Skip to content
The #1 question we get

“Can’t I just paste this into ChatGPT?”

You can get prose from an LLM. You cannot get intent-aware threshold logic, evidentiary sufficiency, and deterministic screening from one.

Not just citation. Intent, fear, signal, verdict.

12
Canonical intents
67
Fear domains
274
Signals in the knowledge graph
53
Structured calculators
LLM-style answer

Sounds polished. Still has no threshold logic.

Unverified

Based on the company’s financial profile, leverage appears manageable and the business seems strategically attractive. However, investors should monitor concentration risk and cash flow trends before proceeding.

No sufficiency threshold
No activated fear profile
No metric-set resolution
No repeatable verdict path
Threshold evidence record

Structured, sourced, and sufficient for a verdict.

Deterministic
Yield acquisition · gross debt / EBITDA
6.95x
CRITICAL
Fear activated
Margin erosion destroys distribution capacity
Sufficiency state
Rejection sufficiency met on leverage threshold
Source stack
Document extraction, calculator pass, cross-check
Metric set
Jurisdiction-aware yield thresholds resolved for this buyer
The practical difference

What changes once the output has to survive real committee scrutiny.

Not just “where did this number come from?”

It asks whether the evidence is already sufficient to conclude the deal breaches the stated thresholds.

It can re-evaluate the same extraction under different buyer intents without rerunning the whole pipeline.

Generic LLM output

Evidentiary sufficiency

An LLM can describe a red flag, but it cannot tell you whether the accumulated evidence is already sufficient to reject the deal.

IACalc system output

Evidentiary sufficiency

IACalc applies rejection sufficiency. One confirmed fatal flaw can justify a critical threshold breach; proving yes requires broader absence checks.

Generic LLM output

Intent -> Fear -> Signal -> Verdict

Generic output ignores buyer context. Yield, strategic IP, and distressed buyers all get the same soft answer.

IACalc system output

Intent -> Fear -> Signal -> Verdict

IACalc activates a fear profile from the thesis, evaluates the relevant signals, and produces a verdict calibrated to that specific buyer.

Generic LLM output

Progressive evidence accumulation

Reads the current prompt in isolation. It does not know what evidence is still missing or what has already become decisive.

IACalc system output

Progressive evidence accumulation

IACalc accumulates evidence from documents, web, and image sources, then shows what is sufficient, inconclusive, and still needed.

Going deeper

What the system is actually doing under the hood.

Negative selection, not generic summary

The core question is not “is this deal good?” It is “does this deal breach the thresholds for this mandate?” The system is designed to find disqualifying evidence early, before sunk cost and deal momentum distort judgment.

The IFKS chain is the product

Intent activates fears. Fears activate signals. Signals roll into verdicts. That Intent -> Fear -> Signal -> Verdict chain is what makes the analysis explainable and thesis-aware.

Three-layer architecture

Knowledge graph, metric sets, and evaluation engine are separated. The computational core stays stable while the analytical coverage keeps expanding.

More than source citation

Every number is citable, but citation is only the trust layer. The deeper value is knowing what that number means for this buyer, under this metric set, at this sufficiency state.

Industrial analytical coverage

The engine runs 53 calculators, 15 qualitative modules, and external enrichment collectors. It is a due diligence system, not a clever prompt.

One extraction, many buyer views

The same extracted data can be evaluated against multiple buyer intents with near-zero marginal cost, because thesis application happens in evaluation, not extraction.

See the difference

Look at a report your team could actually defend.

The point is not to get an answer faster. The point is to get an answer that survives internal review, committee scrutiny, and follow-up diligence.