Functionality

The Deep Research Tool is an advanced AI-powered research assistant, or what is also referred to as a https://blog.langchain.com/deep-agents/ , that performs in-depth analysis and investigation across multiple sources to answer complex research questions that require longer and deeper analysis. This is usually analysis jobs running 10-15 minutes and reviewing 100+ sources

Use Cases

The Deep Research Tool is ideal for:

1. Complex Research Questions: Questions requiring multiple steps and synthesis from multiple sources

   • Example: "What are the current trends in AI safety research and how do different organizations approach it?"

2. Competitive Analysis: Gathering information about competitors, market trends, or industry analysis

   • Example: "Compare the product strategies of the top 5 CRM vendors"

3. Literature Review: Summarizing research papers, articles, or documentation on a topic

   • Example: "Provide an overview of recent advances in quantum computing"

4. Policy Research: Understanding regulations, compliance requirements, or internal policies

   • Example: "What are the GDPR requirements for data processing and how do they apply to our use case?"

5. Technical Investigation: Deep dives into technical topics requiring multiple sources

   • Example: "Explain the architecture and tradeoffs of different vector database implementations"

6. Market Research: Analyzing market conditions, customer sentiment, or industry trends

   • Example: "What are enterprises looking for in observability platforms in 2025?"

Not recommended for:

• Simple factual queries (use standard chat)

• Real-time information updates (though web search provides current data)

• Questions answerable from a single source

Research Engines

The deep research tool requires an “Engine” that powers the research and has access to all the tools and capabilities required for the research. The “Engine” is the underlying agent that guides the agent’s decision making. Unique provides two Deep Research engines:

• Unique Engine - A custom multi-agent research engine that orchestrates multiple specialized agents

• OpenAI Engine - The OpenAI native deep research capability exposed by OpenAI via the Responses API. This engine requires access to OpenAI models hosted directly on OpenAI servers (e.g., litellm:openai-gpt-5)

Unique Engine (Default)

A custom multi-agent research engine that orchestrates multiple specialized agents. Users can leverage external and internal data sources to generate comprehensive answers to their research questions. The Unique Deep Research Engine provides flexible model selection allowing admins to carefully control their to data security and privacy. For more details on how it works see https://unique-ch.atlassian.net/wiki/spaces/sdv2/pages/1496678539/Deep+Research#Unique-deep-research-engine-architecture

Available Tools:

• Web Tools: Search the web and fetch content from URLs with automatic citations (Google Engine)

• Internal Tools: Search company knowledge base and fetch internal documents including looking at uploaded documents. Please not that any space Smart Rules apply to the deep research agents search capabilities

• MCP (In progress): Factset, Outlook, and custom servers

• Connectors (In progress): Quartr, SIX, and custom connectors

• Code Execution (In progress): Run code in a secure container to perform discrete calculations, generate charts, graphs and data files.

Deep research tools can be enabled and disabled, but they are currently not configurable.

Also they are not affected by the other tools configured in the same space. Having Internal Search and Web Search (old - DO NOT PUBLISH) enabled or disabled does not affect the deep research tools

The Unique engine is limited to 300 tool calls and at most 25 different research directions, allowing for thorough multi-step investigation, but keeping token usage limited. This is not configurable in the space configuration, and the values will likely evolve with the tool

To see the configurable fields for Deep Research see the configuration section

OpenAI Engine

The OpenAI Engine connects to the deep research capability exposed by OpenAI via the Responses API. This engine requires access to OpenAI models hosted directly on OpenAI servers (e.g., litellm:openai-gpt-5,litellm:openai-o4-mini-deep-research). Users will not have access to internal data sources or control over model hosting location when choosing this option

Available Tools:

• Web Search: Uses OpenAI's native web_search_preview tool

The OpenAI engine provides a simpler, single-agent research approach with built-in web search capabilities. Unique does not have control over OpenAI’s access to third party data sources or the ability to restrict how OpenAI searches the web given a user query.

Performance and usage cost

Expected token consumption for the deep research can vary drastically depending on the model, tool configuration, and question asked. Unique provides no guarantee on the numbers below. Numbers are based on Azure pricing information from 07/11/2025

Unique's deep research is a type of agent which commonly is referred to as a deep agent. This differ from shallow agents by delegating complex research tasks to specialized sub-agents with explicit planning and memory, rather than executing everything in a single reactive loop. The power of the system comes from handing over full control to the agent and not forcing the research direction or strict limits on what has to be done.

Below is provided a reference for expected token consumption of deep research in different configurations.

We have also provided the RACE scores known from https://deepresearch-bench.github.io/ as an evaluation of performance. This is a score between 0-1 where the higher is better and 0.5 is equivalent to a human expert written report judged by an LLM. Googles Gemini-2.5-pro deep research obtains a 0.4971 and OpenAI’s 0.4645

Unique recommends the blue highlighted entries for Unique AI
[1] OpenAI reports GPT-5 as their leading Deep Research model although a RACE score has not been published.

[2] https://arxiv.org/pdf/2512.01948v1#page=15&zoom=100,94,393

Configuring Deep Research

Documentation is only supplied for the Unique AI Space configuration, not Unique Custom Space

Deep research is a tool available for Unique AI Chat and will, once enabled, appear in the tool section of Unique AI Space

The tool must be set to with is exclusive checked to achieve the desired behavior of the agent. Other than this, the tool follows the normal tool patterns Tools

The tool has the following configurable fields

Additional configurations available for Unique Engine:

Feature flags and environment variables

Deep research requires the enablement of 2 feature flags and has additional system controls to limit the amount of runs that can be executed at any time

Unique deep research agent architecture

Overview

The Deep Research system employs a multi-agent architecture with nested feedback loops to conduct comprehensive research autonomously. The design mimics how human research teams operate: a lead researcher coordinates the investigation while specialized team members focus on specific topics, then findings are synthesized into a cohesive final report. This is based on the work done by Anthropic https://www.anthropic.com/engineering/built-multi-agent-research-system?utm_source=alphasignal and LangChain https://blog.langchain.com/open-deep-research/

Architecture

The system begins with an optional clarification phase where the agent asks follow-up questions to better understand the user's request. Once clarity is established, a comprehensive research brief is generated.

The core research process operates through two nested agent loops:

Lead Supervisor Agent: A lead supervisor agent continuously analyzes research progress and makes strategic decisions. In each iteration, it can choose to (1) use internal reasoning to refine its strategy, (2) spawn new research sub-agents to investigate specific topics, or (3) declare research complete. This iterative decision-making allows the supervisor to adaptively identify knowledge gaps and delegate follow-up research across multiple iterations.

Research Sub-Agent: Each research sub-agent operates independently in its own loop, focusing on a specific assigned topic. Sub-agents execute tools to search the web, fetch web pages, search internal knowledge bases, and retrieve documents. After each tool execution, the agent decides whether to continue investigating or conclude its research. Upon completion, findings are compressed into a concise summary with proper citations and returned to the supervisor.

The critical feedback mechanism enables the supervisor to receive findings from completed sub-agents, incorporate them into its knowledge state, and spawn additional researchers if gaps remain. Multiple sub-agents can execute in parallel (up to a configured limit), enabling efficient multi-topic research.

Finally, a final report generation phase synthesizes all accumulated findings from multiple research iterations into a comprehensive, well-formatted report with validated citations.

This architecture ensures thorough investigation through iterative refinement while maintaining computational efficiency through parallel execution and intelligent task delegation.

The system process design is as depicted below

User flow

1. User actives the Deep Research tool and asks a question. Deep research can never be activated by the agent

2. Agent replies with follow-up questions to clarify the research scope

3. The user provides additional information to guide the research process

4. Deep Research job is queued and waits to start

   

5. Deep Research starts and “Steps” begin to populate, detailing the agents research process. The user can open the Steps panel to follow the agent’s chain of thought and actions

6. Research completes presenting the user with the final output