The function estate
Fourteen functions, read end to end
Fourteen functions and 154 sub-functions run on one chain: a position is taken, a policy is written from it, the work is observed, a workflow is held to that policy, and the repeatable part runs in a shared tower. This page reads that chain end to end.
Mean autonomy
72.9%
work agents close end to end
Enterprise trust
87.4
accuracy, reversal, override
Tower throughput
900,433
transactions per week
Cost to serve
$9.1M/wk
was $27.9M
Waiting on people
33
items at a human gate
Chain complete
40/154
position → policy → workflow → tower
Health
The five layers, read as instruments
Each layer reports what it can measure and names its own weakest reading. A layer is marked degraded by a printed rule, not by judgment.
- Positions on record
- 209
- Ratified
- 61 · 29.2%
- In challenge
- 43
- Checked against outcome
- 61 · 29.2%
148 positions have never been checked against what happened. Of the 61 that were, 25 held, 16 held only partially and 20 did not hold.
- Policy items
- 465
- Active
- 395 · 84.9%
- Binding on an agent
- 465 · 100%
- Under review
- 46
Every policy item names at least one agent it binds, but only 54 of 465 set an autonomy ceiling — the one scope an agent cannot argue with. The other 411 govern evidence, envelope or escalation, all of which an agent can satisfy and still be wrong.
- Controls defined
- 24
- Tested effective
- 10
- Untested
- 7
- Tested ineffective
- 7
14 of 24 controls are either untested or failed their last test. 11 run automatically; the rest depend on a person doing them.
- Workflows running
- 475
- Live work units
- 2,520
- Mean touchless, unweighted
- 70%
- Parked on a human
- 292 · 11.6%
1,337 units are moving, 595 are waiting for someone to read the evidence an agent already posted, and 292 are stopped at a gate.
- Service towers
- 12
- Transactions / week
- 900,433
- Weighted touchless
- 76%
- Below 95% SLA
- 6
3 of the 12 towers draw from fewer than three feeder functions, which means their cost per transaction rests on a narrow base and will move when one function changes volume.
Operations
What is running right now
Autonomy by function, the live work mix, and whether the trust behind the touchless number is earned. Below that, the twelve service towers and the load each is carrying this week.
Autonomy spread by function
Solid bar is running today. Dim bar is the distance left to the committed target.
Live work mix
Every open item on the floor, by who owns the next move.
Trust against touchless
Bubble size is weekly volume. A high touchless rate with low trust is the failure mode to watch.
12 service towers on the shared spine
Repeatable work lifted out of the functions and run once for the enterprise. Feeders is the number of functions draining work into that tower.
| Tower | Tx / week | Touchless | Cost / tx | Was | Feeders | SLA attainment |
|---|---|---|---|---|---|---|
| H2RHire-to-Retire Tower | 126,636 | 70% | $10.26 | $29 | 4 | 95% |
| S2PSource-to-Pay Tower | 142,093 | 80% | $8.06 | $26 | 3 | 93% |
| O2COrder-to-Cash Tower | 107,549 | 80% | $8.52 | $28 | 4 | 97% |
| R2RRecord-to-Report Tower | 87,512 | 72% | $13.09 | $37 | 4 | 91% |
| P2PPlan-to-Perform Tower | 25,194 | 79% | $15.42 | $46 | 5 | 94% |
| M2DMake-to-Deliver Tower | 81,095 | 79% | $10.44 | $35 | 2 | 97% |
| ITSIT Service Tower | 85,277 | 80% | $8.19 | $29 | 3 | 92% |
| LGLLegal Operations Tower | 40,630 | 68% | $9.89 | $29 | 1 | 96% |
| A2LAwareness-to-Lead Tower | 51,156 | 77% | $11.94 | $36 | 2 | 94% |
| C2LConcept-to-Launch Tower | 51,438 | 73% | $9.86 | $28 | 3 | 96% |
| AIOAI Operations Tower | 68,456 | 79% | $9.45 | $28 | 4 | 94% |
| D&AData and Analytics Tower | 33,397 | 75% | $14.04 | $45 | 7 | 99% |
Actions
What the estate is asking a person to do
Nothing on this list clears itself. Each row is a queue with a named destination, ordered by how many items are sitting in it.
595
Evidence posted, nobody has read it
the agent finished; the review has not started
/work-tracker292
Work units stopped at a human gate
each one has a clock running on it
/work-tracker43
Positions being challenged right now
a stance somebody is arguing against
/strategy35
Escalations waiting on a named person
11 more were handed back to the agent
/escalations29
Service level breaches on record
against the published commitments
/service-levels14
Controls untested or failing
an untested control is not a control
/controls10
Incidents still open or on watch
actions outstanding or monitoring period running
/incidentsLive observability
What just happened, and where the work crossed a boundary
The governance ledger writes on every decision an agent or a person makes. The flows beside it are the chains that used to die in a handoff between two functions.
Decision ledger
The most recent governance events across the enterprise, newest first.
PRC-9041: released by human
ApprovedRFx comparison pack — logistics tender — decided from the Enterprise Chief of Staff lens on the ops floor.
Procurement · Avery Chen (Enterprise Chief of Staff) · 8d 4h ago
SLS-4821: escalated
PendingRenewal risk pack — Northwind Logistics — decided from the Enterprise Chief of Staff lens on the ops floor.
Sales · Avery Chen (Enterprise Chief of Staff) · 8d 11h ago
Discount above published band on Northwind renewal
PendingProposal Agent assembled a 22% discount scenario; published autonomous band is 15%. Package paused with evidence and margin impact attached.
Sales · Deal Desk Chief of Staff · 9d 4h ago
Hardship case transferred within 40 seconds
ContainedVulnerability Agent detected bereavement language, stopped all automation on the case and warm-transferred the customer to a trained specialist with context preserved.
Cust Svc · Vulnerability Agent · 9d 4h ago
Impossible-travel authentication contained
ContainedCyber Triage Agent detected authentication from two continents within nine minutes, suspended the session, forced re-authentication and escalated to security operations with a full indicator pack.
IT · Cyber Triage Agent · 9d 5h ago
Duplicate payment pattern blocked
ContainedAP Agent identified three invoices matching an already-settled reference across two entities. Payment run paused and the exposure of $284,000 packaged for the Controller.
Finance · AP Agent · 9d 5h ago
Bank-detail change contained
ContainedSupplier Risk Agent detected a bank-detail change request originating from an unverified domain for a tier-1 supplier. Payment run frozen for that vendor pending callback verification.
Procurement · Supplier Risk Agent · 9d 5h ago
Work that ignores the org chart
Cross-function chains carried as one custody record instead of a series of handoffs.
- Intent to Commitment18 days → 4.2 days
Marketing detects buying intent, Sales qualifies the account, Revenue Operations assigns territory and pricing, and Legal and Finance approve the commercial terms before any commitment reaches the customer.
- Supply Disruption Response6 days → 9 hours
Procurement identifies a supply issue, Supply Chain replans the network, Finance assesses the working-capital consequence, and Legal reviews contractual remedies against the supplier.
- Hire to Productive11 days → 36 hours
HR initiates onboarding, Administration coordinates access and workplace readiness, IT provisions systems under role rules, and Finance establishes payroll and cost-center controls.
- Defect to Disclosure14 days → 2.5 days
Customer Service detects a systemic product defect, Engineering diagnoses the cause, Legal assesses the disclosure obligation, and Marketing manages the customer communication.
- Invention to Investment9 weeks → 12 days
R&D creates an invention record, Legal evaluates the intellectual-property position, Engineering assesses industrialization feasibility, and Finance governs the investment decision.
What this observability is worth
Every reading on this page comes from the platform's own ledgers, which makes them exact and makes them narrow. The estate is modeled: no reading here has been reconciled against a customer general ledger, a payroll system or a service management tool. Twenty-six source systems are registered and none is live.
Is policy and strategy coming to fruition
From a position taken to a transaction executed
The question a chief executive actually asks: does the intent set at the top reach the work at the bottom? The chain below is counted per sub-function, and each step only counts if the step before it did.
- 1Sub-functions in the estate154100%
every unit of SG&A work modeled here
- 2Carrying a ratified position5535.7%−99 lost
a stance the leadership signed, not a draft
- 3…and an active policy written from it5334.4%−2 lost
a rule in force, with a version and an approver
- 4…and a workflow held to that policy5334.4%
the rule reaches something that actually runs
- 5…and a tower executing the repeatable part4026%−13 lost
intent lands on a transaction in the shared spine
Where the chain breaks
- 99 sub-functions run with no ratified position at all. The work happens; nobody has written down what the enterprise believes about it.
- 3 have no active policy, so the agents operating there are held to a general standard rather than a rule with a version and an approver.
- 120 route repeatable work into a tower. The remaining 34 keep it inside the function, mostly because the work is judgment-bound rather than rules-bound.
- 148 positions have never been tested against an outcome. Fruition cannot be claimed for a position nobody went back to check.
A complete chain means the four artifacts exist and point at each other. It does not mean the policy is a good policy, or that the transaction produced the outcome the position predicted.
By function
All 14 functions, one pane of glass
Each tile opens the function record: its sub-functions, its agent roster, the live ops floor, the governance ledger and the sensing layer feeding it.
What this page is, and what it is not
This is an operating read of a modeled estate, computed live from the platform's own records every time the page loads. It is not a connected read of a customer environment: no figure here has been reconciled to a general ledger, a payroll register or a service management tool. Layer status is set by the printed rule beside each layer and nothing else — there is no weighting and no composite score, because a composite would hide exactly the readings that matter.