Politics & Government · Government Contracting
Washington Changed the Name. What Changed in Your Proposal?
October 3, 2026 · Commentary
On September 29, 2026, President Donald Trump signed Executive Order 14434, Inaugurating the Era of Super Intelligence, directing executive-branch agencies, to the maximum extent permitted by law, to use “Super Intelligence” and “SI” instead of “Artificial Intelligence” and “AI” across official communications, websites, reports, policy documents, and other non-statutory materials. [1]
Helix Stratus Group agrees with the change.
“Artificial intelligence” was named in America roughly 70 years ago, when the technology was largely imagined around machines imitating or automating pieces of human intelligence. That description fits less comfortably today. Frontier systems increasingly extend what people can do across science, medicine, engineering, research, analysis, and other fields. The administration’s argument is that the language should reflect that trajectory. We think it should. [5]
But there is an important distinction for anyone selling technology to the federal government:
Washington changed the name. It did not change your past performance.
If your system needed a human analyst on September 28, it still needed one on September 30. If it had never operated autonomously before the executive order, the White House did not test it for you.
And putting “SI” on slide seven does not suddenly make the human optional.
That is where contractors can get themselves into trouble.
The new name is not a software update
Consider a sentence already sitting in a proposal:
“Our AI-enabled platform assists analysts by identifying relationships across large datasets and surfacing potentially relevant information for human review.”
The agency changes its terminology.
So does the contractor:
“Our SI-enabled platform assists analysts by identifying relationships across large datasets and surfacing potentially relevant information for human review.”
Perfectly reasonable.
The customer’s term changed. Nothing else did.
The platform still assists analysts. It still identifies relationships. It still surfaces information for a human to review.
Now imagine a competitor gets a little more enthusiastic:
“Our Super Intelligence platform autonomously discovers hidden threats, predicts emerging risks, and delivers decision-ready intelligence at machine speed.”
That sounds fantastic.
It also raises a few questions.
What does autonomously mean? Which threats? Predicted with what accuracy? Using what data? Who decided the output was “decision-ready”? Is a human still checking it? Has the system actually performed that way outside a demonstration?
Those questions mattered before September 29.
They still matter now.
The administration gave the technology a new name. It did not give contractors a cheat code around evidence.
Do not borrow capabilities from tomorrow
There is another reason to be careful.
In technical discussions, superintelligence has often meant something much more dramatic: intelligence exceeding human capability across most cognitive tasks. [6]
That is not what Executive Order 14434 suddenly declares every current SI system to be.
For purposes of implementing the order, and except where otherwise provided by law, “Super Intelligence” covers the technologies and systems included in the existing definition of artificial intelligence in 15 U.S.C. § 9401(3). At the same time, its purpose section explains why the administration believes the older term no longer captures where the technology is heading. [1] [2]
Those two ideas can exist together.
The administration is looking forward.
A proposal still has to describe what your product can do today.
Within 60 days of the September 29 order, the Assistant to the President for Science and Technology is required to propose legislative language for a federal SI definition, including whether it should modify, expand, or replace the existing statutory definition. That makes the next definition worth watching. It does not make it available for contractors to claim in advance. [1]
Use the administration’s language.
Do not spend a definition that has not been written yet.
Read the sentence like the buyer will
This is where government contracting makes the wording matter more than it might in an ordinary sales deck.
Capture wants the sentence to sound strong.
Engineering knows what the system actually does.
Legal wants to know what everyone just promised.
The government evaluator gets the final sentence.
And in a federal competition, that sentence is not merely marketing. Under FAR 15.305, competitive proposals are assessed using the factors and subfactors specified in the solicitation. The government evaluates whether the proposed approach and the company behind it can actually perform the work described in the solicitation. Claims that create unsupported risk can affect that evaluation. [3]
So before describing any consequential SI capability, separate four things:
What does the customer call it?
What does the system actually do?
What evidence proves it?
What still requires a human, more testing, integration, or customer data?
That does not make the proposal weaker.
It makes the proposal easier to trust.
Before rewriting an entire proposal library, start with the claim that would hurt the most if an evaluator took every word literally.
| Review item | What to record |
|---|---|
| Customer term | The terminology used by the agency or solicitation |
| Promised task | Exactly what the system will do |
| Evidence | Test, deployment, past performance, data, or benchmark |
| Evidence date | When that proof was generated |
| Limitations | Conditions, dependencies, uncertainty, and human review |
| Approved wording | The language the company is willing to defend |
| Accountable reviewer | Who verified the claim |
A defensible version might look like this:
“Using the agency’s SI terminology, our system applies automated entity resolution and model-assisted analysis to customer-authorized datasets to surface previously unrecognized relationships for analyst review. The system identifies candidate relationships; trained personnel determine their significance.”
The sentence still sells the capability.
It simply tells the buyer where the machine stops and the human begins.
That matters.
Mirror the customer. Do not rewrite history.
There is also a simpler business-development lesson here:
Speak the customer’s language.
If an agency has moved to “SI,” use SI. The order directs agency terminology; it does not, by itself, impose a blanket requirement on contractors to rename their products or rewrite existing proposals.
The order is unusually direct about executive-branch terminology, so continuing to lead with the old language can make a contractor look behind the customer rather than aligned with it.
But if the solicitation sitting in front of you still says “AI,” answer the solicitation sitting in front of you.
And do not go back through five years of past performance changing every historical reference. The order specifically says previously issued contracts, grants, regulations, presidential actions, and other historical documents do not have to be altered. [1]
NIST is already showing what that looks like in practice. Its current program material uses SI terminology while preserving earlier work such as the TEVV-Athlon Framework for Evaluating AI Systems and AI Evaluation Scenarios under their historical names. [4]
That is probably the best model for contractors too.
Change the vocabulary where the customer changed it.
Preserve the history where it matters.
And review the claim only when changing the language could change what a reasonable reader thinks the system can actually do.
Close the gap before the evaluator has to
At HSG, we use Intelligence Gap™ to describe the distance between the information in front of a decision-maker and what that person can responsibly conclude from it.
A single capability sentence can create that gap.
The engineer knows the limitations.
The developer knows what happened during testing.
The capture team knows what the customer wants to hear.
The evaluator sees 28 words in a proposal.
HSG’s source-grounded review approach connects those 28 words back to the evidence behind them: what the technology has demonstrated, what it has not, where a human remains responsible, and what the customer can fairly conclude.
The goal is not to make powerful technology sound less powerful.
It is to make sure the words are as advanced—and as defensible—as the technology they describe.
Washington has given contractors a new language for the next era of American technology.
Use it.
Just make sure the evidence can keep up.
Request an HSG Decision Defensibility Briefing. We will take one consequential SI capability claim, trace it to the evidence behind it, and identify the Intelligence Gap™ before the proposal asks the government to cross it.
Request a Decision Defensibility Briefing