Knowledge Assistants
AI Chatbot Accuracy Audit
We test your live AI chatbot with 150 real questions and 30 adversarial prompts and show where it is wrong, with 15…
Knowledge Assistants
AI Company Memory
Decisions, meeting notes and project history become searchable memory, so 'why did we choose this' has a real answer.
Six months after a decision, nobody remembers why it was made or who agreed to it, and teams end up having the same debate again. AI Company Memory turns your meeting notes, decision threads and project retrospectives into a searchable record. Anyone can ask why a call was made and get the answer with its source, and the record keeps growing every month.
- Answers why a past decision was made, with the source
- New hires catch up without re-asking old questions
- The record grows monthly as your team decides more
$12,500
20–30 working days
Optional maintenance $625/mo
What it is
AI Company Memory is a cited Q&A assistant built on your organization's own generated record, meeting notes, decision threads and project retrospectives, rather than static reference documents, so anyone can ask why a past call was made and get the answer with its source. It refreshes monthly.
Why businesses buy it
Lost decisions are expensive. Teams reopen settled debates, repeat failed experiments and onboard new people by word of mouth, all because the reasoning sits in a meeting note or thread nobody can find. AI Company Memory makes that history answerable, with the original note cited, so decisions stick and lessons get reused instead of relearned. Because it refreshes monthly, it gets more useful the longer you run it. The Scope Card fixes the sources, users, permission groups and testing before you order, so you know exactly what gets indexed.
Problems this solves
Nobody can remember why a past decision was made
The assistant reads your meeting notes, decision threads and project retrospectives from up to three sources. Anyone can ask why a call was made and get the answer with its source.
Context leaves the business when long-serving staff move on
Decisions captured in notes and threads stay searchable after the people involved have gone. Each answer cites the original record, so the reasoning survives team changes.
Teams repeat mistakes an earlier retrospective already flagged
Before similar work starts, a team can ask what went wrong last time. Answers draw on cited retrospectives, and a monthly refresh adds new records as they are written.
Some records should only be visible to certain people
Up to three permission groups decide who sees answers drawn from which records. Leadership notes, for example, can stay with the people already allowed to read them.
A good fit if
Your team makes decisions in meetings, Slack threads and project retros, and six months later nobody remembers why, or who agreed to what.
You already write things down, in meeting notes, decision logs or retro documents, even if nobody reads them again afterward.
New hires and other teams keep asking questions that were already answered once, in a thread or a meeting nobody can find again.
You want the record itself to grow every month, not a fixed set of files you upload once and forget.
Not the right fit if
You want answers from static reference material, such as product specs or a manual, not your own decision history. See AI Document Q&A.
You want to search your existing tools yourself rather than get an answer synthesized from them. See AI Internal Search.
Your decision record does not exist yet in any written form. See AI Knowledge Base Builder, which structures scattered material first.
You need company-wide policy and department documents too, not just decisions. See AI Company Knowledge Hub.
How it works
- We connect up to 3 sources where your decisions already live, such as a meeting-notes drive, a Slack decisions channel or a project-retro wiki.
- Up to 2,000 of those documents, meeting notes, decision write-ups and retros, are indexed by meaning and tagged to the team or project they belong to.
- Someone asks why a past call was made, such as why a vendor was chosen or a feature was dropped, in plain language.
- Company Memory retrieves the meeting note, thread or retro that actually recorded the decision, and answers with a citation back to it, not a reconstruction.
- Up to 3 permission groups keep sensitive decisions, such as compensation or legal matters, visible only to the group that should see them.
- A monthly refresh pulls in new notes and threads as your team keeps deciding things, so the memory keeps growing instead of going stale.
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 where your decisions already live, such as a meeting-notes drive, a Slack channel or a project-retro wiki.
- Your user list and the permission groups sensitive decisions should be restricted to.
- 30 sample questions about past decisions, with the answers you expect Company Memory to find.
- 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 your decision sources, permission groups and sample questions agreed.
Weeks 1–2
Meeting notes, decision threads and retros ingested and indexed from your connected sources.
Week 3
Retrieval, citation and permission-group logic built and connected to your chosen interface.
Weeks 3–4
Tested against 60 benchmark questions about past decisions, with an accuracy report delivered.
Week 4
Your review and two rounds of changes.
Go-live
Live, handover report sent, and 30 days of support starts.
| Option | Delivery time | Price |
|---|---|---|
| Standard | 20–30 working days | $12,500 |
| Express | 15–25 working days | $14,050 |
| Rush | 10–20 working days | $15,600 |
What you receive at handover
- The Company Memory assistant in one interface (web app, Slack or Teams) for up to 50 users
- Connections to up to three content sources, covering up to 2,000 documents
- Prompt and guardrail configuration with up to three permission groups and a monthly refresh
- Test report with 60 benchmark questions and an accuracy report
- Admin guide and written handover report
- 30 days of post-launch support
Expected benefits and how to measure them
- Can shorten the time spent piecing together why something was decided.
- Helps new leads get up to speed on a project's history.
- Makes it easier to avoid reopening debates that were already settled.
- Time to find decision context
- Pick five past decisions and time how long it takes to find the reasoning, once before launch and again after.
- History questions to senior staff
- Ask senior staff to tally questions about past decisions for two weeks before launch and two weeks after.
- Answer accuracy over time
- After a monthly refresh, re-run part of the benchmark and compare the result with the launch accuracy report.
Before you start: Note where meeting notes, decisions and retrospectives live today. Time how long it takes to answer three real questions about past decisions.
These are things to track, not promised results. Outcomes depend on your data, your team and your market.
Running costs
You pay for the build once. Running Company Memory costs roughly $50 to $300 a month in your own accounts, covering AI model usage, the vector database and hosting, and it moves with how much new decision history gets added each month.
AI usage during the build and testing is billed to your own accounts too, so the AI model and vector database relationships stay in your name.
Ask about running costsCustom requirements
If you need more than 3 connected 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 go through the Custom AI request form.
Request a custom versionKeep it running
Monthly maintenance covers monitoring, fixes, model and platform updates, a monthly health report and monthly hours for changes, covering a new source or extra permission groups. It starts after the 30 days of support and can be cancelled any month.
$625/mo (5% of the build price a month, includes 6 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 | A meeting-notes or wiki tool charges per seat every month, decisions or not | A custom RAG build is quoted after weeks of discovery | Ours is one fixed price, shown before you buy |
| Freshness | A wiki tool only holds what someone remembers to write and tag | An agency build depends on what was scoped | Ours refreshes monthly as new decisions and notes are added |
| Traceability | A wiki search returns a document, not a specific decision inside it | An agency build depends on what was scoped | Ours cites the exact meeting note or thread behind an answer |
| Delivery | A wiki tool is instant to sign up but holds nothing useful yet | 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 is AI Company Memory different from AI Company Knowledge Hub?
AI Company Knowledge Hub spans static company documents across departments, such as policies and specs. Company Memory is built on your own generated record instead, meeting notes, decision threads and retros, and grows every month as your team keeps deciding things. Choose Memory when the goal is why a decision was made, not what a manual says.
What counts as a source for AI Company Memory?
Anywhere your team already writes decisions down: a meeting-notes drive, a Slack channel dedicated to decisions, or a project-retro wiki. Up to 3 sources and 2,000 documents are indexed, tagged by team and project. If your team does not write decisions down anywhere yet, start with AI Knowledge Base Builder instead.
Does AI Company Memory update itself as new decisions happen?
Yes. A monthly content refresh pulls in newly added meeting notes, decision threads and retros from your connected sources, so the memory keeps growing rather than freezing at launch. Larger changes, such as adding a fourth source, go through the monthly change hours in maintenance or a Change Quote.
How much does AI Company Memory cost to run each month?
Roughly $50 to $300 a month in your own accounts, covering AI model usage, the vector database and hosting. It moves with how much decision history gets indexed and how often your up to 50 users ask questions. The build itself is a one-time payment.
Can sensitive decisions, like compensation, stay restricted in Company Memory?
Yes. Up to 3 permission groups are configured before launch, and answers respect them at the retrieval layer, not just at login. A question about a restricted decision returns nothing to someone outside that group, even if their wording matches the source document closely.
How long does AI Company Memory take to launch?
Delivery runs on a fixed schedule from a complete intake, and the delivery panel shows that date plus express and rush options before you buy. Most of the time goes into ingesting your decision sources, building citation and permission logic, and running the 60-question benchmark test before your review.
Stop losing decisions to a channel nobody can find.
Order AI Company Memory and it goes live on a fixed delivery date, turning meeting notes and decisions into a cited, searchable record.
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 Microsoft 365
Answers staff questions from your documents with citations, connected natively to Microsoft 365 so it reads and cites SharePoint, OneDrive and Teams…