Iterative Search Refinement With Model Directed Queries

via “multi-step reasoning search with iterative refinement”

AI search engine — direct answers with citations, Pro Search, Focus modes, research Spaces.

Unique: Implements explicit query decomposition and iterative refinement where the model generates its own follow-up searches based on intermediate results, rather than executing a single retrieval pass. This mirrors human research behavior (asking follow-up questions based on initial findings) and is architecturally distinct from single-pass RAG systems that retrieve once and generate once.

vs others: Outperforms single-pass search engines and basic RAG systems on complex research questions by dynamically identifying information gaps and filling them, whereas Google Search requires manual query reformulation and ChatGPT lacks real-time web access for iterative refinement.

via “deep-search-with-iterative-refinement”

Search the web and codebases to get precise, up-to-date context for programming and research. Find examples, API usage, and documentation from real repositories and sites to ship faster with fewer mistakes. Extend investigations with deep search, crawling, and business or profile lookups when needed

Unique: Supports search result caching and context preservation across multiple queries, allowing agents to reference previous findings when formulating follow-up searches. Enables stateful research workflows where each search builds on prior knowledge.

vs others: More effective than single-query search for complex research because it allows agents to refine understanding iteratively, similar to how human researchers conduct investigations by following leads and validating findings.

via “contextual query refinement”

Paste in my prompt to Claude Code with an embedded API key for accessing my public readonly SQL+vector database, and you have a state-of-the-art research tool over Hacker News, arXiv, LessWrong, and dozens of other high-quality public commons sites. Claude whips up the monster SQL queries that safel

Unique: Utilizes a dynamic feedback mechanism that adapts to user interactions, enhancing the relevance of search results through contextual understanding.

vs others: Offers a more interactive and adaptive search experience compared to static query systems that do not learn from user input.

via “iterative-search-refinement-with-model-directed-queries”

Tongyi DeepResearch is an agentic large language model developed by Tongyi Lab, with 30 billion total parameters activating only 3 billion per token. It's optimized for long-horizon, deep information-seeking tasks...

Unique: Implements a closed-loop search strategy where the model's reasoning directly controls search execution and evaluation, rather than treating search as a separate tool invoked once. The model maintains state across search iterations and makes explicit decisions about strategy pivoting, enabling adaptive research workflows.

vs others: More adaptive than static RAG systems that execute a single retrieval pass, and more transparent than black-box search ranking because the model's reasoning about search strategy is part of the output.

via “iterative-query-refinement-with-feedback-loops”

Sonar Deep Research is a research-focused model designed for multi-step retrieval, synthesis, and reasoning across complex topics. It autonomously searches, reads, and evaluates sources, refining its approach as it gathers...

Unique: Implements query refinement as an internal reasoning loop where the model evaluates search result quality and autonomously decides whether to reformulate, rather than exposing refinement as a user-facing interaction

vs others: More adaptive than single-pass search APIs; more autonomous than systems requiring explicit user feedback between search iterations

via “contextual query refinement”

MCP server: web-search

Unique: Incorporates a feedback loop that captures user interactions to continuously improve query suggestions, unlike static search engines.

vs others: Offers a more personalized search experience by learning from user behavior, which traditional search engines do not provide.

via “contextual query refinement”

MCP server: brave-search

Unique: Incorporates a feedback loop mechanism that allows the search engine to learn and adapt to user preferences over time.

vs others: More adaptive than traditional search engines, which often require manual query adjustments.

via “context-aware search query formulation”

GPT-4o Search Previewis a specialized model for web search in Chat Completions. It is trained to understand and execute web search queries.

Unique: Search query formulation is implicit and trained into the model weights rather than explicit (no separate query-generation step or function call); the model learns to recognize search-worthy intents from conversational context and reformulate queries for optimal retrieval during training.

vs others: More natural and context-aware than rule-based search triggers, but less transparent and debuggable than explicit query-generation agents with separate LLM calls for query refinement.

via “iterative refinement chat with context persistence”

Microsoft announces a new version of its search engine Bing, powered by a next-generation OpenAI model. Microsoft blog, February 7, 2023.

Unique: Treats search as a conversational experience rather than a stateless query-response model. Each turn re-executes the full search-and-synthesis pipeline with updated query intent, maintaining conversation context in the model's input rather than in a separate state store.

vs others: More natural than traditional search because users can refine queries through conversation rather than reformulating keywords, but slower than stateless search because each turn incurs full web indexing latency.

via “natural-language-query-refinement”

via “conversational-query-refinement”

via “query-refinement-and-correction”

via “conversational search and follow-up queries”