The Portal

The working surface where the Engagement runs.

The Portal is not a standalone product and is not sold separately. It is the display layer for ongoing Engagement clients, refreshed monthly from the maintained Excel model. If you are not in an Engagement, the Portal does not have a useful version of itself to show you.

Reviewed monthly Reconciles to the model Excel calculation engine sits in your workbook, not behind a login
Where it fits

Sprint builds it. Engagement maintains it. Portal displays it.

The Sprint produces the calculation engine: a working Excel model that runs cash, margin, and scenarios from your data. That model is yours from delivery day.

The Engagement keeps it current. Each month, the model is updated against actuals, reviewed, and re-published. The numbers in the workbook drive the numbers on the Portal.

The Portal renders selected outputs from the maintained model in a permissioned, named-user view. Six tabs, each tied to a named range or block in the underlying workbook. Every line in the Portal reconciles back to a cell in your model. There is no opaque calculation layer between the two.

Stage 01 10 BUSINESS DAYS
Sprint
Builds the model. Cash, margin, scenarios from your data. Client-owned workbook.
DELIVERSExcel calculation engine
Stage 02 MONTHLY · 30 DAYS NOTICE
Engagement
Maintains the model. Updated to actuals each month. Reviewed. Re-published.
SUSTAINSReviewed monthly outputs
Stage 03 DISPLAY · NAMED-USER
Portal
Displays the model. Six tabs. Every line reconciles to a named range in your workbook.
RENDERSPORTAL_OUTPUT block
Reconciliation The Portal does not run a parallel calculation engine. Every Portal line traces back to a named range in your workbook. If the Engagement ends, the workbook leaves with you and the Portal is decommissioned. The calculation engine continues to run on your machine.
Six views

What the Portal displays.

Six tabs, one per operating question. Each one ties back to a specific decision class.

01Overview 02P&L 03Cashflow 04Forecast 05Hire Impact 06KPIs
Tab 01 / 06Overview

Overview.

The one-page read on the month.

  • Five headline tiles with month-on-month and prior-year variance arrows
  • Industry benchmark gauges on margin metrics where configured
  • Reviewed commentary card, three paragraphs, naming what to act on
  • Top-three anomaly strip surfacing the period's exceptions
  • Momentum card tracking the business against its own twelve-month trend
Decision useThe screen the owner reads first on review day to set what the conversation needs to cover.
Tab 02 / 06P&L

P&L.

Every line variance-analysed against budget or prior year.

  • Department, service-line, or location split with a Consolidated view
  • Favourable / unfavourable variance colouring
  • Ledger-level detail on hover for every aggregate line
  • Every row reconciles back to a named range in the Excel model
Decision useIdentify the line items that moved, separate noise from signal, decide what to investigate before the next cycle closes.
Tab 03 / 06Cashflow

Cashflow.

Opening, in, out, closing for the month, with the 13 weeks ahead.

  • Base, Upside, Downside scenario toggle on the 13-week forecast
  • Trough weeks highlighted on the chart
  • Monday-morning cash alert emailed to mapped users when the forecast crosses a configured minimum
  • Reconciles weekly to bank balance inside the Excel model
Decision useSee the cash squeeze before it hits, decide whether a cost, a hire, or a payment timing needs to move.
Tab 04 / 06Forecast

Forecast.

Twelve months of draggable revenue bars, full recompute live.

  • EBITDA, net profit, and closing cash recalculate as inputs change
  • Configurable minimum-cash threshold flagged in red
  • Baseline forecast available as a one-click reset
  • Same calculation engine and working-capital cycle as the Cashflow tab
Decision useStress-test a revenue assumption or a spending decision in the room and see the cash impact in the same minute.
Tab 05 / 06Hire Impact

Hire Impact.

Model any hire against actual margins.

  • Salary plus on-cost loading, configurable start month
  • Six-month ramp on expected revenue contribution
  • Break-even month highlighted on the chart
  • Cash impact runs through the same working-capital cycle as the main forecast
Decision useDecide whether the hire is a yes, a not yet, or a no, based on a number rather than a feeling.
Tab 06 / 06KPIs

KPIs.

KPI scorecard per period, with industry context where configured.

  • Current value, target, and on / watch / off status per tile
  • Six-period sparkline with month-on-month delta on every tile
  • Industry benchmark overlay
  • KPI definitions held stable across periods, editable per client in the configuration layer
Decision useTrack the operating metrics that drive cash and margin, not the vanity metrics that drive a dashboard.
Review Standard

Every output is reviewed before it publishes.

AI-assisted analysis is used inside the build process: data preparation, anomaly detection, draft variance narrative, draft commentary. Every output is then reviewed and signed off by the founder before it reaches the Portal. Nothing publishes unreviewed.

This matters more than it sounds. AI produces analysis. It does not take accountability for what the numbers mean. The published Portal pack is the operator's call, not the model's. That is the line that distinguishes a reviewed financial product from an automated one, and it is held deliberately.

Principle 01 01 / 03

Reviewed, not automated.

AI drafts. Founder reviews and signs off. Nothing publishes unreviewed.

Sign-offFounder
CadenceMonthly
Principle 02 02 / 03

Grounded in your context.

Outputs constrained to your twelve-month history, published commentary, and industry context.

Window12 months
SourceYour data
Principle 03 03 / 03

Client-owned engine.

The Excel model is yours. The Portal is decommissioned at exit; the workbook leaves with you.

EngineClient workbook
Lock-inNone
Operator accountability The published Portal pack is the operator's call, not the model's. That line is held deliberately. AI shortens the path between raw data and a reviewed pack. It does not become the pack.
Where the logic lives

The Excel model is the engine. The Portal is the display.

The calculation engine sits in the client's Excel workbook. Three-way forecast logic, working-capital cycle, KPI logic, scenario layer, and assumption inputs are all named ranges the client can see and audit. Trial balance flows in through an INPUT block. The CALC tab runs the model. A PORTAL_OUTPUT block is the rendering target for the Portal.

The Portal renders selected cells and ranges from PORTAL_OUTPUT. Every Portal line maps to a cell or named range in the model. There is no separate calculation happening inside the Portal that cannot be traced to the workbook. This is intentional. It is the central reason the Portal can be decommissioned at engagement exit without taking the calculation engine with it.

If the Engagement ends, the workbook leaves with the client. The Portal is decommissioned. The calculation engine continues to run on the client's machine. No residual dependency on Infraxus to keep it operational.

Client workbook · MODEL_v.2026.05
Block 01 INPUT Trial balance · client data
Block 02 CALC Named ranges · auditable
Block 03 PORTAL_OUTPUT Rendering target
INPUT.tb_revenue$1,420,000
CALC.revenue=INPUT.tb_revenue
PORTAL_OUTPUT.revenue$1.42M
INPUT.tb_cogs$822,860
CALC.cogs=INPUT.tb_cogs
PORTAL_OUTPUT.cogs$822.9k
CALC.gross_profit$597,140
CALC.gross_profitCALC.revenue - CALC.cogs
PORTAL_OUTPUT.gp$597.1k
CALC.gross_margin0.42056
CALC.gross_marginCALC.gp / CALC.revenue
PORTAL_OUTPUT.gm42.1%
INPUT.tb_opex$413,200
CALC.opex=INPUT.tb_opex
PORTAL_OUTPUT.opex$413.2k
CALC.ebitda$183,940
CALC.ebitdaCALC.gp - CALC.opex
PORTAL_OUTPUT.ebitda$183.9k
INPUT CALC PORTAL_OUTPUT KPI SCENARIO ASSUMPTIONS +
NAMED RANGES · audit any line, any time EXIT · workbook leaves with client · Portal decommissions
Reconciliation Every Portal number reconciles to a named range. You can audit any line, any time.
Query layer

Ask Your Numbers.

A query layer sitting over your reviewed financial context. Type a question in plain English; receive an answer constrained to the twelve-month history, published commentary, and industry context already inside your Portal. The query layer cannot reach beyond your reviewed context, cannot invent data, and does not generate new analysis without grounding.

It is a faster way to ask a question of the same numbers your model already runs, not a parallel intelligence with its own opinions. AI calls route through Anthropic under zero-data-retention API terms; queries are not retained and are not used to train models.

Decision usePull a number, a trend, or a comparison without opening the workbook, before deciding whether the workbook needs to be opened.
Next Step
01Fit call 02Sprint 03Engagement 04PortalYOU ARE HERE
Entry Continuation Display layer

The Portal sits inside the Engagement. The Engagement starts after a Sprint.

Most clients reach the Portal by completing a Sprint and choosing to continue. The fit call is the first step in that sequence.