> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bono.network/llms.txt
> Use this file to discover all available pages before exploring further.

# Deliberation Modes

> Multi-perspective AI analysis for complex legal questions

<Note>
  **Beta Feature** - Deliberation and Grand Deliberation modes are currently in beta. Functionality may change as we refine the multi-analyst system.
</Note>

<Frame>
  <img src="https://mintcdn.com/bononetwork/iesyQdBb07VQFOI9/images/deliberation.png?fit=max&auto=format&n=iesyQdBb07VQFOI9&q=85&s=cfa8482d39e64898bbf9c5d95e43e5c1" alt="Deliberation mode showing three AI analysts" width="2278" height="1296" data-path="images/deliberation.png" />
</Frame>

## Why Multiple Perspectives Matter

Legal problems rarely have a single right answer. The best counsel considers multiple angles: what's legally sound, what minimizes risk, and what achieves the client's objectives. Arbiter's Deliberation modes replicate this by deploying three AI analysts simultaneously, each approaching your question from a distinct vantage point.

This adversarial approach is particularly valuable when:

* You're advising a client on strategy and need to present balanced options
* There are competing interests or potential conflicts to navigate
* The legal question involves risk/reward tradeoffs
* You want to stress-test your own analysis before presenting it

## The Adversarial Trinity

<CardGroup cols={3}>
  <Card title="Arbiter" icon="scale-balanced">
    **Lead Strategist**

    Balanced, practical analysis weighing precedent, policy, and real-world outcomes. Synthesizes the other perspectives into actionable guidance.
  </Card>

  <Card title="Anais" icon="shield-check">
    **Risk Guardian**

    Conservative perspective focused on compliance, ethics, and downside protection. Flags potential pitfalls, regulatory concerns, and reputational risks.
  </Card>

  <Card title="Abaddon" icon="chess">
    **Aggressive Tactician**

    Creative, assertive approach that identifies leverage points, novel arguments, and maximum-value strategies. Proposes options you might not have considered.
  </Card>
</CardGroup>

***

## The Three Modes

### Standard Mode

**Single Analyst • Fast Turnaround**

In Standard mode, only Arbiter responds. This is appropriate for:

* Routine legal questions with straightforward answers
* Quick lookups of statutory provisions or regulations
* Preliminary research before deeper analysis
* Low-stakes or time-sensitive inquiries

**Response time:** Approximately 30 seconds\
**Token usage:** \~10-30 tokens per query

***

### Deliberation Mode

**Three Analysts • Cross-Aware Dialogue**

All three analysts work in parallel. Critically, each analyst can see what the others are saying and refines their analysis in response. This produces more nuanced, battle-tested conclusions than any single perspective.

Deliberation is appropriate for:

* Questions involving multiple jurisdictions or regulatory regimes
* Strategic decisions with meaningful risk/reward tradeoffs
* Situations where you need to present options to a client
* Legal questions with unsettled or developing doctrine

**Response time:** 2-5 minutes\
**Token usage:** \~30-60 tokens per query

***

### Grand Deliberation Mode

**Premium Reasoning • Maximum Depth**

Grand Deliberation uses the same three-analyst approach but with extended reasoning capabilities. Each analyst engages in deeper, multi-phase analysis before contributing. This is Arbiter's most thorough mode.

Reserve Grand Deliberation for:

* Critical strategic decisions (significant liability, major transactions)
* Complex multi-jurisdictional or multi-party matters
* High-stakes advice requiring maximum analytical rigor
* Situations where you need comprehensive documentation of the analysis

**Response time:** 10-30 minutes\
**Token usage:** \~60-150 tokens per query

***

## How the Cross-Aware Dialogue Works

In Deliberation modes, the three analysts don't simply work in isolation. The process unfolds in phases:

<Steps>
  <Step title="Initial Analysis">
    Each analyst independently forms their initial perspective on your question.
  </Step>

  <Step title="Cross-Reading">
    Each analyst reviews what the other two have said.
  </Step>

  <Step title="Refinement">
    Each analyst refines their position in response to the others, acknowledging valid points and distinguishing where they disagree.
  </Step>

  <Step title="Synthesis">
    A final synthesis integrates all three refined perspectives, highlighting areas of agreement, key disagreements, and a balanced recommendation.
  </Step>
</Steps>

This process produces analysis that has been stress-tested against alternative viewpoints, similar to how a good internal memo might be reviewed and challenged by colleagues before going to a client.

***

## Reading the Three-Column Output

When you receive a Deliberation or Grand Deliberation response, you'll see three panels:

| Arbiter                       | Anais                                      | Abaddon                      |
| ----------------------------- | ------------------------------------------ | ---------------------------- |
| Balanced strategic assessment | Conservative risk analysis                 | Aggressive tactical options  |
| Weighs all relevant factors   | Identifies compliance and ethical concerns | Proposes creative approaches |
| Practical recommendations     | Flags potential pitfalls                   | Maximizes client leverage    |

Below the three panels, a **Synthesis** section:

* Identifies points of agreement among the analysts
* Highlights key areas of disagreement and why
* Provides a balanced recommendation
* Notes which perspective to prioritize depending on your client's risk tolerance and objectives

<Tip>
  Don't skip directly to the synthesis. Understanding *why* Anais flagged a particular risk or *what* aggressive option Abaddon proposed often provides the most valuable insight for client counseling.
</Tip>

***

## Reasoning Transparency

All modes include expandable reasoning summaries (click the purple card to expand):

<Tabs>
  <Tab title="Standard Mode">
    * Single reasoning card showing Arbiter's thought process
    * Key factors considered
    * How conclusions were reached
  </Tab>

  <Tab title="Deliberation Mode">
    * Three reasoning cards (one per analyst)
    * Shows how each analyst considered the others' perspectives
    * Documents what influenced their refinement
  </Tab>

  <Tab title="Grand Deliberation">
    * Extended reasoning summaries
    * Full multi-phase thinking process visible
    * Maximum transparency for auditing and compliance documentation
  </Tab>
</Tabs>

This transparency is valuable for:

* Understanding the basis for conclusions
* Documenting your diligence process
* Explaining AI-assisted analysis to clients or supervisors
* Identifying which assumptions to verify independently

***

## Practical Examples

<AccordionGroup>
  <Accordion title="M&A Structure Analysis (Grand Deliberation)">
    **Question:** "Analyze the tax and regulatory implications of structuring this acquisition as an asset purchase versus a stock purchase. The target is a Delaware corporation with California employees. We're concerned about federal antitrust review."

    **Why Grand Deliberation:** Multi-jurisdictional analysis across tax, corporate, employment, and antitrust. High-stakes strategic decision where maximum analytical depth is warranted.

    **Expected Output:** Arbiter provides a balanced comparison of the two structures. Anais flags regulatory risks (antitrust, employment liabilities assumed). Abaddon proposes creative structuring options to optimize tax position or minimize regulatory exposure.
  </Accordion>

  <Accordion title="Contract Negotiation Strategy (Deliberation)">
    **Question:** "The counterparty rejected our proposed indemnification cap of 1x the contract value. They want unlimited indemnification for IP claims. What are our options?"

    **Why Deliberation:** Need balanced market context (what's standard), risk perspective (exposure if we accept), and aggressive tactics (leverage points and alternatives).

    **Expected Output:** Arbiter analyzes market standards for similar deals. Anais quantifies the risk exposure and flags specific IP concerns. Abaddon proposes counter-strategies (carve-outs, insurance requirements, knowledge qualifiers).
  </Accordion>

  <Accordion title="Litigation Response Strategy (Deliberation)">
    **Question:** "We received a demand letter alleging trade secret misappropriation by a former employee we hired six months ago. The letter demands immediate injunctive relief and threatens suit. What are our response options?"

    **Why Deliberation:** Need risk assessment (liability exposure, injunction likelihood), strategic options (settle, fight, negotiate), and balanced recommendation.

    **Expected Output:** Arbiter outlines the landscape and likely timeline. Anais assesses litigation risk and potential damages. Abaddon proposes aggressive defense strategies and potential counterclaims.
  </Accordion>

  <Accordion title="Regulatory Filing Question (Standard)">
    **Question:** "What are the filing requirements and timeline for Form D under Regulation D after closing a Series A round?"

    **Why Standard:** Straightforward regulatory question with a clear, settled answer. No meaningful tradeoffs to analyze.
  </Accordion>
</AccordionGroup>

***

## When to Use Each Mode

| Situation                               | Recommended Mode         | Why                                           |
| --------------------------------------- | ------------------------ | --------------------------------------------- |
| Quick legal question                    | Standard                 | Fast answer, low token cost                   |
| Research for a memo                     | Standard                 | Single perspective sufficient                 |
| Strategic advice to client              | Deliberation             | Multiple perspectives for balanced counseling |
| Negotiation strategy                    | Deliberation             | Need risk vs. opportunity analysis            |
| Major transaction structuring           | Grand Deliberation       | Maximum depth for high-stakes decision        |
| Litigation strategy on significant case | Grand Deliberation       | Comprehensive analysis of options             |
| Compliance question                     | Standard or Deliberation | Depends on complexity                         |
| Multi-jurisdictional analysis           | Deliberation or Grand    | Multiple perspectives essential               |

***

## Token and Time Summary

| Mode               | Typical Token Usage | Response Time |
| ------------------ | ------------------- | ------------- |
| Standard           | 10-30 tokens        | \~30 seconds  |
| Deliberation       | 30-60 tokens        | 2-5 minutes   |
| Grand Deliberation | 60-150 tokens       | 10-30 minutes |

Grand Deliberation uses more tokens but provides correspondingly deeper analysis. Match the mode to the importance of the question.

***

## Best Practices

<CardGroup cols={2}>
  <Card title="Match Depth to Stakes" icon="balance-scale">
    Don't use Grand Deliberation for simple questions, and don't use Standard for critical strategic decisions. The analytical depth should match the importance of the matter.
  </Card>

  <Card title="Read All Three Perspectives" icon="eye">
    In Deliberation modes, resist the temptation to skip to the synthesis. The individual analyst perspectives often contain the most actionable insights.
  </Card>

  <Card title="Use for Client Presentations" icon="user-tie">
    The three-perspective format translates well to client advice. It demonstrates you've considered the question from multiple angles and can present balanced options.
  </Card>

  <Card title="Enable GitLaw Research" icon="globe">
    Enable web research in Deliberation modes. Each analyst will incorporate current law, recent cases, and regulatory developments into their analysis.
  </Card>
</CardGroup>

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="Analysis taking longer than expected">
    Deliberation (2-5 minutes) and Grand Deliberation (10-30 minutes) involve multiple analysts working in sequence. Complex questions with web research enabled take longer. This is normal.
  </Accordion>

  <Accordion title="Analysts reaching similar conclusions">
    For clear-cut questions, the three perspectives may converge. This indicates strong consensus and should increase your confidence in the conclusion. To see meaningful divergence, ask about questions with genuine strategic tradeoffs.
  </Accordion>

  <Accordion title="Need to leave during Grand Deliberation">
    You can navigate away during a Grand Deliberation session. The analysis will continue, and results will be available when you return.
  </Accordion>

  <Accordion title="Unsure which mode to use">
    When in doubt, start with Standard mode. It's fast, reliable, and sufficient for most legal questions. Escalate to Deliberation when you specifically need multiple perspectives on a strategic question, and reserve Grand Deliberation for high-stakes matters requiring maximum analytical depth.
  </Accordion>
</AccordionGroup>

***

## Next Steps

<CardGroup cols={2}>
  <Card title="AI Research" icon="brain" href="/guides/ai-research">
    Overview of research capabilities and the Adversarial Trinity in action
  </Card>

  <Card title="GitLaw" icon="scale-balanced" href="/guides/gitlaw">
    How Arbiter performs verified legal research
  </Card>
</CardGroup>
