Executive advisory

AI is an org-design problem disguised as a technology problem.

Most organizations now have an AI strategy. Very few have a plan for the workforce that has to deliver it. We work on the layer in between: what the work actually becomes, what the roles become, and who owns getting there.

The Translation Gap™

One question tends to change the temperature of the room.

"What does your workforce need to look like in three years, and who owns getting you there?"

Most leaders answer the first half in detail. They can describe the tools, the pilots, the roadmap, the vendor decisions. The second half is usually where it goes quiet.

That silence is not a failure of leadership. It is a gap in the org chart. The technology strategy has an owner. The workforce consequences of that strategy do not. So the tools go in, the work quietly changes underneath them, and the roles, the levels, and the pay bands keep describing an organization that no longer exists.

L1
ToolsOwner: CIO or CTO
What AI has been bought, built, and put into production. Almost always the most developed layer, and the only one most organizations can describe fluently.
L2
WorkOwner: usually contested
Which tasks are being absorbed, which are being augmented, and which are newly created. Not systems. Tasks. This is where fluency typically breaks.
L3
RolesOwner: nominally HR
What a job becomes once its task mix has changed. A role is a container for work. Change the work, leave the container, and the container stops describing anything real.
L4
ArchitectureOwner: typically nobody
Job families, levels, career paths, and pay bands. The scaffolding that decides who advances, who is paid what, and what a career looks like here. Nothing breaks loudly when this goes unattended. Your strongest people simply outgrow it and leave.
L5
Capability and accountabilityOwner: this is the question
Whether the people you will need in three years exist, and whether one person is accountable for closing the distance. A policy is not an owner. A steering committee is not an owner. An owner has a name.

An AI strategy converts into results only as deep as the lowest layer that has an owner. The organizations pulling ahead right now are not the ones with better tools. They are the ones who went below the first layer while everyone else was still buying.

How we work

A method, not a workshop.

Every engagement runs the same five moves. We call it FRAME. It scales from a three-week diagnostic to a full operating model redesign, and each move ends in a decision rather than a deck.

F

Find

The work as it actually runs today. Workflows, systems, roles, decision points, and the outcomes leadership is genuinely measured on.

R

Rank

Every workflow scored for value at stake, repeatability, complexity, data readiness, and risk. Most should be left alone, and saying so is part of the work.

A

Architect

The redesigned human and agent workflow, with decision rights named, data requirements specified, and success agreed before anything is built.

M

Mobilize

Adoption designed rather than hoped for. Oversight model, escalation paths, change leadership, and a named owner at every handoff.

E

Evaluate

Business outcomes, not usage statistics. Learning loops, and explicit criteria for what earns the right to scale and what does not.

Engagements

Diagnosis, then pilot, then scale.

Most teams start with the diagnostic. It is deliberately small, and it is designed so that the decision about what comes next is obvious by the end of it.

Start here

AI Opportunity and Readiness Diagnostic

Two to three weeks

Where your AI strategy stops converting, and what it will take to go deeper. Run against the five layers with your executive team.

  • Executive interviews and layer scoring
  • Task-level view of absorb, augment, create
  • Named gaps, named owners, sequenced first moves
The core engagement

Executive Workflow Redesign Sprint

Six to eight weeks

Take two or three high-value workflows and rebuild them for human and agent collaboration, with the role and governance consequences designed in rather than discovered later.

  • Workflow redesign with explicit decision rights
  • Role impact and job description rewrites
  • Oversight model and escalation paths
  • A live pilot with defined success metrics
Then scale

Responsible Scale-Up Roadmap

Eight to twelve weeks

The three-year workforce blueprint. What the organization needs to look like, and the sequence and governance that gets it there without breaking trust on the way.

  • Capability and competency mapping
  • Job architecture, levels, and compensation design
  • Governance and accountability model
  • Sequenced roadmap with a named owner per phase

Engagements are fixed scope and fixed fee. Pricing follows the shape of the problem rather than an hourly rate, and you will have a number before you commit to anything.

Also available

Executive workshopsGovernance reviewsBoard and leadership briefingsOperating model design sessionsFractional advisory
The practice

One discipline, three ways to reach it.

Human architecture is the design of how people, roles, and organizations are structured to do work that AI has changed. It is a discipline, not a project, which is why it is built to be taught as well as delivered.

The firm

Human Architecture Group

Advisory engagements with executive teams. Diagnosis, redesign, and scale-up, delivered directly.

The method

The Translation Gap™

The five-layer framework used in every engagement. Published openly, because the value is in the redesign, not in withholding the map.

The credential

Certified Human Architect™In development

A practitioner program for internal leaders and independent advisors who want to run the method inside their own organizations. Registration opening in due course.

About

I sit between the technology strategy and the people strategy, because almost nobody does.

For fifteen years I built and led people functions at an elite academic institution, which meant owning the unglamorous machinery most AI conversations never reach: job architecture, levels, compensation design, capability building, and the governance that keeps all of it defensible. I am a Certified Compensation Professional, which is a dull sentence describing the exact expertise that Layer 4 requires and that almost no AI advisor has.

Then I went and got fluent in the technology itself, because the advice being given to executives was being written by people who understood the models but not the org chart, or the org chart but not the models.

That combination is the whole offer. I am not here to explain what a large language model is, and I am not here to run a change management program around a decision someone else already made badly. I am here for the layer in between, where the work is redesigned and the organization has to actually become something different.

I wrote Know Your Worth, Get Your Worth: Salary Negotiation for Women, which was about the same underlying question this practice asks at organizational scale: what is this work actually worth, and who decides?

Olivia JarasFounder, Human Architecture Group
Certified Compensation ProfessionalCCP. The credential behind the job architecture and pay design work.
Harvard Data Science InitiativeAgentic AI, completed. AI Strategist, in progress.
MBA and MA[confirm institutions]
AuthorKnow Your Worth, Get Your Worth: Salary Negotiation for Women
Forbes Coaches CouncilContributing member
Fifteen years in people strategy[confirm how you want your current role referenced]
Start a conversation

Tell me how you would answer the question.

If the second half of it is harder to answer than the first, that is the conversation worth having. Thirty minutes, no pitch, and you will leave with a clearer read on where your own gap sits.