blocked work, and current proof in one view.

Requirements ledger

Submitted requirements, accepted scope, generated tasks, test criteria, and delivered artifacts reconciled line by line.

Change request path

PMs and business owners submit changes that are risk-scored, routed, estimated, and attached to the existing workstream.

Agent run observability

Agent assignments, run history, files touched, confidence, blocked decisions, and human-review triggers become visible.

Approval gates

Architecture, security, QA, compliance, and production release gates are explicit instead of buried in chat or code review.

Evidence vault

Control mappings, screenshots, manifests, test runs, vulnerability results, and release packets are browseable by audience.

Stakeholder views

Leadership, PM, BA/QA, developer, security, and auditor views show different depth without creating different truth.

Integration fabric

GitHub, GitLab, Jira, Slack, CI/CD, SAST, test runners, and deployment systems feed status instead of manual updates.

SprintLoop product direction

SprintLoop should become the interface between product leaders and the agentic workforce.

SprintLoop Code already owns specs, tasks, agents, runs, reviews, testing, evidence, audit, and release readiness. The upgrade is to package those pieces into a PM-visible product experience: request intake, status, traceability, and governed promotion.

Product lane SprintLoop Code

Owns AI SDLC control, source governance, agent work, requirements reconciliation, evidence, and release readiness.

Collaboration lane SprintLoop Communicator

Owns human collaboration, update channels, messaging, calls, notifications, and stakeholder conversations.

Leadership lane Visibility Console

A shared surface that translates agentic delivery into status, risk, decisions, proof, and business outcomes.

Build brief

Agentic Portfolio Visibility Console

Build a role-aware console where product owners can submit change requests, see requirement-to-delivery reconciliation, track agent and developer work, review lower-environment output, and approve controlled promotion.

Show what was requested, what was accepted, what changed, and what evidence proves it.
Integrate GitHub/GitLab issues, branches, PRs/MRs, CI checks, reviews, and release notes.
Separate low-risk review deployment from governed production merge and promotion.
Provide executive status summaries that are generated from live work evidence, not manual slide assembly.

Recommended build sequence

Move in controlled increments: visible first, automated second, governed promotion third.

The first release should make the status truth visible. The second should reconcile requirements to work. The third should safely route change requests into agentic delivery.

Phase 01 Visibility Console

Surface portfolio health, workstreams, requirements, agent runs, Git status, evidence, and decision queue.

Phase 02 Requirements Reconciliation

Link submitted asks to decomposed requirements, tasks, acceptance criteria, tests, PRs, and evidence.

Phase 03 Change Intake

Create PM-facing change requests with AI scope assessment, risk tiering, route recommendation, and review target.

Phase 04 Controlled Promotion

Route low-risk changes to review environments and escalate production approvals to developers and release owners.

WHPS proof alignment

The current transformation site is already the prototype pattern.

MarketLink evidence, Broker/Agent pathways, GroupLink modernization, AI SDLC, and ControlFrame-style evidence collection can become the first proof set for the portfolio visibility model.

MarketLink Platform walkthrough

End-to-end operating path from broker access to enrollment and evidence.

Evidence vault Control evidence walkthrough

Folder-style evidence repository with screenshots, manifests, mappings, and proof artifacts.

GroupLink Modernization proof

Current-to-future portal narrative with user experience, workflow, and modernization strategy.

AI SDLC Governed delivery method

Risk tiering, model/tool gates, human-in-the-loop review, evidence, and release controls.