From solicitation to award, SquarePact helps government agencies draft, redline and check documents inside the most-commonly used contracting tool: Microsoft Word.
Co-Developed with Tampa International Airport's procurement department.
Safe by design. We only use Zero Data Retention AI.
Read our Privacy Guarantee →
Most Word AI tools hand you a closed assistant, which means their prompt, their logic and their idea of a good redline. SquarePact is the opposite. You decide what runs and you approve every change it proposes. Here's what procurement teams ask us.
SquarePact was spun out of Professor John Licato's Advancing Machine and Human Reasoning Lab at the University of South Florida, built on 20 years of research into AI that helps people reason better, not AI that tries to replace them. That focus shapes the product. We do not rewrite your judgment and we do not decide anything for you. What we do is get the document mechanics right, meaning the formatting drift, broken cross-references, stray styles and inconsistent defined terms that other AI tools quietly introduce or ignore. We also published the measurement rather than the claim. On 86 real Word documents every add-in we tested left defects behind, including ours at 12.8%, and the whole table is at /formattingBench/. More about who we are ↓
Much more. Formatting integrity is the foundation, and on top of it you get the review routines that do the work. Save the checks your office runs as sequences written in plain English, mark them Critical, Normal or Low, and run them on any document, with a report that states how much of the document it covered. Keep your clauses and boilerplate in a playbook you can search, tag and insert, with a conflict check against language already in the document. Set usage caps so a department knows what it is spending. Nothing is applied to your document until you click Apply, and what you approve goes in as a tracked change. It carries a solicitation from first draft through final review, all inside Word.
Most Word AI tools are a closed assistant, meaning their prompt, their logic, their idea of a good redline. SquarePact is the opposite. Bring the agent you already built and run it against the open document with full structure, not a chat window guessing at content it can't see. Or use the built-in SquarePact agent for the checks you don't want to build yourself. The choice is yours.
Totally fine. SquarePact gives you optional access to our own AI agent, built specifically for the documents a procurement office has to defend. Use ours or bring your own, it's entirely up to you.
Numbered lists reset. Styles disappear. Cross-references break. Anything that touches your document through a chat window is guessing at content it can't see. SquarePact edits at the XML level, where Word actually lives, so it never leaves Word's native format. That is why it damaged documents least of the four tools we measured, at 12.8%. It is not zero, which is exactly why nothing is applied until you approve it.
You should expect costs to go up, especially on long documents. So SquarePact does two things about it. You can set usage caps, so a department knows what it is spending before it spends it, and you can pick how hard the agent works on each task rather than paying top rate for a lookup. You can also bring your own API key and pay your provider directly. We have our own agent you can use, and we never force you to.
We designed SquarePact for you. Connect it to your own API keys and your own endpoints. You can route entirely to local or private models, so document content never leaves your infrastructure. Your data, your model, and an answer you can put in the file.
It doesn't. SquarePact augments them. Prefer Claude? Get an API key and use Claude inside Word through the SquarePact add-in. Prefer an advanced agent you wrote yourself? Use that. Whatever you want as the brain, SquarePact is the front-end UI for your document workflow. We also measured what each one does to a Word document. Across 86 real Word documents, SquarePact left defects in 12.8%, Claude for Word in 20.9%, Microsoft Copilot Premium in 31.4% and Grok for Word in 32.6%. The full table and the method are at /formattingBench/.
No problem. SquarePact isn't tied to OpenClaw or any one agent framework. Point it at your own API keys, or at your own local and private model endpoints, and it runs them against the open document from inside Word. Route to a local or private endpoint and document content never leaves your infrastructure. Either way you get the same formatting-safe editing, whether the model runs on your own machine or in the cloud.
Good. The gap is that those tools do not sit beside you in Word, and they do not preserve formatting when they edit a document you are about to publish. SquarePact gives it that ability: bring your existing AI into Word, or use our agent. In an enterprise or government environment it's even easier, we can set up SquarePact entirely within your existing Azure setup.
Not another platform. Just a panel inside the Word you already have. You don't open SquarePact; you open Word, same as every day, and it's sitting in the ribbon. It installs from Microsoft AppSource like any other Office add-in, then admins push it to users, groups, or the whole organization through the Microsoft 365 admin center.
Procurement offices choose SquarePact because it can be set up entirely within their existing Azure environment, and because there is nothing new for staff to learn in the week a solicitation is due.
SquarePact is unlike any tool I've used so far. It saves me a ton of time and gives me answers that I can trust, because it shows me exactly where it gets the answers from and links me right to that spot in the document.
SquarePact did a great job in evaluating my proposal, I wish I had it sooner! It provided detailed critiques and recommendations that were really helpful. It went beyond a surface level critique and provided concrete points linked to specific proposal criteria. I can't wait to use it for all of my future proposals!
SquarePact stands out from all the 'use AI for contracts' offerings, it provides professional-grade, hallucination-resistant, and trustworthy output.
SquarePact has caught subtle gaps in my contracts, around compensation, equity, and IP, that would otherwise have been vague, and let me raise those points with the other party using its suggested language. It does this without ever pulling me out of Word.
Actualization AI, Inc. is a Tampa, Florida company spun out of Professor John Licato's Advancing Machine and Human Reasoning Lab at the University of South Florida. Dr. Licato's research has been driven for 20 years by a simple motto:
AI should help people find and fix errors. It should not introduce more errors. It should not try to replace people.
Not AI that pretends to be a contracting officer. It's AI that fights the quirks of editors like Word, so you spend less time re-reading a document line by line and more time on the procurement itself.
Learn more about us →Recent notes from the team on practical AI adoption, document quality, procurement workflows, and the places small errors get expensive.
We're so confident we'll find an issue worth fixing, we'll give you a free month if we can't.
Take the challengePublished prices and usage caps you set, with no separate system to procure before you start. Use the pricing calculator to estimate a department's usage, or bring your own API keys and route to your own models.
See pricingThanks, your message is on its way. We've received your note and will get back to you within one business day.