AI approach
Math for the number.
AI for the work around it.
People for the call.
Value Alpha computes the valuation with deterministic engines and spends AI on the work around it: reading documents, structuring messy financials, clearing busywork. The decision stays with you. In private markets, the alpha is people.
We believe a valuation must be explainable, auditable, and repeatable. Where real money moves on a number, those standards are not negotiable.
So the number is computed, not generated. The engines are deterministic code: the same inputs return the same range, every time, and every method, weight, and assumption is on the page. AI works where it speeds the job, never where it would soften the number. It reads documents, structures messy financials, files the paperwork, and explains results. It does not set a multiple, a discount rate, or the range.
What that buys you is time. The busywork clears out of the way, and your attention goes where the alpha is: judgment, relationships, and the deal itself. In private markets the alpha has always been people. The software only makes sure that is where your hours go.
Author: Tomasz Felpel, CEO, Value Alpha
Unstructured in, structured out
VA reads PDF and spreadsheet financials into standardized statements, and every figure keeps the page and line it came from. Deterministic checks then re-verify the accounting: the balance sheet must balance, margins must add up, and anything that does not tie is flagged before it can reach the valuation. Every extracted figure is put in front of you to confirm or correct.
Illustrative figures. Not a real company and not a real valuation.
How we calculated this
EBITDA = Gross profit − Operating expenses + Add-backs
Where each figure came from
Select a source to open the document.
FY2025 accounts.pdf · Gross profit, page 4
Northwind Roofing · Profit and loss · FY2025
Highlighted where we read this figure.
1 thing to check
FY2024 gross profit does not tie to revenue less cost of sales, off by $18K. Flagged before it reached the valuation.
Same inputs, same range, every time
Every method a buyer, lender, or broker would run works in parallel: DCF, comparable companies, precedent transactions, the asset floor, and the model built for your sector. 29 valuation methods across 43 industries, and none of them samples from a language model. Each conclusion is shown next to the weight it carried, so the range is checkable rather than asserted.
Illustrative figures. Not a real company and not a real valuation.
Northwind Roofing
$5.5M
Revenue
$2.9M
LTM
EBITDA
$895K
30.8% margin
EV / EBITDA
5.9x
blended
A workflow that points at the alpha
The Deal Book holds every target by stage, from first contact to close. VA does the sweeping: it files documents, pre-fills the model, and names the next step on each deal, so a Tuesday afternoon goes into the two deals that deserve it rather than the folder that does not.
Illustrative figures. Not a real company and not a real valuation.
Northwind Roofing
$5.5M
Next: confirm add-backs
Cedar Mechanical
$2.1M
Halden Plumbing
$3.4M
Next: request payroll detail
Ardent Facades
$8.2M
Next: draft the LOI
None
None
Marek Insulation
$2.6M
People are the alpha
Private markets do not clear on a screen. Prices are found by people: in diligence questions, in a seller's reasons, in the judgment to walk away. The analyst explains any number on the page and names the assumption driving it. The decision it supports stays where it belongs, with you.
Illustrative figures. Not a real company and not a real valuation.
Why is the DCF higher than the comps?
The DCF assumes margin holds at 30.8% through the forecast, while the comps set prices this sector nearer 22%. Most of the $1.2M gap is that single assumption. Lower it to 25% and the two land within 8%.
People to people
We help you buy, sell, and invest in private businesses. The software carries the work. People close the deal.
Questions people ask
Does Value Alpha use AI to calculate my valuation?
No. The valuation engines are deterministic code, so the same inputs return the same range every time. AI reads your documents, suggests the sector, and explains results. It does not set a multiple, a discount rate, or the range.
What does AI actually do in Value Alpha?
It reads PDF and spreadsheet financials and cites the page it took each figure from, suggests the sector classification you confirm, files documents and pre-fills the deal workflow, and answers questions about the valuation on screen.
How do I know where a number in the valuation came from?
Every extracted figure records the page and line it was taken from, and deterministic accounting checks flag anything that does not tie before it reaches the valuation.
Is the output AI generated?
The reading and explaining layers are, and their output is marked for your review. The valuation math itself contains no language model: no multiple, discount rate, weight, or figure in the range is model-generated.
Can AI value a business?
Parts of the job. AI can read financials, classify a company, and explain a result. Value Alpha does not let it set the number: a valuation that real money moves on must be explainable, auditable, and repeatable, so the range is computed by deterministic engines and AI does the reading and explaining around that computation.
How does AI value a business in Value Alpha?
It does not value it; it prepares and explains the valuation. AI turns uploaded statements into structured figures with the page each one came from and suggests the sector, then 29 deterministic valuation methods compute the range and show each method's conclusion beside the weight it carried.
Footnotes
- Extraction, sector classification, and the analyst use large language models, and their output is marked for your review. A semantic embedding model also helps rank candidate comparables; the multiples themselves come from the observed evidence. The valuation math contains no language model.
- Deterministic means nothing is sampled: a re-run on the same inputs against the same market evidence reproduces the range exactly. The comparable-company and transaction datasets themselves update as markets move, so a valuation re-run months later prices against fresher evidence.
- Figures on this page are illustrative. Northwind Roofing is an invented company, and nothing shown here is a real valuation.
Every method shown, every weight explained, every figure sourced.
Price a business