AI搜索agent可以在演示中工作，但当检索返回不完整证据，重复不成功的行动，或执行超过时间和成本限制时，在生产工作量下失败。 为AI agents提供网络基础设施的TinyFish已经出版了两份应对这些挑战的技术指南。 第一个定义了生产代理搜索架构，而第二个则分享了建设深层研究系统的经验教训。 他们共同研究自主搜索agent如何规划检索，评价证据，处理故障，管理模型推理与决定性执行之间的界限。
TL；数据
TinyFish 推出其代理搜索架构和从构建深层研究系统中吸取的关键工程教训。 该系统建于四个网络原始体(TinySearch,TinyFetch,TinyBrowser，和TinyAgent)上，通过基模型周围的七相控制循环运行。 在OpenBenchmarks上，TinyFish在提供最低中位任务成本(0.050美元)和象征性足迹的同时，实现了79.0%的完成率，凸显了硬性控制循环，故障特定重试，以及统一推理而不是独立规划者的重要性。
什么是代理搜索架构？
代理搜索架构使AI agent能够规划检索，访问来源，评价证据，并随着新信息的出现而调整其方法。 该模型提供推理，而周围的应用程序管理状态，工具执行，证据处理，运行时间政策和输出验证。 这些组件共同使agent能够处理不完全的结果，可收回的故障，并改变信息要求。
TinyFish的架构从2026年2月的架构预印中抽取了基础概念，通过打字工具接口将推理与执行脱钩。 该文件建立了必要的系统护栏，包括详细的追踪，同时对信使、执行时间、总成本和工具引用设置严格的上限。
生产搜索是如何运作的？
小费分成七个阶段：
- 确定任务：将目标变成“证据合同”：必要的领域、来源、新鲜度和完成规则。
- 计划检索：将独立需求与依赖需求分开。 依赖性查询等待更早的答案。
- 选择第一个路线：根据系统对证据的了解选择一个工具。
- 发现、选择、检索：在页面内容到达之前将搜索结果作为候选人对待。
- 评估：标签证据充足、不完整、相互冲突、无法获取或预算用尽。
- 恢复或升级：修改查询、重试、更改来源或更改访问方法。
- 停止和验证：返回完成、不完整、失败、取消或需要人审查。
第二阶段处理已知的失败。 ChainRAG(ACL 2025)称其为"丢失的检索"问题，关键实体在多跳问题分解过程中消失。 第六阶段借鉴了Search-R1，它展示了模型在早期检索不足时如何在推理过程中学习生成额外的搜索查询。 在7个QA数据集中，其所学的多回合搜索比RAG基线高出41%(Qwen2.5-7B)和20%(Qwen2.5-3B).
搜索 agent 路由工具应该如何？
排队应该跟随信息差距，而不是固定的梯子。
升级并不总是固定的。 弱的结果需要更好的查询。 冲突来源需要更有力的证据。 一个更重的工具也不会修正。
检索应以故障类型为基础，并保持在规定的限度内。 短暂的错误可能证明有理由重复请求，而不完整或相互冲突的证据则要求采取不同的做法。 浏览器重试还必须说明为避免意外后果而已完成的行动，例如两次提交相同的表格。
TinyFish学了什么深层研究？
深入研究agent搜索，浏览，并生成全面，引用的报告。 大多数工程挑战围绕背景管理，系统管弦，以及产品设计。
第1课：删除单独的规划员
初始迭代使用独立规划器，每次称为新鲜。 只有结构化的州进入它。 在诸如“只关注东南亚的定价风险”等后续询问中，agent重新回到广泛的市场研究。 TinyFish后来将计划转移到了管弦乐团自己的统一推理圈中。
课 #2：删除子剂：
TinyFish最初为每个搜索查询指定了单独的副代理，增加了上下文切换的间接费用，空闲性和成本。 团队后来用工具调用这些副剂替换了可重复的任务，如排序搜索结果，分类来源，提取证据。 这减少了传递给管弦乐团的信息，使其得以专注于研究规划，并决定下一步要调查什么。
第3课——匹配研究者：
TinyFish最初优化了源覆盖的研究系统，但用户更重视支持证据的质量和相关性。 该小组调整了源的排名，以说明研究目标，并增加了对优先排序、排除或限制域的控制。 它还利用LLM法官开发了内部评价工具，以评估报告的相关性、证据强度、引用信任和来源多样性。
成本和耐性如何加起来？
TinyFish将全部任务预算
-任务延迟=规划+搜索+检索+评价+重复+互动步骤+排队。
- 任务成本 = 符号 + 工具调用 + 页面或浏览器使用 + 重复 + 基础设施。
Search Agent-X(ICLR 2026) 显示为什么。 精确的检索增加了间接费用，粗糙的检索也迫使额外推理。 它的修补达到3.4x吞吐量和5x低于测试基线。
TinyFish如何比较搜索和获取？
Open Benchmarks)修正模型(gpt-5.6-sol),100项任务和预算，并且只不同提供者。 得分是3跑的手段。
资料来源：开放奔驰板，2026年9月26日更新。 任务只涵盖企业文件。
Open Benchmarks也分别报告只搜索结果。 在该配置中，TinyFish完成了59.3%的任务，Perplexity为77.3%,Firecrawl为70.3%,Exa Fast为66.3%. 由于现有检索能力不同，不应直接将这些结果与搜索和查询分数进行比较。
关键外卖
- 控制整个轨道，而不是一个搜索电话。
- 按信息差距划分路线；只有在访问失败时才升级。
- 继续规划和执行在一个推理循环。
- 将决定性步骤移到工具呼叫中，而不是潜剂中。
- 按研究目标排列的排名来源，而不是一个质量得分。
读"建筑指南","深度研究课"和"小费"文献。
感谢小鱼团队为这篇文章提供的思想领导/资源。 这篇文章由小菲赞助。
Asif Razzaq是Marktechpost AI Media Inc.的首席执行官。 作为一个有远见的企业家和工程师，阿西夫致力于利用人工智能的潜力促进社会福祉。 他最近的努力是人工智能媒体平台的发布，该平台以深入报道机器学习和深层学习新闻而著称，这些新闻在技术上既可靠又为广大观众所容易理解。 该平台每月有超过200万次观看，表明其受观众欢迎。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
An AI search agent can work in a demo but fail under production workloads when retrieval returns incomplete evidence, retries repeat unsuccessful actions, or execution exceeds time and cost limits. TinyFish, which provides web infrastructure for AI agents, has published two technical guides addressing these challenges. The first defines a production agentic search architecture, while the second shares lessons from building a deep research system. Together, they examine how an autonomous search agent can plan retrieval, evaluate evidence, handle failures, and manage the boundary between model reasoning and deterministic execution.
TL;DR
TinyFish introduced its agentic search architecture and key engineering lessons from building deep research systems. Built on four web primitives (TinySearch, TinyFetch, TinyBrowser, and TinyAgent), the system operates via a seven-phase control loop around the base model. On OpenBenchmarks, TinyFish achieved a 79.0% completion rate on coding tasks while delivering the lowest median task cost ($0.050) and token footprint, highlighting the importance of rigid control loops, failure-specific retries, and unified reasoning over standalone planners.
What is Agentic Search Architecture?
Agentic search architecture enables an AI agent to plan retrieval, access sources, evaluate evidence, and adapt its approach as new information becomes available. The model provides reasoning, while the surrounding application manages state, tool execution, evidence handling, runtime policies, and output validation. Together, these components allow an agent to handle incomplete results, recoverable failures, and changing information requirements.
TinyFish’s architecture draws foundational concepts from a February 2026 architecture preprint, which decouples reasoning from execution through typed tool interfaces. The paper establishes essential system guardrails, including detailed tracing alongside strict caps on tokens, execution time, overall cost, and tool invocations.
How does a production search run work?
TinyFish splits a run into 7 phases:
- Define the task: Turn the goal into an “evidence contract”: required fields, sources, freshness and completion rules.
- Plan retrieval: Separate independent needs from dependent ones. Dependent queries wait for earlier answers.
- Pick the first route: Choose a tool based on what the system knows about the evidence.
- Discover, select, retrieve: Treat search results as candidates until the page content arrives.
- Evaluate: Label evidence as sufficient, incomplete, conflicting, inaccessible or budget exhausted.
- Recover or escalate: Revise the query, retry, change sources or change access method.
- Stop and validate: Return completed, incomplete, failed, cancelled or human review required.
Phase 2 addresses a known failure. ChainRAG (ACL 2025) calls it the ‘lost-in-retrieval’ problem, where key entities disappear during multi-hop question decomposition. Phase 6 draws on Search-R1, which demonstrates how models can learn to generate additional search queries during reasoning when earlier retrieval is insufficient. Across 7 QA datasets, its learned multi-turn search beat RAG baselines by 41% (Qwen2.5-7B) and 20% (Qwen2.5-3B).
How should a search agent route tools?
Routing should follow the information gap, not a fixed ladder.
Escalation is not always the fix. Weak results need a better query. Conflicting sources need stronger evidence. A heavier tool fixes neither.
Retries should be based on the failure type and remain within defined limits. Transient errors may justify repeating a request, while incomplete or conflicting evidence requires a different approach. Browser retries must also account for actions already completed to avoid unintended consequences, such as submitting the same form twice.
What did TinyFish learn building deep research?
The deep research agent searches, browses, and generates comprehensive, cited reports. Most engineering challenges centered around context management, system orchestration, and product design.
Lesson #1: Removing the separate planner
Initial iterations used a standalone planner called fresh each time. Only structured state crossed into it. On follow-up queries like “focus only on pricing risks in Southeast Asia,” the agent drifted back to broad market research. TinyFish subsequently moved planning into the orchestrator’s own unified reasoning loop.
Lesson #2: Removing subagents:
TinyFish initially assigned a separate subagent to each search query, increasing context-switching overhead, latency, and cost. The team later replaced these subagents with tool calls for repeatable tasks such as ranking search results, classifying sources, and extracting evidence. This reduced the information passed back to the orchestrator, allowing it to focus on research planning and deciding what to investigate next.
Lesson #3- Matching researchers:
TinyFish initially optimized its research system for source coverage, but users placed greater importance on the quality and relevance of supporting evidence. The team adjusted source ranking to account for the research objective and added controls for prioritizing, excluding, or restricting domains. It also developed an internal evaluation harness using an LLM judge to assess reports for relevance, evidence strength, citation trust, and source diversity.
How do cost and latency add up?
TinyFish budgets the full task, not single calls:
- Task latency = planning + search + retrieval + evaluation + retries + interactive steps + queueing.
- Task cost = tokens + tool calls + page or browser usage + retries + infrastructure.
SearchAgent-X (ICLR 2026) shows why. Exact retrieval added overhead, and coarse retrieval forced extra reasoning. Its fixes reached up to 3.4x throughput and 5x lower latency versus tested baselines.
How does TinyFish compare on search and fetch?
OpenBenchmarks fixes the model (gpt-5.6-sol), 100 tasks and budgets, and varies only the provider. Scores are means of 3 runs.
Source: OpenBenchmarks board, updated 26 Sept 2026. Tasks cover enterprise documentation only.
OpenBenchmarks also reports search-only results separately. In that configuration, TinyFish achieved 59.3% task completion, compared with 77.3% for Perplexity, 70.3% for Firecrawl, and 66.3% for Exa Fast. These results should not be directly compared with the search-and-fetch scores because the available retrieval capabilities differ.
Key Takeaways
- Control the whole trajectory, not 1 search call.
- Route by information gap; escalate only when access fails.
- Keep planning and execution in 1 reasoning loop.
- Move deterministic steps into tool calls, not subagents.
- Rank sources by research objective, not one quality score.
Read the architecture guide, the deep research lessons and the TinyFish docs.
Thanks to the Tinyfish team for the thought leadership / resources for this article. This article is sponsored by TinyFish.
Asif Razzaq is the CEO of Marktechpost AI Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.