Excel model → memo, deck, emailInspectable open-source core

From Excel model to reviewer-ready memo in Microsoft Office.

The analyst selects the Excel tabs and approved data. ValuClaw drafts the Word memo section with every number tied to a cell, every assumption flagged, and save locked until the reviewer approves.

Boutique investment banksLean PE and private credit teamsFamily offices

Built for draft-and-review investment work. It does not trade, make investment decisions, or send external communications without an approved human review path.

Model-to-memo runReview first
Model routeAuto or an approved named model
SkillWeekly update v0.1
ContextExcel model + approved data + memory

Office draft

Model commentary section

Drafted with citations attached, one assumption flagged, and save/send locked until approval.

5 sources1 review itemeditable in Word
Does not crawl the full mailbox, overwrite files, or write to CRM without a reviewer.

Start with a deliverable

The first win is a draft your MD can edit in Word before the IC meeting.

The Monday-morning model commentary. The diligence note the associate writes at 11 PM. The IC deck page nobody wants to draft. Pick one, name the inputs, and let ValuClaw produce the first version — the senior reviewer still decides what is right.

Excel model to memo

The analyst selects the Revenue Build, Debt Schedule, and Assumptions tabs from the model they already use. ValuClaw drafts the Word commentary section with every number tied to a cell.

Diligence note

The analyst drops in the CIM, data-room files, management presentation notes, and the latest 10-K. ValuClaw drafts red flags, open questions, and source-backed sections for the deal lead to review.

IC or client pack

Portfolio company metrics, board minutes, and approved fund data become the Monday IC deck page or the LP update your team sends every month.

Follow-up draft

After the IC meeting, the approved memo becomes the Outlook email to the client, the Teams note to the deal team, or the DealCloud update — drafted but not sent until a reviewer approves.

Workflow review

Four weeks. One deliverable. A draft your team can actually use.

Week 1: map the deliverable and allowed inputs. Week 2: configure the skill, memory, and model. Week 3: produce a source-backed draft in Word or PowerPoint on real context. Week 4: measure review time, source coverage, and the edits the reviewer still makes.

01

Bring one painful deliverable

The Monday-morning model commentary. The diligence note the associate writes at 11 PM. The IC deck page nobody wants to draft. The LP update that takes all of Friday.

02

Name the allowed context

Which workbook tabs, which folders, which Capital IQ screens, which FactSet pulls, which internal memos, and which model the LLM is allowed to read for this task.

03

Draft inside Microsoft work

The output lands as an editable Word section, PowerPoint page, or Outlook draft — not in a separate portal. Sources sit beside the text, not buried in a chat log.

04

Review the evidence

Time to first draft. Source coverage percentage. Unsupported claims flagged. Assumptions that changed since the last version. Edits the reviewer still makes. All measured, all visible.

Every claim has a trail

$48.2M came from cell F87. The reviewer can check that.

Every number in the draft links to the cell, document, or data source it came from. The churn assumption that changed from 3.2% to 4.0% since v11 is flagged. The model route — GPT, Claude, or Auto — is logged beside the text. The reviewer checks the draft without reconstructing the prompt.

IC memo — draft sources attached

Topline reached , up as the new enterprise cohort ramped, while net leverage held at . The base case assumes .

Tap a highlighted figure to see where it came from.

$48.2MFY24 revenue
Model_v12.xlsx
Revenue Build · cell F87

Pulled live · model v12, today 08:14

No model or workflow lock-in

Your models. Your data. Your memory. Your reviewer.

Pin GPT for the memo, switch to Claude for the diligence note, or route to a local model behind the firewall. Your files stay in SharePoint. Your memory is a plain list you can edit. The reviewer approves before anything saves or sends.

Model choice stays visible

Start with Auto, then pin GPT, Claude Opus, Gemini, DeepSeek, or a local model behind your firewall. The model used for each run is logged beside the draft, not hidden in a prompt.

Your data stays where it is

ValuClaw reads the Excel tab you selected, the SharePoint folder you pointed to, the Capital IQ screen you approved. It does not copy your files into a new database or become a system of record.

Memory is explicit

Team memory is a list of approved facts — house style, sector assumptions, review preferences. You see exactly which items will enter each model call. Edit them, remove them, or turn them off before the run.

Try a sample workflow

Pick a model, toggle the context, run the draft.

Choose Auto or pin a model. Toggle the Excel model, the approved data connector, and team memory. Pick Word, PowerPoint, or Outlook. Run it. You will see the drafted commentary, the source pills, the changed-assumption flag, and the approval gate — on synthetic data, but the control pattern is what we build with each customer.

Model-to-memo draftsample data

Tell ValuClaw the job

Prepare the weekly update from the model, notes, and approved data.

Start with Auto or pin a familiar model, attach only the context you approve, run a reusable skill, and keep the output editable in Word, PowerPoint, or Outlook.

Model routeAuto
SkillWeekly update v0.1 · inspectable

Auto chooses from the approved fast and balanced model pool; the actual route is shown after the run.

Choose what the LLM sees

Pick the Office draft

What happens before it runs

Will readselected workbook tabs · approved data connector · approved memory
Will not touchmailbox crawl · CRM writes · SharePoint crawl
Approvalrequired before send or save

Run this synthetic sample

A server-side OpenRouter key is used only for the selected sample context. The selected and actual model route are shown with the draft.

Word section Ready for review

Topline and leverage

Topline reached $48.2M in FY24, up 21% YoY. Net leverage held at 3.1x.

Assumption requiring review

The base case assumes 4.0% churn, changed from 3.2%; confirm with the deal team before distribution.

Open review items

Confirm churn assumption change, then approve the Outlook draft or shared-folder save.

Review one change

Assumption changed

Churn assumption 3.2%Churn assumption 4.0%

Still proposed; reviewer can reject.

Verified

Citation coverageok
Numerical traceabilityok
Stale source checkok
Unsupported claim guardok

Approval

The draft stays local until you approve the save or send.

Data integrations

Connect the finance systems your team already pays for.

Capital IQ screens, FactSet feeds, PitchBook searches, Preqin fund data, Datasite and Intralinks data rooms, DealCloud CRM, broker research portals, internal APIs, SharePoint folders, and Office add-ins. Each connector respects the entitlements and redistribution limits in your vendor agreements.

Microsoft 365SharePointOneDriveExcelOutlookTeamsCapital IQFactSetPitchBookPreqinDealCloudDatasiteIntralinksFilingsBroker research portalsData warehousesInternal APIsOffice add-ins

FAQ

Straight answers.

ValuClaw is an Office-native AI associate for investment teams. An analyst selects the Excel tabs, notes, and approved data sources for a specific deliverable. ValuClaw drafts the Word memo section, diligence note, or IC update with every material claim tied to a cell, document, or data source — and a reviewer approves before it goes anywhere.

Start small

Send us the deliverable your team dreads drafting. We will map it in four weeks.

The Monday-morning memo. The Friday LP update. The diligence note at 11 PM. Tell us which one, who reviews it, and where it lives. We will build a source-backed draft path your team can inspect, edit, and keep.