Home / Break Your Research Filter Bubble with Multi-AI Synthesis

Break Your Research Filter Bubble with Multi-AI Synthesis

⋅

Ng Chong

Why We Built This

Traditional research is broken. You rely on a single search engine or database, and what do you get? A “filter bubble” — algorithmic biases that hide critical information and diverse perspectives, leaving you with an incomplete, skewed understanding of your topic.

That’s the problem we set out to solve with the Deep Research Agent AI.

Breaking Out of the Filter Bubble

Here’s the key insight: you can use different AI providers for research, fuse the results, and let an AI judge create something better than any single source could produce.

Think of it like tuning into different media channels. When you want a fuller picture of a complex topic, you don’t rely on just one source – you switch between networks and local outlets. Each offers its own perspective, strengths, and blind spots. Comparing them brings you closer to the full story.

The Deep Research Agent works the same way:

  1. Research with different providers — Run the same research plan through multiple LLMs (Claude, Perplexity, Gemini, OpenAI, etc.). Each one brings different training data, different search strategies, and different blind spots.
  2. Fuse the reports — Compare what each AI found. Notice where they overlap (consensus) and where they diverge (unique insights).
  3. Judge verifies against the plan — Choose your preferred LLM to act as an impartial judge. It evaluates each report against your original research plan, scoring how well each objective was met.
  4. Pick the best or synthesize something superior — Select your best report (highest scoring report) or top 2+ reports and have the AI Judge create a final synthesis that combines the strengths of all versions into something better than any individual report.

The result? A more balanced, thorough understanding than you could ever get from manual searching or a single AI.

How It Works: A Simple 3-Step Process

Step 1: Define Your Topic

Start with a clear research question. The more specific, the better. Think:

  • “How does quantum computing threaten modern cryptography?”
  • “What are sustainable urban planning strategies for coastal cities?”

Then pick your planning engine — the AI that will create your research blueprint. Maybe you choose Claude for its structured thinking, or perhaps Gemini for its breadth of knowledge. Hit Generate Research Plan.

Step 2: Review and Refine

The AI generates a structured plan with objectives and subtopics. But here’s the thing: you’re still the expert. Edit the plan. Add missing angles, rephrase things, tweak the search queries. A few minutes here transforms good research into great research.

Step 3: Configure and Execute

Now comes the fun part — choose your engines:

  • Search Engine: Which AI will hunt down the sources? Perplexity is great for web search, while others might excel at academic databases.
  • Synthesis Engine: Which AI writes the final report? Claude’s prose is clean and structured; OpenAI might bring a different voice.
  • Judge Engine: When you compare multiple reports later, which AI plays referee?

Set your target word count and click Generate New Research. Then go grab a coffee — your AI team is working.

The Multi-Source Workflow: Research, Fuse, Judge, Synthesize

Once you have your first report, the real power kicks in. Here’s what makes the Deep Research Agent different:

Research with Different Providers

Hit Re-Synthesize and choose a different AI model. Maybe your first report used Perplexity for research and Claude for writing. Now try Gemini for research and OpenAI for writing. Same plan, different perspectives.

Fuse and Compare

Now you have multiple reports. Compare them side by side. Notice the differences? One might have better technical depth, another stronger analysis, a third more comprehensive examples. This is where insights emerge.

Judge Verifies Against the Plan

The scoring system provides a detailed evaluation assisted by AI-powered analysis tools. Draft Statistics automatically calculates word count, citations, citation density, and coverage. Citation Overlap Analysis identifies consensus sources (cited by multiple reports) versus unique insights.

Your chosen Judge LLM evaluates each report against your original research plan, scoring metrics like:

  • Objective Fulfillment
  • Question Coverage
  • Depth & Insight

Pick the Best or Create Something Superior

Now you have options:

  • Select your top report based on the scoring and use them as-is
  • Let the AI Judge create a synthesis that merges the strongest elements from each version, resolving contradictions and filling gaps

The judge applies strict quality filtering: it removes unsubstantiated statistics, unsupported claims, vague filler, and redundancy. Every factual claim must include a proper citation. The judge prioritizes consensus sources (information verified by multiple engines) while preserving unique insights. Because of this filtering, the final report may be shorter but more accurate and well-sourced than the originals.

The final synthesis is a coherent report that combines the best elements from each version into something better than any individual source could produce.

Why This Matters

In a world of information overload and echo chambers, having a tool that deliberately seeks diverse perspectives is invaluable. Whether you’re a student, a professional researcher, or just someone who loves learning deeply, the Deep Research Agent turns what used to take hours into minutes while giving you a more balanced, comprehensive view.

Coming Soon: Open Source

We believe in the power of community-driven innovation. That’s why we’re planning to open source the Deep Research Agent soon. Developers will be able to:

  • Add new search engines — Integrate additional web and academic search APIs
  • Add new LLM providers — Extend support for more AI models as they emerge
  • Devise advanced judging mechanisms
    • Multi-Plan Comparative Analysis
      • Compare reports against multiple research plans on same topic
      • Identify coverage gaps: “What questions did Report A answer that Report B missed?”
      • Detect contradictory conclusions and flag for resolution
    • Specialized Judge Personas
      • Peer Review Judge: Simulates academic peer review criteria
      • Skeptic Judge: Actively looks for weak evidence and logical fallacies
      • Policy Analyst Judge: Evaluates for decision-making suitability
      • Technical Validator Judge: Checks code accuracy, technical claims
    • Fact-Verification Layer
      • Automated fact-checking against primary sources
      • Citation chain validation (does source actually support the claim?)
      • Statistical significance verification
  • Customize workflows — Modify the research pipeline to suit specific use cases
  • Build on top of this foundation — Create new features and applications

Stay tuned!

Explore More

Deep Research Agent was developed by Dr. Ng Chong at UNU Campus Computing Centre. It’s open, modular, and designed to make deep, unbiased research accessible to everyone.

Ready to flip between some channels and see what turns up?