AGENTIC ORCHESTRATION · ENTERPRISE DEPLOYMENT · SAN FRANCISCO

I let bots agents handle the boring details of the business.

The interesting ones still need a signature.
10 years from deterministic orchestration to agentic orchestration, deployed into enterprise operations and run in production.
MICROSOFT-PUBLISHED · THE FORRESTER WAVE Q1 2021 · UIPATH AI10 PRESS . FOR THE WORK
RUNNING NOW · HOME CARE
26M jobs. 99% straight through, for 5 years.

One control plane runs every operation: agents on the repeatable work, a person where a decision needs one, cost per outcome on the meter, an audit record of what ran.

CONTROL PLANE IDENTITY GUARDRAILS OBSERVABILITY COST PER OUTCOME AUDIT TRAIL AGENT PERSON WHERE decision = regulated → PERSON
99%straight through
$0.03per outcome
16 in 10,000reach a person
RUN · 5 OPERATIONS

Where the operations run.

5 operations in production. Every run traced, every exception routed to an owner, cost per outcome on the dashboard.

SUM runs
GROUP BY value_stream
AVG duration_sec
WHERE reached_person = 1 → PERSON
WHERE status = failed
ONE DAY ON THE HOME-CARE PLANE
One day on the home-care plane, by hour: runs started, runs that reached a person, runs that failed.
HourRunsReached a personFailed
00:00800
01:00100
02:00401
03:00700
04:00411
05:00400
06:00300
07:001000
08:001200
09:001311
10:00811
11:00810
12:00300
13:00710
14:00911
15:00610
16:00500
17:00320
18:00300
19:00500
20:00600
21:00300
22:00200
23:00800

The orchestration is the job.

26M JOBS, 99% STRAIGHT THROUGH

One home-care company's operations on one plane for 5 years. 16 in 10,000 reach a person.

660,000 FRAUD DECISIONS A MONTH

Assembled by an agentic system from 8 systems, every one signed by a person. 9:00 to 4:00 per case.

700,000 BILLS A MONTH

A $2.5B payments operation on one twin. 1 invoice in 100 reaches a person. $0.58 to $0.21 a bill.

HOW I WORK
Discover, redesign, decide the tooling, build, secure, observe, run.
Agents go on top of the mess, or the mess gets fixed first. Every enterprise chooses one. These 5 operations run in production because I redesigned the workflow first and built the governance in.
BUILD

Agents where the operating model is the starting point.

RUN

Agents and people on one plane, at fleet scale, with a cost per outcome.

GOVERN

Identity, guardrails, a human in the loop, observability and an audit trail on every action.

DISCOVER · 1
Discover
I walk into the operation and map the process as it runs.
70+ systems read in a 5-day discovery sprint.
DECIDE · 2
Redesign the operating model
Agents take the repeatable work and people keep the judgment. Every decision gets a named owner and a place in the workflow. A person sits only where a decision needs one.
100% of regulated decisions pass through a person.
DECIDE · 3
Decide the tooling
I run build versus buy as the buyer, on live volume, and deploy each tool where it wins.
Workato, Power Platform, Appian and Azure AI, each where it won.
BUILD · 4
Build the orchestration
An orchestrator runs specialized agents on shared state, with search and risk scoring. A person acts on the recommendation.
8 systems, one record per case.
GOVERN · 5
Identity and guardrails
Every agent has an identity and scoped access. Every action has a fail-closed path, and every regulated decision stops for a person.
Only a name change and a card change close themselves.
GOVERN · 6
Observe and route
The system traces every run, step by step. Every failure has a code and a route to the team that owns it.
1 error topic, 1 consumer, 108 routes.
RUN · 7
Run it on cost per outcome
I hire and direct the team and run it in production as a standing operation, with a named executive sponsor. Cost per outcome is on the dashboard, and the exception trains the system. Process intelligence first. Every step is measured and priced before an agent takes it.
$0.03 per outcome · 99% straight through · 9 acquisitions absorbed as configuration.
SECURE EVERY AGENT
IDENTITY·GUARDRAILS·HUMAN IN THE LOOP·OBSERVABILITY·EVALUATION·FINOPS·AUDIT TRAIL
DECISION RIGHTS

I draw the decision rights before any agent ships. That drawing is the redesign.

STRATEGIC INITIATIVE QUICK WIN DEPRIORITIZE BACKLOG 01 · Rate-increase letter release
ASSIST · QUICK WIN
A person reviews
The system builds the case. A person approves, rejects, or sends it back.
40% OF NOMINATED USE CASES CUT AT REVIEW · 5-DAY SPRINT
THE RECORD

What ran, where, and what it returned.

I ask where the check goes, what fails closed, who gets the ticket and with what evidence, and what an outcome costs.

US telecom carrier
Enterprise automation governance, intake and delivery frameworks. Invoicing errors down 85%, processing time down 90%.
2016 – 2018
Fortune 20 energy
70+ automations. 200,000 hours returned, $20M saved.
2018 – 2019
Top US telecom
Enterprise operating model, designed and stood up. 250 automations, $18M returned. One redesign alone, $988K a year.
2019 – 2021
The practice
Center of Excellence grown from 4 to 30+ across the US, Mexico and India. Discovery runs as a 5-day sprint reading 70+ systems.
2019 – present
Governed operations
26M jobs at 99% straight through over 5 years, 9 acquisitions absorbed as configuration, no rewrite since 2021.
2021 – present
EXPERIENCE MICROSOFT-PUBLISHED · FORRESTER 2021 · UIPATH AI10
Concentrix Grew the automation Center of Excellence from 4 to 30+. 2019 – PRESENT
UiPath Professional Services, Delivery Engineering. 2018 – 2019
ProKarma Automation Consultant. 2016 – 2018
Agents inherit the operating model they land on.
I build the gate, and I answer for it when it breaks.
THE PRACTICE
Enterprise readiness is agentic readiness.
Agentic OrchestrationProcess MiningRobotic Process AutomationDocument IntelligenceML OpsFoundation ModelsConversational AIEnterprise Search · RAGObservabilityIntegration · iPaaSData Warehouse
Headless · agents call the systems by API · a person only at the gates.
CONTACT