AI DEMOS · USE CASE 01 · TRADING

One question,
three AIs, one conclusion.

A demo of a parallel analysis: the same investment question about Komerční banka, asked to Claude, ChatGPT and Gemini at the same time. Each answer stands on its own, carries the exact model version, and the page closes with the joint discussion of all three.

  • Fictional demo — illustrative answers
  • Not investment advice
  • Model versions labelled

THE SETUP

What was asked,
and how.

One prompt, written once and sent to all three assistants on the same day. No extra documents, no private data — only public information each model already has.

THE PROMPT

“Analyse Komerční banka as a potential investment today: how it earns money, what the dividend looks like, the main risks, and a clear buy / hold / sell recommendation with reasoning.”

THE ANSWERS

Three independent answers.

Each answer below is illustrative demo content written to show the format. The version badge is what makes the format auditable — anyone can rerun the question against the same model.

ANTHROPIC

Claude

claude-sonnet-5-5

Verdict: Hold

  • Earnings lean on net interest income, so Czech rate cuts press on the core business.
  • The dividend stays the strongest part of the case, with a yield well above the sector.
  • Ownership by a large Eurozone group caps takeover upside, so re-rating needs the dividend to hold.

Confidence: medium — public data only.

OPENAI

ChatGPT

gpt-6-sol

Verdict: Buy

  • Trading well below book value, which leaves room for the price to follow the dividend.
  • The capital position is comfortable enough to keep payouts stable through a slower year.
  • The real risk sits in credit losses if the economy turns; size it as a position, not a bet.

Confidence: medium-high — valuation-driven.

GOOGLE

Gemini

gemini-3.8-pro

Verdict: Hold, with a written trigger

  • The case is fair value plus the dividend; nothing in the numbers suggests obvious mispricing.
  • The rate path is the swing factor: one more cut squeezes margins further.
  • A hold with a written trigger — yield above a set level, or guidance raised — beats buying today.

Confidence: medium — depends on the rate path.

JOINT DISCUSSION

What the three said together.

Read side by side, the answers converge more than they clash. The joint discussion below is the part a team would actually act on.

01

Where all three agreeThe dividend is the core of the case; the price alone is not the argument. All three treat the capital position as solid.

02

Where they splitChatGPT buys on valuation; Claude and Gemini wait. The gap is not about the bank — it is patience versus price, and the demo shows it in one line each.

03

What a human decidesPosition size, horizon, and whether the dividend fits the portfolio’s tax picture. The models narrowed the work to three lines a person can argue with.

WHY THE BADGES

Why every answer names its model.

Models change under you: an answer that impressed you in March can differ in October from the same product. A version badge turns a screenshot into an audit trail — anyone can rerun the question against the same model and check the reasoning.

Demo content. The answers above are fictional illustrations of the format — not real model outputs, and not investment advice.

WANT THIS ON YOUR OWN QUESTION?

We will build the same
demo for your case.

Your question, your approved tools, the same auditable format — usually ready within days of a first call.