Paste the teaser, the CIM summary or the broker's email. Your browser pulls out the numbers,
checks them against each other and screens them against your fund's box as you type - free,
before you sign in. Then the screen writes a one-page memo with a verdict, and
the meeting prep turns what the screen could not answer into the questions for
the first call. Every quote it writes is checked against the material.
Each example comes with a saved model run, one per verdict plus a meeting prep, so you can see the whole page for free.
Screening
criteria met
missed
red flags
prescan flags answered
statements checked
disagreements
Changes since your last screen of this deal
Compared with the most recent saved screen under the same deal name. It is in the memo .md too.
Deal facts
Against your box
Criterion
Target
Actual
Call
Comment
Bull case
Bear case
Red flags
Questions for a first call
Prescan flags, answered
Reconciliation with the material
Every quote was searched for in the material, every criterion was compared with the browser's arithmetic, every prescan flag had to be answered, and every dollar figure in the facts had to be in the material. A disagreement means the screen, not your material, is wrong.
Bottom line
questions
must-ask
concerns addressed
statements checked
disagreements
Objectives
Question list
* must-ask. Lead with the open questions; let them talk, then follow up.
Benchmarks to reference
Metric
Value
Source
Use
Red flags to probe
Follow-up requests
Reconciliation with the material and your concerns
Every concern had to get a question, every reference had to point at a real concern, flag or screen question, no question may lead the witness, and benchmarks from the material had to carry figures the material has.
How to run it
Raw model reply
The reply did not parse as the structured result, so it is shown as it arrived.
Your recent runs
If that did not work
Check that the material has some numbers in it, that a meeting prep has its meeting type set,
that you are signed in, and that your balance covers the amount reserved next to the button. The
three examples always work and cost nothing, so they are the quickest way to tell whether the
problem is your material or the service.
What this does, and what it does not
The prescan reads text. Each dollar figure is read with its scale ($8.1M, $8,100K and 8.1 under a
"($M)" header are one number) and filed under the metric named nearest to it and the period next
to it. Stated EBITDA margins are recomputed from EBITDA and revenue, stated growth and CAGRs from
the revenue series (any span that produces the rate passes), and a stated multiple from the price
and EBITDA. The box is compared row by row, and anything the material does not say is
"unknown", never assumed. Sector, geography and management are keyword reads, so the screen may
overrule them with a reason.
It knows nothing the material does not say: no market data, no comparable companies, no
background on the people. Benchmarks it cannot source are listed as things to pull, without
figures. The screen and the meeting prep are first-look work, not investment advice. Derived
from the agent skills
@anthropics/deal-screening
and @anthropics/dd-meeting-prep
(anthropics/financial-services-plugins, Apache-2.0). The example companies are fictional.