OpenAI Just Built a Consulting Firm. What In-House IT Teams Should Do About It.

On May 11, 2026, OpenAI announced something that had nothing to do with a new model, a benchmark, or a pricing update. They announced a consulting company.

The OpenAI Deployment Company — which the industry is already calling DeployCo — is a $4 billion standalone business unit backed by TPG, Bain Capital, and about a dozen other PE and growth-equity firms. Capgemini is an early investor, which tells you something about where the large system integrator market thinks this is heading. The company acquired Tomoro on day one, bringing around 150 Forward Deployed Engineers into the fold immediately. The job of those engineers is straightforward: embed inside client organizations, map workflows, and redesign them around AI.

This is a different story from the one about AI software resolving 90% of Level-1 helpdesk tickets. That story is about software replacing tasks. This one is about a well-funded organization that intends to do the work that in-house IT ops and automation teams currently do. Worth thinking about clearly if that's you.

What Forward Deployed Engineers Actually Do

The model isn't new. Palantir built its entire enterprise business around FDEs — engineers who spent months at a time inside government agencies and large companies, making the software actually run. OpenAI is borrowing that playbook wholesale.

In practice: an FDE comes in, maps the current state of your workflows, identifies where AI can automate or augment, builds integrations, and leaves a running system behind. Sometimes they stay for the full deployment. The pitch is that instead of buying software and trying to implement it internally, you get the software and the engineers who make it work, together.

For organizations with no in-house AI capability, that's a real offer. The question is how wide the category of "no in-house AI capability" turns out to be.

Who They're Actually Competing With

DeployCo's first-wave clients will most likely be large enterprises that currently spend heavily on system integrators — Accenture, IBM, Deloitte — but want to move faster than those firms typically allow. The economics of a $4 billion valuation only work if you're pricing against a meaningful share of the global IT services market, which is well north of $1 trillion per year.

But smaller organizations are clearly on the list too. For a funded company that runs on five SaaS tools and has one IT generalist, the proposition is direct: pay OpenAI to send someone in, automate your support queue and your onboarding process, and get back to building the product. That proposition lands squarely in the territory where in-house IT ops specialists operate.

I'm not framing this as a threat. I'm framing it as a signal about what "IT operations" needs to mean at organizations below the Fortune 500 tier. An FDE who can ship a working AI workflow in two weeks is a credible alternative to an IT person whose main offering is maintaining the MDM and keeping the CRM clean. They are not a credible alternative to someone who does what FDEs do and also has three years of organizational context.

The people who end up fine are the ones who offer something a well-resourced generalist can't replicate in the first six months — and the name of that thing is institutional knowledge.

The Gap That Doesn't Close Without You

Here's what no FDE has on day one: the history of the place.

I've built automation for a US top company I've worked with for several years. The ticket routing logic I wrote a while back has a specific branch that handles a specific edge case from a specific team — because of something that happened that nobody ever wrote down. The offboarding workflow I maintain has different handling for contractors in two countries because I was the one who found out, through a mistake, that the standard process breaks for them.

None of that context is in any documentation. It exists in the people who built the systems and stayed.

An FDE arrives with API access and a clean slate. In month three, when the automation they built breaks because it doesn't account for an edge case that was never in any spec, someone has to explain what's actually going on. That's the person who's been running things. Institutional context is not a soft skill. It's the difference between a workflow that runs and a workflow that runs reliably, across the messy exceptions that real organizations accumulate over time.

What to Build Now While You Still Have the Advantage

The part of IT ops work that is genuinely at risk from a well-resourced consulting arm is the reactive, generalist part: tickets, password resets, MDM, basic admin. That work was already being compressed by AI tooling. FDEs accelerate the compression.

What stays valuable:

• Deep workflow ownership. Not just the ability to configure a workflow, but the ability to explain why it works the way it does and what breaks if you change it. Pick two or three workflows that run through your hands and become the authoritative expert on every edge case. Document them not out of fear of being replaced, but because documented workflows compound — they become the institutional knowledge FDEs will need to consult you for.
• API-level tool fluency. I work daily with n8n, the OpenAI and Anthropic APIs, Pipedrive, HubSpot, and Help Scout. That combination lets me build context-sensitive automation that a cold FDE cannot ship on week two. If you're using AI products but not building with AI APIs, that is the meaningful skill gap to close this year. The technical barrier is genuinely low — the bigger barrier is deciding to start.
• A visible record of results. Every workflow I've redesigned, every hour of manual work I've removed — those are measurable, and I've made them visible internally. FDEs produce that kind of evidence for client organizations as part of their deliverable. In-house IT ops people should be doing the same. "This automation reduced onboarding time by roughly a third" is a different kind of defensible position than "I keep the lights on."

The Honest Version of the Threat

DeployCo is not going to eliminate in-house IT roles. But it is a credible signal that the consulting and implementation layer of enterprise AI will be contested by organizations with substantially more resources than most IT departments. The people who end up fine are the ones who have built something an FDE team can't replicate in the first quarter — organizational context, deep workflow ownership, and the standing to hold an external team accountable on the things they've missed.

Every enterprise AI implementation that succeeds needs that counterpart on the inside. Whether you're that counterpart by the time DeployCo's teams start arriving, or whether an FDE has to improvise the role themselves, is a decision you have a reasonable runway to make.

If you're working through what this means for your team or your own role, get in touch — it's worth thinking through with someone who's operating in the same landscape.