Visual Field Guide
How honest AI tooling works, in one diagram each
A growing set of plain, reference-grade diagrams that explain the ideas behind RedHub's tools — the gate that overrides a good score, the two finish lines every AI output has to cross, and the deterministic stack that makes a verdict reproducible. Each one is free to share and embed with a link back.
Anatomy of a Gate
Every RedHub judgment tool builds a weighted score, then runs it past a gate that can only make the verdict worse — which is why a deliverable can score 89 out of 100 and still be told DO NOT SHIP. The score is context; the gate is the answer.
View & embedVisual Field Guide · No. 02Two Finish Lines
AI is built to reach the first finish line — it works — but real stakes need the second: it's defensible. The distance between them is the whole job, and every RedHub tool exists to carry you across that gap.
View & embedVisual Field Guide · No. 03The Deterministic Stack
RedHub's judgment layer isn't AI. One logic renders three ways — a runnable engine, a spreadsheet, and a live demo — verified to produce the same verdict to the byte, so the result is reproducible enough to hand an auditor. The AI drafts; the deterministic engine decides.
View & embedThe Math
What the calculators really add up to
The same shareable treatment for RedHub's calculator kits — the exact arithmetic behind each number, worked through one real example, as a diagram you can embed.
What a meeting really costs
A recurring meeting is a standing salary expense no one approved. The Meeting Cost Calculator prices every one — attendees × (salary ÷ 2,080 hrs) × length × yearly frequency — so a single weekly 9-person standup surfaces as ~$19,700 a year, and you see exactly how much is flagged to trim or kill.
View & embedThe Math · How it worksWhich invoices to chase first
The kit ranks every open invoice by amount, age, and risk into Current / Remind / Chase now / Write-off review. In the built-in six-invoice example, three flagged Chase-now invoices total $19,800 recoverable this week out of $33,200 outstanding.
View & embedThe Math · How it worksWhat a small task really costs
A two-minute task you do all day is a full salary line in disguise. The calculator prices every task — frequency × time × loaded rate — then nets out AI run-cost to return payback and ROI, including the tasks it tells you not to automate.
View & embedThe Math · How it worksWhat this AI tool really nets
The kit nets each AI tool as (hours saved − oversight) × loaded rate − subscription − setup/12. The example AI outbound tool saves 6 hours but costs 5 hours of oversight, so 1 net hour × $50 against $600 spend lands at −$550/mo — a Kill.
View & embedThe Math · How it worksWhat idle seats really cost
For a Right-size tool the kit recovers cost × (paid − used) ÷ paid — the idle-seat share of the bill. On a $250/mo plan with 4 of 10 seats idle, that's $100/mo recoverable, $1,200/yr on one line item.
View & embedThe Math · How it worksThe channel that quietly loses money
One acquisition channel can cost more to win a customer than that customer ever pays back — and blended CAC hides it in the average. The calculator runs CAC, payback, LTV and LTV:CAC on every channel and totals the spend sitting in Cut channels you can reallocate.
View & embedThe Math · How it worksThe bench you never billed
A billable team at 71% utilization feels slammed while sellable hours walk out the door every week. The planner prices the bench, flags who's overloaded versus idle, and checks whether you can take a new client before you say yes.
View & embedThe Math · How it worksWhat an empty seat really costs
Cost-per-hire is tracked; the cost of the seat sitting empty is the bigger number nobody prices. The calculator puts an honest daily and net figure on every open role — gross impact minus the salary you're not paying while it's vacant.
View & embedThe Math · How it worksThe promo that lost money
A 20%-off promo can move 1,000 units and still lose money — because most of those buyers would have paid full price. The analyzer separates incremental from cannibalized demand and gives every promo a keep / limit / kill verdict.
View & embedThe Math · How it worksCash stuck in slow stock
Inventory isn't cash until it sells. One overstock SKU — 10 months of stock when you need 2 — has $4,000 frozen on the shelf. The finder flags every dead and slow SKU and totals the trapped cash plus the carrying cost to hold it.
View & embedThe Math · How it worksWhat slow replies quietly cost
Reaching 80% of leads instead of 40% doubles customers — 20 to 40 a month — and at $2,000 each that's $480,000 a year left on the table. The lift comes entirely from your own contact-rate inputs, never a baked-in multiplier.
View & embedThe Math · How it worksThe client that loses you money
A $30,000-a-year client can still run at a loss once you count the hours you burn servicing it. The kit ranks every account by true profit and returns a keep / reprice / fix / fire verdict.
View & embedThe Math · How it worksThe 37% margin that isn't
A product with a healthy 37% gross margin can still lose money once returns are netted out. The kit nets return cost against kept margin per SKU and returns a keep / fix returns / drop verdict.
View & embedThe Math · How it worksThe customers you can afford to lose
A price raise pays off until the customers you lose outrun the extra margin each remaining sale earns. The kit solves that break-even and returns a raise / hold-and-test / restructure verdict.
View & embedThe Math · How it worksWhat one piece of content really becomes
One real source piece a week, repurposed into 5 channels, is a 5x ratio that hands back about 30 hours - roughly 3.8 workdays - every month. The Content Engine makes that ratio repeatable without the posts turning to slop.
View & embedThe Math · How it worksWhy a "reliable" agent quietly fails
Eight chain steps that are each 85% reliable succeed end-to-end only 27% of the time - 0.85 to the 8th power. The kit computes true end-to-end reliability for every agent and flags the long chains before they quietly fail.
View & embedThe Math · How it worksWhat a 2/10 net 30 discount is really worth
A 2/10 net 30 early-payment discount is a 37.2% annualized return - (2 divided by 98) times (365 divided by 20). On an $18,000 invoice that is a real $360 for paying 20 days sooner, far above a 12% cost of capital - if the cash buffer can fund it.
View & embed