Four Dollars of People for Every Dollar of Technology: Staffing the People Side of an AI Budget

Last updated: October 25, 2026

By Robert Ardell, Co-Founder and Strategic Advisor, KORE1

Bain finds that for every dollar companies spend on technology, four go to people and process, and on an AI program that people spend is data engineering, integration, and governance work. Those seats are contract-shaped. A mid-level contract data engineer bills $98 to $122 an hour nationally, about $57,200 for three months (KORE1 Contract Rate Index, October 2026).

Somebody in your company approved a line item for AI tools this year, and it covers licenses, a cloud commitment, and maybe a model vendor with a very confident account executive who closed the deal in December. That line is easy to find. It has a purchase order.

The other line is the one this post is about. It has no purchase order, because it is made of people, and it is usually four times bigger than the tools line that got all the attention in the steering committee deck.

One thing up front. We sell the hourly seats described below, so read the argument with that in mind. I will tell you which seats should stay on your payroll, because a few of them should. The rest of the seats are the kind of work that starts, ends, and is better bought by the hour through contract staffing.

Why Is the People Side of an AI Budget Bigger Than the Tools?

The people side of an AI budget is the spend on humans instead of licenses and compute. Data engineers who prepare the sources. Integration engineers who connect a model to the systems of record. Analysts who redesign the process around it. Governance staff who decide who may use what, and write it down.

Bain put a ratio on it. In its Technology Report 2026, published September 29, the firm wrote that companies it works with “have found that for every dollar spent on technology, four are spent on people and process.”

Technology. Not AI technology. I keep that distinction on purpose, because the stat is already circulating as “four dollars on people for every dollar on AI,” and Bain did not say that. Bain also described its own client base, not a census of every company. Treat four-to-one as a planning rule of thumb from a good source. It is not a law of nature.

Still, run the arithmetic on a budget you recognize. Say the tools line is $250,000. Four-to-one puts $1,000,000 of people and process behind it, and “people” there includes your own employees, not just contractors. A five-seat contract crew working six months, which I price below, comes to $488,000. That is roughly half of the people side of a $250,000 tooling line, bought by the hour and gone when the work is done.

Has AI Moved Productivity Yet?

A little. Which is the point.

NBER Working Paper 34836, “Firm Data on AI,” surveyed nearly 9,000 CFOs, CEOs, and senior executives across seven countries, and the current version was revised in October 2026. Executives in that survey report that AI adoption raised annual labor productivity growth by 0.2% over the last three years. They expect 0.6% a year over the next three, and on jobs the authors say executives report essentially no net effect on firm employment so far, which fits a technology that is still being absorbed.

Read those numbers next to Bain’s ratio and the story gets less mysterious, because the tools are bought on day one and the gain waits on everything else. The gain arrives when somebody cleans the data the tool reads, connects it to the order system, rewrites the approval step it changes, and writes the rule for what it may not touch. Bain titled its report “Absorption Is the New Advantage,” and absorption is a staffing problem before it is a technology problem.

I would not oversell 0.2%. It is what executives reported, in a survey, about a technology three years into mass adoption. But it explains why the CFO who funded a pilot in January is asking in October where the return went.

Which AI Program Seats Are Contract-Shaped?

A seat is contract-shaped when it has an end date you can name on day one. Pipeline builds end. Integrations ship. Process maps get approved. Policies get written, adopted, and handed over. Most of an AI program’s people work looks like that.

Data engineers are the clearest case. Somebody has to land the sources, fix the keys that do not match between the CRM and the billing system, build the pipeline, and document it so your own team can run it. That is a quarter or two of work, then a handoff. The data scientist and data engineer staffing page covers how we fill them, and the data engineering staff augmentation page covers the augmentation model when you already have a lead and need hands. If your plan is to hire instead, the cost to hire a data engineer breakdown has the permanent-seat math.

Integration engineers and systems consultants connect the model to the places work actually happens, which means NetSuite, Salesforce, Workday, the ticketing tool nobody loves, and whatever homegrown system holds the numbers finance actually trusts. The project ends when the connection is live and tested.

Business systems analysts translate, sitting between the people who own the process and the people building the pipeline, and they write down what the process really does, which is rarely what the org chart says it does.

AI and data governance is the one I would split in two. Standing up the program is contract work. The inventory of models and data sources, the use policy, the access controls, the review checklist. A contractor with audit or security-compliance experience can build all of that in a quarter. Owning it afterward, answering to the board and the regulator when something goes sideways, belongs to an employee. Hire that person. Contract the build.

Project managers and a security reviewer round out the usual crew. Neither is exotic. Both are the first seats cut from an AI budget and the first ones missed.

Devin Hornick, a partner here, said something on our October 16 call that explains why these seats get bought the way they do. “With AI, I’m seeing that so many more managers are involved… HR kind of ran things.” More hiring managers own AI outcomes directly now, and the person who owns an outcome tends to want a named seat on a clear timeline, not a headcount request that takes two quarters.

What Does Each Seat Cost per Hour and per Quarter?

Less than the pilot that stalls without it. Here is every seat priced at the national cell in the KORE1 Contract Rate Index.

Source: KORE1 Contract Rate Index 2026-Q4, effective October 16, 2026, United States national cells. These are modeled from BLS OEWS May 2025 wages (released May 15, 2026) at a 24% employer burden. Ranges are the 25th to 75th percentile bill rate. Three months is 13 weeks at 40 hours a week at the median, and six months is 26 weeks. Your metro will move these numbers.

Price your own AI program seats. Get a bill rate quote and send us the roles, the levels, your metro, the start date, and how long you need them.

The index has no separate row for AI governance. In practice I would price it from the business systems analyst and security cells, depending on whether the work leans toward process or controls. That is my judgment, not an index cell, so treat it as a starting point for a conversation.

Build a crew and the total is easier to hold in your head. Two mid-level data engineers, one mid-level integration consultant, one business systems analyst, and one project manager come to $244,000 for three months and $488,000 for six. Five seats. That is a real AI program team, and the number is smaller than most of the tooling contracts it supports.

Need a faster way to run your own version? The contract staffing cost calculator does the same math by role, level, and city, and the 2026 Contract Rate Card lists the full set of cells.

What Breaks If the Project Does Not Finish?

Picture a composite, built from several intake conversations rather than one client. A mid-sized distributor renews its model license in October. The data engineer who was cleaning up customer and order records had a contract that ended in September, because someone in finance called it “done” when the first load ran. By November the pilot has a license, a champion, and nothing reliable to read. The renewal was automatic. The finish was not.

No one voted to stall it. The tool had a purchase order and the people did not, so the tool survived the budget review.

Devin framed the question in an email to our team this month. “What technology initiatives are already funded that you can’t push into 2027?” He added that companies “may not approve three or four permanent hires, but they may absolutely approve the resources or project team necessary to complete a $300K or $500K initiative that has already been funded.” That second sentence describes the budget most AI programs actually have. Money for a finish, no headcount to do it.

We wrote a longer piece on what breaks when an IT project does not finish, with a four-question test for ranking funded work. It applies to AI programs without changes.

What Is Inside a Contract Bill Rate?

The number on the quote is not the number the contractor takes home, and the gap is not all profit.

Start with a contractor paid $100 an hour. Add the employer burden, which we model at 24%. Payroll taxes, workers’ compensation, required insurance, and in California, mandatory paid sick leave. That is $124 an hour before anyone has recruited, run payroll, or carried liability coverage, and a bill rate for that contractor lands around $150 to $160 an hour. Devin’s summary was short. “First off, it’s not a markup. Secondly, there’s a margin in there. And thirdly, there’s a bunch of other costs that you’re not thinking of.”

Practical advice. When a firm quotes you a percentage, ask what the percentage sits on, the pay rate or the loaded cost. The two produce very different invoices, and burden is often left out of the quoted figure. Ask for the all-in hourly bill rate and compare those. We walk through the method in how to compare staffing agency quotes.

How Do You Staff These Seats Hourly Without Fixed-Fee Scope?

You hire the person, not the deliverable. Hourly, you direct the work, and the engagement grows or shrinks as the program does.

Devin on the call: “Hourly, which we prefer. It’s really more of a staffing model to do the hourly billing and you manage the person.” Handing an outside team an outcome to own is a consulting engagement. Staffing supplies the person, and the direction stays with you.

So scope the seat and not the whole program. Write down the three things you want done by week 2, week 6, and week 12, name the single person who signs off, and start with a three-month term that you can extend once the first deliverable lands. Our post on the 90-day deliverable list for a data contractor has the template.

Then plan for a longer run than you think. Across 1,887 ended KORE1 contract assignments from 2019 to 2026, the median ran 116 days, and IT assignments ran 136 days at the median (n=410). If an AI program needs a pipeline built and handed over, fund four to five months, not six weeks. And if one of these contractors turns out to be the person you want on your payroll, they become eligible to convert at no fee after 1,040 hours.

Common Questions From Budget Owners

So how much of an AI budget actually goes to people?

About four dollars on people and process for every dollar on technology, according to Bain’s Technology Report 2026, which describes companies Bain works with and covers technology generally, not AI alone.

Use it as a planning ratio and not a forecast, since a $250,000 tooling line points to roughly $1,000,000 of people and process, including your own employees, and the contract share depends on how much of that work has an end date.

What does an AI implementation contractor cost per day?

$784 to $976 for a mid-level data engineer, at the KORE1 Contract Rate Index national range of $98 to $122 an hour over an eight-hour day, with a median near $880.

Senior data engineers run $118 to $140 an hour, which is $944 to $1,120 a day. Remote work does not have its own cell in the index, so use the national figure as a floor and let the metro move it. Days are billed hourly, so a six-hour day is a six-hour invoice.

Can AI governance really be staffed on contract?

The build can. A contractor with audit or security-compliance experience can stand up the model inventory, use policy, and access controls in about a quarter.

Ownership cannot, because somebody has to answer to the board and the regulator for the policy long after the contractor has left, and that person should be an employee. Contract the build. Hire the owner.

Who manages the contractor on an hourly AI seat?

You do. On an hourly seat the contractor works under your direction, and KORE1 handles recruiting, payroll, and the employment side.

That suits AI work, where the plan changes as the program learns. Billing is time and materials, so the seat grows or shrinks with the work. If you would rather hand an outcome to an outside team and not manage anyone, that is a consulting engagement, and a different purchase.

Is a contractor cheaper than hiring the same seat?

Not per hour, and that is not the point. A contractor is cheaper when the work has an end date, because you stop paying when it ends.

Direct hire at KORE1 is 20% of annual base salary with a 60-day replacement guarantee. It makes sense when the seat outlives the project. For a pipeline build or a governance stand-up, the seat does not.

Fund the People Line Before the Tools Renew

Before the next license renewal, list the people work the tools depend on, meaning data prep, integration, process change, and governance, then put a name and a length on each seat and price it from the table. If a seat ends, contract it. If it stays, hire for it.

For the permanent side of that plan, our guide to workforce planning for AI transformation covers headcount models. For the contract side, AI staff augmentation is how we fill these seats.

When the list is ready, talk to a recruiter about staffing the people side of your AI program. Tell us the roles, the metro, and the date the work has to be done by. We will tell you honestly what it costs and whether we are the right fit.