AI Workforce · Module
Investment & ROI Ledger
All-in investment and evidence-backed return for every agent, squad and client
The Investment & ROI Ledger records everything a business invests in its Digital Employees and every return they help produce, with attribution, baselines and holdouts. It produces a monthly Agent Investment Statement where every figure drills down to its evidence.
The heart of the rental model
Most AI pricing leaves buyers guessing whether it pays for itself. The Investment & ROI Ledger, included with every rental, is designed to answer that with evidence. A v0 is planned for W1, v1 for W2 and v2 for W3.
What gets counted
Ledger accounts exist per agent, squad, campaign, goal, client and agency. The investment side captures AI inference cost per trace, messaging pass-through per message from provider webhooks, daily accrued rent prorated on hire and return, agent-managed ad spend, avatar, video and voice render costs including re-renders, call costs, Human Expert Cloud escalations, amortised setup and human supervision time (approval and review timings times a tenant-entered hourly rate, always shown). Pass-through, ad spend and rent are reconciled against provider invoices, ad accounts and One3 invoices within a tolerance.
What gets credited
Outcomes such as won CRM deals, Billing payments and verified Desk resolutions link to attribution. Attribution models start with first and last touch (with a warning label), linear and position-based, with time-decay and data-driven models later. Rules are versioned per tenant. A baseline period is frozen at contract start. Confounder flags mark seasonality, discounts, price changes, existing pipeline, data gaps and human overrides. Holdout groups give lift with a confidence interval, and only holdout-validated returns are labelled caused; everything else is assisted. Pipeline value is shown beside ROI, never inside it. Below the data-quality floor, the statement shows "insufficient evidence" with fixes.
Guardrails in real time
A live spend meter shows today's spend against the daily cap and month-to-date investment. Daily and monthly caps apply per agent, squad and tenant, with a hard stop on R3 spend Actions. Outcome guardrails pause external-facing Actions when ROI stays below a minimum after enough opportunities. On a cap or pause, non-critical skills stop and conversations move to a human fallback queue with a holding message.
The monthly statement
The Agent Investment Statement has sections for summary, investment, returns, experiments, confounders, guardrail events and recommendations. Every figure drills down to ledger lines, traces, CRM records, invoices and payments. Recommendations (continue, adjust, retrain, replace or stop) are recorded as decisions when the owner acts. Disputes on specific entries within the window lead to a restated version; the original stays immutable. Roll-ups run from agent to squad, client and agency portfolio.
Features
What Investment & ROI Ledger includes
Ledger accounts per scope
Agent, squad, campaign, goal, client and agency, with currency and baseline reference.
Inference and pass-through capture
AI cost per trace span and messaging cost per message from provider webhooks.
Rent, ad spend and render costs
Prorated daily rent, agent-managed ad spend and avatar, video and voice renders.
Human supervision time
Approval and review timings times the tenant's hourly rate, always shown.
Reconciliation
Pass-through, spend and rent reconciled against provider invoices and ad accounts.
Live spend meter and caps
Daily and monthly caps per agent, squad and tenant with hard stops on spend Actions.
Outcome guardrails and graceful degradation
Low ROI pauses external Actions; caps route conversations to a human queue.
Outcome links and attribution
Won deals, payments and resolutions attributed under versioned tenant rules.
Frozen baseline
Baseline period fixed at contract start and immutable after.
Confounder flags
Seasonality, discounts, price changes, existing pipeline, data gaps and overrides.
Holdouts
Lift with confidence intervals; only holdout-validated returns labelled caused.
Monthly Agent Investment Statement
Eight sections with drill-down from every figure to evidence.
Recommendations and decisions
Continue, adjust, retrain, replace or stop, recorded when the owner acts.
Disputes and restatement
Dispute entries in the window; restated versions keep the original immutable.
Workflow
Capture to statement
- 1
Capture costs daily across rent, credits, pass-through, spend and supervision
- 2
Reconcile against invoices and ad accounts
- 3
Link outcomes and attribute them
- 4
Flag confounders and check data quality
- 5
Generate the statement and recommendation
- 6
Owner decides; disputes lead to restatement
AI
AI skills and Digital Employees
Cost-line classification
Maps unlabelled charges to cost lines with an "unclassified" bucket.
Reconciliation mismatch explanation
Ranks likely causes of variances with evidence.
Statement narrative
Writes prose around computed figures only.
Holdout design assistant
Proposes holdout size and duration, refusing to exclude legally required contacts.
Dispute evidence pack
Assembles traces, approvals, logs and records for a dispute.
Digital Employees that use this module
They act through the same governed Actions as people; risky steps wait for approval.
Frequently asked questions
Does AI calculate my ROI?
No. Figures come from deterministic computation. AI only writes the narrative, and a check blocks any number not in the computed statement.
What does "caused" mean on a statement?
Only returns validated by a holdout group are labelled caused. Everything else is labelled assisted.
Is supervision time counted?
Yes. Human approval and review time is costed at your hourly rate and always shown.
See Investment & ROI Ledger in your workflow
Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.