The Portfolio App

The Portfolio surface gives you a rolled-up view of your programs and initiatives — their health, what's at risk, and how work aggregates from individual items up to the program level.

Because InaiBridge maps every source into one canonical record, the Portfolio spans all your connected tools at once: programs from Jira, Asana, and Azure DevOps appear together, each carrying its own honest confidence label.

Mira sits alongside that view, and answers from it. Every claim carries its own source, named inline as Mira makes it — the query that ran, and the individual register entry behind each item listed.

Beneath the answer Mira lists the sources drawn on, entry by entry — so any figure in the reply can be traced back to the record that produced it rather than taken on trust.

What you see

The Portfolio Assurance dashboard is organized top-to-bottom to help leaders reason about program health, financial exposure, and grounded risks. Accessing the dashboard initiates a silent Single Sign-On (SSO) check; if you are unauthenticated, you will be redirected to your organization's identity provider to sign in before live portfolio data is loaded.

The dashboard includes the following key areas:

• Program Health: An instant-loading visual breakdown of your programs. Each program is represented by a horizontal bar showing the proportion of initiatives that are done, in progress, or overdue.

• Portfolio Totals: High-level metrics showing the total number of programs, initiatives, overdue items, and the overall percentage of work on track.

• Money at execution risk (containment): The hero figure shows the whole budget of every program that is either overdue or has no confirmed (human-authored) risk assessment while work remains open. This is distinct from the per-program expected exposure figures below, which are computed as budget × (overdue share). The UI labels the hero explicitly as containment and explains the difference so the two measures are never conflated.

• Portfolio Insights: A single, foldable reasoning surface structured by Elios Tiers (Tier 1 Executive, Tier 2 Managed, Tier 3 Sustained). These tiers allocate executive attention and stewardship rather than ranking worth. Within this surface, you can view programs through two distinct financial lenses:

  • Cost Exposure (Consumes): A breakdown of what the program consumes in budget, including capex and opex allocations across cost centers. To prevent budget inflation, portfolio-wide financial totals automatically deduplicate shared cost centers that are linked to multiple programs, ensuring each center's budget is counted exactly once across the portfolio.
  • Opportunity Cost (Forgoes): The value foregone per month while a program is delayed — a figure your organization records with a written basis, not one the system derives. It applies only to revenue and strategic programs; for any other type the measure does not apply, which is not the same as zero.

• Risk Records: A foldable section grouping grounded risk records by program. Each card displays a risk score, category, status (such as Suggested, Authored, or Rejected), impact, likelihood, and confidence, alongside root causes, mitigations, and the actual field-drift evidence that triggered the risk.

• Risk Matrix Configuration: A foldable section displaying the active risk matrix. This matrix maps impact (from Negligible to Critical) and likelihood (from Rare to Almost Certain) on a standard 1–5 scale to resolve governed severity levels (Low, Medium, High, or Extreme). All users can view this configuration, while editing is restricted to owners to maintain governance.

• LLM Cost Ledger: A foldable section showing durable, organization-scoped LLM costs from the verified ledger. It breaks down spend and calls by surface (e.g., Mira replies, risk sweep, docs agent) and by program, with separate all-time and month-to-date windows. The month-to-date view includes a clearly labeled month-end run-rate projection (an estimate, not an actual).

Drilling in

Select a program in the graph to focus on it: the view narrows to that program's at-risk initiatives. Select it again (or choose "show all") to return to the whole portfolio.

Portfolio Insights takes this further, letting you move through three levels of altitude — from the whole portfolio down to a single program:

• Overview: Portfolio-wide totals and the cross-cutting themes behind the risk.
• Stewardship tier: Programs grouped by Elios Tier (Executive, Managed, Sustained) — showing where executive attention is concentrated, not which programs matter more.
• Program detail: The deepest level, where one program opens into four grounded views:
  • Cost centers: How the program's budget breaks down — its capital (capex) and operating (opex) allocations across each cost center.
  • What's driving this: A grounded, plain-language read on why the program is drifting — traced to the actual initiatives and field-level movement behind it, never generic advice.
  • Signal & context: A briefing that connects where the program stands now to comparable situations from the past and the leadership context on record.
  • Risk records: The program's grounded risk records — each with its score, category, and status, alongside impact, likelihood, confidence, root cause, mitigation, and the field-drift evidence that raised it.

Every figure and narrative here is grounded in your real data — nothing is invented. In a demonstration portfolio, financial figures are illustrative, not billed amounts.

Ask about what you're looking at

The Portfolio includes Mira, a grounded conversational companion for querying your live portfolio data. Mira answers only from your real portfolio data, and the panel always states what it's grounded in, so the scope of an answer is never ambiguous.

Grounding Scope

Mira is always aware of your active dashboard selection:

• Scoped Queries: When you select a program on the dashboard, Mira automatically scopes its reasoning to that specific program.
• Unscoped Queries: When no program is selected, Mira answers across the entire portfolio.

The panel explicitly displays its active grounding scope (e.g., "scoped to [Program Name]" or "whole portfolio") at all times, ensuring you always know the context of its answers.

Program Health Lenses

You can ask Mira to apply a visual overlay to the Program Health chart. When you ask Mira to "size by budget," "weight by risk," or "apply a financial lens," the chart's horizontal bars are dynamically reshaped:

• Bar Width: Scaled to represent the magnitude of the selected dimension.
• Bar Color: Remains mapped to program health (done, in progress, or overdue), preserving the health proportions within the scaled width.

Supported lens dimensions are strictly grounded in live portfolio data:

• Budget: Total program budget.
• Capex / Opex: Capital and operating expenditure splits.
• Opportunity Cost: The monthly value foregone due to program delays (for revenue or strategic programs).
• Risk Severity: The summed severity of all risks identified for the program.

You can also request a department-group lens, which aggregates bars by department (summing initiative counts across programs) while preserving health coloring. If you ask for a dimension that is not grounded in the live data (such as headcount or vendor), Mira will decline to apply the lens and explain what is missing.

You can reset the chart to its default equal-width view at any time using the reset button on the dashboard.

Voice and Language Controls

To support diverse executive workflows, Mira includes built-in speech and localization capabilities:

• Multilingual Support: Select from 25 supported languages (including English, French, German, Spanish, Japanese, Arabic, Russian, Cantonese, Mandarin, and Tamil) via the language selector. Mira replies in your chosen language — with speech input, spoken replies, and Right-to-Left (RTL) layout support. Ask about a specific program or risk by name in your language, and Mira grounds its answer in your live risk register and returns it in that language. Broader conversational understanding across languages is expanding; naming the program or item you're asking about gives the most precise results today.
• Voice Input (Speech-to-Text): If supported by your browser, you can use the microphone button to speak your questions directly to Mira instead of typing.
• Spoken Replies (Text-to-Speech): You can toggle spoken replies on or off. When enabled, Mira reads its responses aloud using your device's native text-to-speech voices.

What you're entitled to see

The Portfolio shows you the programs your role reaches. A program manager sees the programs they own; a portfolio lead sees a wider slice; an executive or risk officer sees the whole estate. This is applied when the data is read, not as a filter over a view already built.

One consequence is worth stating plainly, because it affects every number on the page: totals, counts and health proportions describe the programs visible to you, not necessarily your whole organization.

Honest confidence

Health and at-risk status come from your actual work items, mapped into one canonical record. A program is not shown as on track just because a source couldn't prove otherwise — each figure carries the confidence its source supports.

Program health is calculated deterministically based on the volume of overdue initiatives:

• On track (Green): Zero overdue initiatives.
• At risk (Amber): One or two overdue initiatives.
• Critical (Red): Three or more overdue initiatives.

This ensures that health is an objective reflection of actual work items rather than subjective manual updates.

Risk Promotion and Governance

To maintain absolute integrity in your risk register, InaiBridge separates risk detection from risk authorization through a governed, two-tier workflow.

Automated Drift Detection

The platform continuously monitors key project indicators—such as schedule health, due dates, status categories, and program health—for changes (drift). When multiple fields degrade simultaneously, the system consolidates these signals into a single Suggested risk. Sweeps are protected by per-organization rate limits and daily cost ceilings drawn from the verified ledger; these bounds prevent unbounded spend from automated loops while still allowing immediate assessment after each ingest. Refusals surface clearly so operators can tune the limits.

To determine the severity of a suggested risk, the system uses a deterministic lookup against your organization's active 5×5 Risk Matrix:

• Impact (1–5): Evaluated against explicit criteria ranging from Negligible to Critical.
• Likelihood (1–5): Evaluated against criteria ranging from Rare to Almost Certain.
• Severity Level: Resolved purely by mapping these coordinates to your configured grid (Low, Medium, High, or Extreme). This calculation is entirely deterministic and free of model bias.

The risk sign-off queue

All AI-detected risks arrive in the Inbox's risk sign-off queue as proposals. They do not become part of the official project record until they are reviewed by an authorized human. Each proposal displays:

• The Consolidated Risk: A clear description of the risk, its root cause, and potential consequences.
• Grounded Evidence: The exact field-drift transitions and project snapshots that triggered the detection, showing you exactly why the risk was suggested.
• Confidence and Rationale: An honest assessment of the placement clarity, including notes on any fuzzy data or single-signal limitations.

Human-in-the-Loop Decisions

Authorized organization owners or administrators can review these proposals in the Inbox and make a formal decision:

• Approve: Promotes the risk to Authored status, moving it into the active risk register as an open, tracked item. You can optionally append your own notes to the record.
• Reject: Soft-archives the proposal as Rejected, keeping an auditable record of the decision without cluttering the active register.

Owners and administrators may also author a new concern directly where the engine remained silent (human nomination). A revoke action on an already-authored risk creates a new signed ledger event moving it to REVOKED; the original record is never edited. Every promotion, rejection, nomination, or revocation writes a signed, cryptographically chained event to the governance ledger. This ensures that every change to your risk profile is backed by a verifiable, tamper-evident record of human authorization.

A decision made at any surface (inbox or workflow gate) produces the identical signed record and automatically resolves any related open gates so that no run is left waiting. A separate quiet panel for any program shows the field state the drift engine examined and found unchanged, explaining why no suggestion was raised.

Grounded Program Voice

The Portfolio surfaces one grounded executive sentence per program, built only from settled facts such as governed severity, overdue counts, and your organization's own anchor criteria. Lines proposed by the model are passed through a deterministic code guard; any invention of numbers, dependencies, forward time claims, or severity inflation causes an automatic abstain to a fact-only fallback line. Programs with no risk posture remain silent.

Two governance questions, two vocabularies

⚠ The Portfolio shows governance in two different senses, and they answer different questions. Reading one as the other is the most common way to misjudge a portfolio.

	The question	The states you'll see
Review posture (the program overlay)	Has anyone assessed this program's risk yet?	reviewed · awaiting sign-off · not yet reviewed · nothing currently flagged
Control compliance (the Governance Matrix and its findings)	Does this program comply with a control we put in force?	breach · checked-clean · not-assessed

A program can be reviewed for risk and still breach a control. A program can be checked-clean against every control and still have had no risk review at all.

And in both senses, the unexamined state is its own answer: not yet reviewed is not "no risk found", and not-assessed is not "compliant". Neither is a clean bill of health — they mean nobody looked. InaiBridge reports them as absences rather than letting silence read as safety.

For the figure that answers "how much of our budget is at risk of overrun?" — what it means, and why it is a risk flag rather than a spend forecast — see Budget at overrun-risk.

The mechanics behind these figures

How the numbers are produced and configured — stewardship inputs, cost centres, how the Elios tier is computed, the two money-at-risk measures, and the governance Controls register — is covered in Portfolio mechanics.