Records and document AI
AI records management and document AI that work from your agency’s own files
AI for public records requests, extraction, report drafting and FOIA support, and knowledge assistants built on governed SharePoint, with a person reviewing every output that matters.
Free · about one week · view-only · no obligation
View-only access your IT administrator approves and removes. Nothing in your systems changes.
On this page
Short answer
We build records and document AI for public agencies on Microsoft 365: extracting data from forms and scans, drafting recurring reports, supporting records and FOIA requests, and answering staff questions from your own policies and files, with the source cited. It runs on governed SharePoint libraries, so the AI only sees what each user may see. A first pilot takes 6 to 8 weeks and typically costs $25K to $50K.
Starting points
Where AI records management pays back first
The cost of waiting
Document work is where staff time goes, and where AI exposure starts
An assistant answers from what it can reach.
If a library is overshared, so is every answer drawn from it. That is why we govern the library before we build on it.
What it really is
Document AI is a governed library with an assistant on top, not a chatbot with a folder.
The model is the easy part. Accuracy, permissions and records rules come from how the documents are organized.
Your options
Chatbot, document AI product, or built on your SharePoint
The difference is where the documents live and whose permissions apply.
| Generic chatbot | Document AI product | Built on your SharePoint (us) | |
|---|---|---|---|
| Answers from | The public internet | Files uploaded to the vendor | Your own libraries, in your own Microsoft 365 |
| Respects permissions | No | A separate permission model | The same permissions as SharePoint |
| Cites sources | Rarely | Sometimes | Every answer links its source |
| Records retention | No | Separate | Your retention labels apply |
| Cost model | Per user | Annual subscription | Fixed-price build on licenses you own |
What is included
What we build
Data extraction
Read forms, scans and PDFs, pull the fields into SharePoint lists or Dataverse, and route exceptions to staff.
Report drafting
First drafts of recurring reports, memos and summaries from your data, for staff to edit and approve.
Records and FOIA support
Find, classify and assemble responsive records, flag sensitive content for redaction review, and log each step.
Knowledge assistants
Assistants that answer staff questions from policies, procedures and past cases, and cite the source document.
Governed libraries first
Metadata, retention and permissions in SharePoint, so the AI only sees what each user is allowed to see.
Human review
Approval steps and an audit trail on every output that leaves the agency or changes a record.
Our approach
From messy library to trusted answers
Govern the documents, build the assistant, test it on real requests, then run it.
- GovernOrganize the libraries the AI will use: metadata, permissions and retention.
- BuildExtraction, drafting or an assistant, inside your own Microsoft 365.
- TestRun it on real past requests and measure accuracy with your staff.
- RunGo live with review steps, then monitor and improve it.
Our promise: every answer cites its source document, and nothing leaves the agency without a person approving it.
Governance, security and data handling
Where your data goes, and who stays in control
- Inside your own Microsoft 365. We build in your Microsoft 365 and Azure environment, using the identity and permissions you already govern.
- Exposure fixed first. An AI assistant sees whatever its user can see, so oversharing and unlabeled sensitive content are dealt with before anything is switched on.
- Hosted in Microsoft Foundry on Azure. We use the model that fits the job: OpenAI, Anthropic or open-source models, deployed in Microsoft Foundry on Azure. Your data is never used to train them (see Microsoft’s data terms for Foundry). Anthropic operates its Claude models on Azure under its own commercial terms, and we show you which model and which terms apply before work starts.
- People decide. Approval rules, audit trails and human review wherever an output affects a resident, a record or a decision.
What it runs on
Built on SharePoint document management
Document AI is only as good as the library beneath it. Our SharePoint document management service builds the governed libraries this AI answers from.
Proof
Government work this builds on
Document AI is only as good as the document system beneath it. These are the systems we built.
The offer
One week, read-only, and a plan you keep
The AI readiness assessment shows what an AI assistant could find in your Microsoft 365 today, which uses are ready, and what each one needs before it goes live. The report is yours whether or not you build with us.
Exposure map
What an assistant could reach today: oversharing, permission sprawl and unlabeled sensitive data, ranked by risk.
Use-case shortlist
Three to five uses your agency could start with, scored for value, risk and readiness.
Plan and fixed price
The controls each use needs and a fixed-price estimate for the first build, ready for your procurement process.
- Free, with no obligation
- About one week
- View-only: nothing in your systems changes
- The report is yours to keep
Why free? It is how we scope a fixed price we can stand behind, and it shows you what to fix whether or not you hire us.
How it runs
Day 030-minute scoping call with the architect who would run the work.
Days 1–2Your administrator grants read-only consent. About an hour of your team’s time.
Days 2–4We review your Microsoft 365 and talk to the use-case owners you choose.
Day 5Findings session: exposure map, shortlist and fixed-price plan.
We reply within one business day.
Read-only Microsoft Graph permissions only, consented by your administrator and revoked at the end. No content leaves your Microsoft 365. NDA and data processing terms before any access.
Why agencies choose us
A Microsoft partner that builds and runs it

Transparent investment
The assessment is free. Here is what the work costs.
Every agency is different, and the assessment gives you the firm number. You deserve a real range first.
| Engagement | Typical timeline | Typical investment |
|---|---|---|
| Pilot: one document process | 6–8 weeks | $25K–$50K |
| Production assistant or extraction service | 2–4 months | $50K–$100K |
| Library clean-up before AI, where needed | 2–8 weeks | Fixed price, set in the assessment |
| Managed AI operations | Monthly | $2.7K–$10.5K a month |
Want a figure for your organization? Get an estimate in 2 minutes with our pricing calculator.
Model usage, SharePoint configuration and training are in the estimate, not a change order.
Questions buyers ask
Before you decide
Will the AI make things up?
It can, which is why every answer cites the document it came from, staff can check it in one click, and anything that leaves the agency goes through a person first. We measure accuracy on your real requests before go-live.
Are AI prompts and outputs public records?
They can be, depending on your state. We configure retention for prompts and outputs to match your records schedule and your counsel’s guidance.
Our documents are a mess. Do we have to fix that first?
Only the libraries the AI will use. The assessment shows which ones, and the clean-up is priced separately so you can see what it adds.
Can it handle scanned paper records?
Yes. OCR turns scans into searchable text. We built this for the Town of Bluffton’s scanned records.
Put your documents to work
The free assessment shows which libraries are ready for AI, what they would expose today, and what a first pilot would cost.

