How Augur Works

Not one opinion.
Fifty debating.

Augur doesn't predict the future with a single model. It spawns 50 independent AI analysts, gives them the same data, and watches what they argue about. The disagreement is the signal.

Scroll to explore
The Problem

Most predictions come from
one voice.

A single model, a single analyst, a single opinion. When they're wrong — and they often are — there's no way to know why or how confident they should have been.

🎯
The old way
One model, one answer
A single AI or analyst looks at the data and gives you a number. "BHP will beat earnings." No explanation of what could go wrong. No measure of confidence. Just an answer.
Black box
The Augur way
Fifty analysts, one debate
50 AI analysts with different personalities and approaches look at the same data — and argue. The result isn't just a prediction. It's a map of where the real uncertainty lives.
Transparent
Think of it this way

It's like asking a
room full of experts.

Imagine you're about to make a big financial decision. Would you rather ask one person — or put 50 specialists in a room and listen to what they argue about?

One friend's opinion
Your mate works in finance. He says "looks good to me." You have no idea how confident he is, what he missed, or what would make him change his mind.
Limited perspective
vs
A room of 50 specialists
The optimists make their case. The pessimists push back. The cautious ones flag risks everyone else missed. After 3 rounds of structured debate, you know exactly where they disagreed — and why.
Richer picture
The 50 Agents

Each one thinks
differently.

Augur creates 50 unique AI analysts before every simulation. Not clones — each has a distinct personality, a different way of reading data, and a different threshold for changing their mind.

📈
The Optimist
10 agents
Focuses on growth signals. Believes in management. Sees opportunities others miss.
📉
The Sceptic
10 agents
Questions everything. Looks for hidden costs. Doesn't trust guidance.
⚙️
The Numbers Person
10 agents
Only trusts patterns and history. Ignores stories. Follows the data.
🛡️
The Risk Watcher
10 agents
Always asks: what could go wrong? Spots tail risks others overlook.
📱
The Everyday Investor
10 agents
Reacts to news and sentiment. Represents how most people actually think.
The Moderator

Not just a referee.
An analyst.

Between every round, a neutral moderator reads all 50 agent responses and structures the next debate. It doesn't take sides — but it makes sure weak arguments get challenged and strong minority views get heard.

📋
Argument Extraction
Finds the 3 strongest bull and bear cases from all 50 agent reasoning strings. Must be specific and evidence-grounded — generic arguments don't qualify.
Outlier Challenge
Identifies agents holding extreme views with low conviction. Sends them a targeted challenge: what specific evidence would move you toward the group?
🎯
Dissent Protection
Flags analysts holding an unusual view with high conviction. Ensures all 50 agents explicitly consider the minority case before updating.
📊
Swing Factor Tracking
Counts which data points agents cite most across all rounds. The top 5 become the simulation's swing factors — surfaced in the final output.
The Debate

They argue.
Three rounds.

Each round, every agent sees the group's current distribution and must decide: hold my position, or update it? Bulls who can't defend their thesis against bear arguments drift down. Bears facing strong data drift up. After 3 rounds, the positions that survive are the ones with the strongest evidence behind them.

Watch how opinion shifts across rounds for a real BHP simulation.

BHP Group — Earnings Simulation
Round 0 / 3
Probability of beating earnings consensus
Round 1
P(beat): 55.4%9 bears · 20 neutral · 21 bulls
Agents anchor to real yfinance data — analyst consensus "hold" with +7.8% upside shapes opening positions.
Moderator
Extracts strongest bull and bear arguments. Challenges outliers. Flags high-conviction dissenters. Shapes the next round.
Round 2
P(beat): 54.8%11 bears · 19 neutral · 20 bulls
Bears push back on iron ore pricing. Quant traders hold near centre. Bulls defend copper thesis.
Moderator
Updates argument ranking. Tracks which data points drove the most position changes. Identifies the debate's swing factors.
Round 3
P(beat): 53.6% → TOSS-UP13 bears · 18 neutral · 19 bulls
Final verdict: a contested call. The swarm is genuinely divided — real data grounds the debate.
See It In Action

See it
happen.

A real simulation. Watch the debate unfold — from data to verdict in 3 minutes.

The Output

Not just a prediction.
A map of the debate.

The most valuable part isn't the verdict. It's the swing factors — the specific variables the moderator identified as most contested across all rounds. Computed from what agents actually argued about, not generated afterward.

augur · simulation complete · BHP.ASX
BHP.ASX
Probability Distribution · 50 agents · 3 rounds
Will beat expectations26%
Will meet expectations18%
Will miss expectations56%
Verdict
Lean Miss
56% of agents expect a miss. Moderate conviction.
Swing factors — identified by moderator
Cost inflation — are rising costs temporary or permanent?
High contention
Management guidance — do they typically deliver on promises?
Contested
Operational execution — is production running to plan?
Contested
China demand — is the slowdown already priced in?
Consensus
Currency moves — AUD/USD impact on reported earnings
Consensus
Understanding Results

What the
verdicts mean.

Every simulation ends with a verdict. Here's what each one tells you — in plain language, not finance jargon.

LIKELY MISS
The group strongly expects a miss
Most agents believe earnings will fall short of what analysts predicted. High conviction.
LEAN MISS
More agents lean bearish, but it's not a strong call
The group tilts toward a miss, but there's meaningful disagreement. Moderate conviction.
TOSS-UP
Genuinely uncertain — the group is split
Roughly equal arguments for beat and miss. The swing factors section tells you where they disagreed.
LEAN BEAT
More agents lean bullish, but it's not a strong call
The group tilts toward a beat, but bears made valid counter-arguments. Moderate conviction.
LIKELY BEAT
The group strongly expects a beat
Most agents believe earnings will exceed expectations. Strong analyst consensus and financial data support.
Each verdict is based on 50 agents debating across 3 rounds, anchored to real financial data.
Why It Matters

Three things you get
nowhere else.

Augur isn't built to replace your thinking. It's built to show you what the debate looks like before you make up your mind.

01 — Disagreement
Where experts don't agree
When 48 of 50 agents agree, the outcome is fairly predictable. When they're split 26/24, you're looking at genuine uncertainty. Augur shows you which scenario you're in — before the announcement.
02 — Reasoning
Not just what — but why
Every swing factor comes with a bull argument and a bear argument, written in plain English. You don't need to be a finance expert to understand what the debate is about.
03 — Honesty
It tells you when it doesn't know
When agents stay widely split after 3 rounds, Augur flags HIGH UNCERTAINTY. It won't give you false confidence. A tool that admits uncertainty is more useful than one that always sounds sure.

Ready to hear
all 50 voices?

Pick any ASX 100 company. Run a simulation. See what 50 AI analysts argue about — in under 3 minutes.

Run a simulation →
Free to try · No account required · Not financial advice
Used Augur? Share your feedback →