Knowledge Assistants
AI Knowledge Agent for Microsoft 365
Answers staff questions from your documents with citations, connected natively to Microsoft 365 so it reads and cites SharePoint, OneDrive and Teams…
Knowledge Assistants
AI Document Search
One question in plain language returns a synthesized, cited answer drawn from your whole document library.
Your team knows the answer is in a document somewhere, but finding it means opening ten files across three tools. AI Document Search reads your library and answers a plain question with one clear answer, citing the documents it came from. People only get answers drawn from files they are allowed to open. You choose the sources and the permission groups.
- One cited answer instead of a pile of files
- Answers respect each person's permission group
- Monthly refresh keeps answers current as documents change
$8,900
20–30 working days
Optional maintenance $445/mo
What it is
AI Document Search indexes your document library across up to 3 content sources and answers plain-language questions with one synthesized, cited answer, scoped to what each person is allowed to see. Every answer links back to its source document, and the index refreshes monthly as your files change.
Why businesses buy it
Searching by keyword works when you know the exact file name. Most of the time people do not, so they ask a colleague, interrupt an expert or give up and guess. That costs hours each week and spreads wrong answers. After this build, anyone can ask a question in plain words and get one answer with links to its sources, filtered by what they are allowed to see. The Scope Card sets the sources, document limits and permission groups in writing before you order, so there is no open-ended project to manage.
Problems this solves
Staff search several drives and still cannot find the answer
Document Search indexes your library across up to three content sources. Staff ask a plain question and get one answer drawn from all of them.
Search results show files, not the answer inside them
Each question gets one synthesised answer, with links back to the source documents. People can check the source in one click, without opening a dozen files to confirm it.
Some documents must stay hidden from some staff
Answers are scoped to what each person is allowed to see, across up to three permission groups. Confidential files stay out of answers for people without access.
A good fit if
You have a document library spread across up to 3 sources, such as Google Drive, SharePoint or Notion, that people ask questions about instead of searching for.
You want a single synthesized, cited answer to a question, not a list of results you have to open and read yourself.
Your library runs up to 2,000 documents, about 10,000 pages, and different teams should see different answers based on what they're allowed to read.
You're comfortable defining 30 sample questions and the answers you expect, so accuracy can be measured before launch.
Not the right fit if
Your questions really come from one defined document set, such as a single manual, not a broader library. See our AI Document Q&A product, built for exactly one set at a smaller price.
You want a ranked list of results to open yourself, not one synthesized answer. See our AI Internal Search product, which works like a search bar across your tools instead.
Your library isn't captured or sorted yet. See AI Document Intake and AI Document Classification, earlier pipeline stages that capture and organize documents before a library can be indexed well.
You need contract-specific terms and obligations extracted, not general questions answered. See AI Contract Intelligence, built for that document type alone.
You want staff to ask everyday questions of living documents, wikis and drives. See AI Knowledge Agent or AI Company Knowledge Hub.
How it works
- We connect up to 3 of your content sources, such as Google Drive, SharePoint, Notion or Confluence, and index up to 2,000 documents from them, about 10,000 pages.
- Every document is embedded so it can be matched by meaning, not just keyword, and tagged to the permission group it belongs to.
- Someone asks a question in plain language, on your web app, Slack or Microsoft Teams, the same way they'd ask a colleague.
- Document Search finds the passages across your whole library that answer it, and drafts one synthesized answer with a citation back to each source.
- Up to 3 permission groups keep answers scoped to what each person is already allowed to see, across every connected source, not just the one they asked in.
- A monthly refresh keeps the index current as documents change, and we test it against 60 benchmark questions with an accuracy report before launch.
What you provide within 72 hours
Your portal lists these the moment you order. Delivery starts the working day after everything is in and your access is checked.
- Access to the content sources this should cover, such as Google Drive, SharePoint, Notion or Confluence.
- Your user list and the permission groups answers should respect.
- 30 sample questions, with the answers you expect for each.
- Workspace admin access for sign-in on your chosen interface.
- Admin access and API keys for every connected system, shared by adding our team as a user or creating a key you control, never by email.
- Your business information (company basics, hours and brand voice), shared during intake.
- One decision-maker who can approve the permission groups before launch.
- An API key for your own AI model account (for example OpenAI or Google Gemini), created for this build with a spending limit.
The 72-hour rule. Send everything on your checklist, including access, within 72 hours of ordering so your build starts on time.
Timeline
Start
Intake confirmed, with content sources, permission groups and sample questions agreed.
Weeks 1–2
Documents indexed and embedded from every connected source.
Week 3
Cited Q&A, permission-group logic and interface connected.
Weeks 3–4
Tested against 60 benchmark questions, with an accuracy report delivered.
Week 4
Your review and two rounds of changes.
Go-live
Live, with a recorded handoff session, and support starts.
| Option | Delivery time | Price |
|---|---|---|
| Standard | 20–30 working days | $8,900 |
| Express | 15–25 working days | $10,000 |
| Rush | 10–20 working days | $11,150 |
What you receive at handover
- AI Document Search in one interface (a web app, Slack or Microsoft Teams) for up to 50 users
- An index of up to 2,000 documents from up to three content sources, with a monthly content refresh
- Cited answers with up to three permission groups
- A test report of 60 benchmark questions with an accuracy report
- An admin guide and a recorded 60-minute handoff session
- 30 days of post-launch support
Expected benefits and how to measure them
- Can cut the time staff spend hunting for information.
- Helps new starters find answers without asking colleagues.
- Keeps answers checkable because every one links to its source.
- Time to find an answer
- Time a set of common questions before launch and repeat them after.
- Benchmark accuracy
- Start from the accuracy report on 60 benchmark questions. Rerun them after each monthly refresh.
- Repeat questions to colleagues
- Ask a few teams to log questions they ask each other for a week, before and after launch.
Before you start: Write down twenty questions staff often ask and time how long each takes to answer today. List where the relevant documents are stored.
These are things to track, not promised results. Outcomes depend on your data, your team and your market.
Running costs
The build price is one-time. Running Document Search costs roughly $50 to $300 a month in your own accounts, covering AI model usage, the vector database and hosting, and it moves with document volume and how often your up to 50 users ask questions.
AI usage during the build and testing is billed to your own accounts too, so every provider relationship stays in your name.
Ask about running costsCustom requirements
If you need more than 3 content sources, more than 2,000 documents, or a second interface, add a note at checkout for a written Change Quote. Your order still runs on its normal timeline. Larger or bespoke projects use the Custom AI request form.
Request a custom versionKeep it running
Monthly maintenance covers monitoring, fixes, updates, a health report and change hours for things such as a new content source or updated permission groups. It starts after the 30 days of support and can be cancelled any month.
$445/mo (5% of the build price a month, includes 4 change hours)
Optional. Starts after the post-launch support period. Cancel any month.
How maintenance worksHow this compares
| Do it yourself with a SaaS tool | Hire a custom agency | Upmark Labs | |
|---|---|---|---|
| Price | An enterprise search tool charges per seat every month regardless of use | A custom search build is quoted after weeks of discovery | Ours is one fixed price, shown on this page before you buy |
| Answer format | A document search SaaS often returns a bare list of filenames | An agency build depends on what was scoped | Ours returns one synthesized answer with a citation to its source |
| Access control | A document search SaaS may index everything with no permission scoping | An agency build depends on what was scoped | Ours scopes every answer to up to 3 permission groups |
| Delivery | An enterprise search tool is instant to sign up but slow to tune well | An agency build usually takes several weeks | Ours has a fixed delivery time shown before you buy, with faster options |
Questions about this product
How much does AI Document Search cost?
The fixed price for each delivery speed is shown in the price and delivery panel on this page, so you see it before you order. It covers everything on the Scope Card: indexing up to 2,000 documents across three sources, up to three permission groups, the monthly content refresh and the benchmark accuracy report before launch.
How is AI Document Search different from AI Document Q&A?
Document Q&A, a smaller build, answers questions from one defined document set, such as a single manual, up to 100 documents. Document Search covers a broader library, up to 2,000 documents across up to 3 sources, with permission groups scoping answers by team. Choose Search when your questions span more than one document set.
Does AI Document Search give one answer, or a list of results to open myself?
One synthesized, cited answer, drawn from across your whole indexed library. If you'd rather get a ranked list of documents to open and judge yourself, see our AI Internal Search product, which works like a search bar instead of a question-answering assistant, across the same kinds of connected tools.
How much does AI Document Search cost to run each month?
Running costs are paid in your own accounts, not to us. Expect roughly $50 to $300 a month covering AI model usage, the vector database and hosting. The figure moves with how much you index and how often your users ask questions. There is no per-question fee from us, and monthly maintenance is optional.
Do AI Document Search answers respect who is allowed to see what?
Yes. Up to 3 permission groups are configured before launch, and every answer is filtered against them before it's shown, across all connected sources at once. A document someone can't open in its source system won't inform an answer they receive, even when its words match the question closely.
Is our document content kept confidential with AI Document Search?
Yes. Every credential and connected account is shared by adding our team as a user or creating a key you control during setup, never by email, and indexed content is used only to answer questions within your permission groups. Nothing is shared outside your connected sources, and the 30 sample questions can be based on redacted examples if needed.
How is AI Document Search different from AI Knowledge Agent?
The build sizes are the same, so choose by content. AI Document Search is for a stored archive of files, such as PDFs, scans and records, where access differs by team. AI Knowledge Agent answers everyday staff questions from living documents. AI Company Knowledge Hub brings wikis, drives and documents into one assistant.
Can I add a new content source to AI Document Search later?
Yes. A fourth source beyond your initial 3 is usually a Change Quote, turned around, since it needs its own indexing and permission mapping. Refreshing or re-indexing your existing 3 sources is included every month as part of the standard build, at no extra cost.
Stop searching five places for one answer.
Order AI Document Search and it answers questions across your document library, with citations, on a fixed delivery date. The scope above is what you get, in writing, before you pay.
Works well with
Knowledge Assistants
AI Knowledge Agent for Google Workspace
Answers staff questions from your documents with citations, connected natively to Google Workspace so it reads and cites Drive files, shared drives…
Knowledge Assistants
AI Knowledge Agent for Confluence
Answers staff questions from your documents with citations, connected natively to Confluence so it reads and cites spaces and pages your users…