Search “Search”

Finout executive argues that AI agents need a clearly designated human owner and a team to which costs are attributed

Asaf Liveanu, co-founder and CPO of Finout, points out that an AI agent can start spending money with just a prompt and an API key, yet lacks a clear accountability process like that of an employee. He notes that a single agent task may call multiple model providers, search services, and SaaS tools, with costs varying according to the number of retries, loop duration, and model choice, and that two agents built from the same template may end the month with vastly different bills. To address this, he recommends designating a human owner and the team that bears the cost before the agent receives an API key, and using the enterprise's existing identity directory to map the logins, licenses, or API keys on the bill to specific people or agents and their teams. This claim is currently reported by CTech, and there are no subsequent developments of fact yet.

W3C releases discussion document on the Web and AI agents in the AI era

The W3C Strategy Team has released a discussion document, "W3C and the State of the Web in the AI Era," proposing the use of Web standards to make AI agents safer, with a focus on prompt injection protection, controllable permissions, and accessibility. The report states that the document was released on October 21 to build momentum for the TPAC meeting; it is a proposal rather than an approved standard, with an official presentation scheduled for October 29. Using travel booking as an example, the text points out that having an agent search for flights does not amount to authorizing it to pay for tickets, and recommends setting spending limits, revocable authorization, and traceable credentials for payments. The document is currently still in the discussion stage, with no further progress yet.

Nvidia adds Agent Knowledge Foundation to Agent Toolkit

At GTC Berlin, Nvidia added the Agent Knowledge Foundation to its Agent Toolkit. The component packages the Nemotron Retriever models, a new Auto-Ontology library, the Kumo family of models for structured data, and CUDA-X data processing libraries; data remains in place and is governed by enterprises' existing permissions. Nvidia says Nemotron Retriever improved search relevance by 65% in tests in the open-source OpenSearch environment, SAP's fine-tuned reranker runs nearly three times faster, and ServiceNow's reranking offers better quality with nearly 25% lower latency. Auto-Ontology provides a REST API and an MCP server for agents to query. Nvidia says it does not compete with data platforms such as Snowflake, but partners with them. The news was reported by The New Stack; there are no subsequent developments yet.

YGO completes €8 million Series A funding round

YGO has completed an €8 million Series A funding round. According to a Tech.eu report on October 21, 2026, the round was led by Ventech, with participation from Sake Bosch, Caesar Capital, and angel investors. YGO's AI infrastructure is used to connect travel companies' existing systems with AI agents, enabling flight, hotel, and activity inventory to be searched, recommended, and booked directly by AI. The report states that YGO already processes bookings for European travel companies such as Condor, Bentour, and GetAway Group, and that one airline customer reduced manual operations by 90% and increased conversion rates by 20%. There are currently no subsequent developments.

Google launches Personal Intelligence for Android Auto

Google launches Gemini Personal Intelligence for Android Auto. According to a t3n report on October 21, 2026, the feature can access data from Gmail, Google Calendar, Google Photos, Google Maps, YouTube, and Search—for example, to look up a flight's boarding gate or set navigation to a location from a photo. The report says the feature is rolling out in all regions except the EU, Switzerland, the UK, and Nigeria, and it is not yet clear when it will be available to users in Germany. Personal Intelligence is off by default and requires authorization in the Gemini app before it can access these Google services. There are currently no subsequent developments.

Cloudflare launches Cloudflare OS managed service

Cloudflare has announced a managed service for its AI business platform Cloudflare OS, with a waitlist now open for registration. Cloudflare OS was open-sourced in August this year and can be deployed in Cloudflare environments, on-premises, or in the cloud. It supports integration with Cloudflare Workers AI as well as AI services from Anthropic, OpenAI, Google, and others, and has the ability to securely access enterprises' internal data. This release also adds two new features: first, connecting GitHub repositories, allowing AI agents to add features, debug, view PRs, create commits, and push code based on repository code; second, enhanced Google Workspace integration, enabling agents to read and search Gmail conversations, draft and send emails, and connect to Google Drive, Docs, and Sheets.

Google says it may use Gmail and Photos data for targeted ads

Google said it may use data from Gmail and Google Photos for targeted advertising. On October 24, 2026, The Verge AI reported that Google updated its policy, stating that users who connect Drive, Gmail, or Google Photos to search tools such as AI Mode through the Personal Intelligence feature may have their content used to serve targeted ads; Google also said that advertisers will not receive this data or identity information such as names and email addresses, and that ads will not be targeted based on sensitive categories such as race, sexual orientation, religion, or health, and that users can disconnect the relevant apps on the search service's personalization settings page or completely turn off personalized recommendations. On October 26, heise online reported that Google added a new exception clause to its privacy policy, allowing personalized ads to be served based on content from Gmail, Google Drive, or Google Photos when users make that data available to services that support personalized advertising, such as Search and Gemini; users can turn off personalized recommendations in search services in the account's "Data & privacy" settings, but data already collected will not be deleted as a result. There is currently no further development with evidence.

Rauch says AI Gateway itself is a tool marketplace

On October 23, 2026, Guillermo Rauch reposted Vercel's post introducing TakoViz's launch on AI Gateway, calling TakoViz one of the coolest real-world examples of generative UI. He also said he likes that AI Gateway itself is a tool marketplace: adding a tool in aisdk takes just 1 line of code, and you get "Google"—not just search, but also weather, sports, stocks, elections, and more. This statement currently comes from Rauch's own public remarks, with no subsequent developments yet.

IskayPet builds enterprise knowledge assistant based on Amazon Bedrock

IskayPet has built an enterprise knowledge assistant based on Amazon Bedrock Managed Knowledge Bases, covering retail operations, veterinary care, HR onboarding, and procurement scenarios, supporting four languages and three countries, with no need to self-manage RAG infrastructure. The assistant directly ingests existing documents through the native SharePoint connector, combining hybrid search of semantic and BM25 keyword search, managed reranking, and document-level permission control, and returns answers with source citations. The entire solution, from store front end to AWS architecture deployment, was completed by Kiro with Claude Haiku 4.5. There are currently no subsequent development facts.

MiniMax releases the M3.1 Flash model

On October 19, 2026, MiniMax released the M3.1 Flash model. The company says it costs one-third as much as M3 and is 3 times faster, with a unified API price of $0.15/$0.60 per 1M input/output tokens and cached reads at $0.015 per 1M tokens; it had previously been tested anonymously on OpenRouter under the name Space Bunny, ranking first on the model leaderboard for 10 consecutive days. On October 20, the model launched on platforms including Ollama, GMI Cloud, Cline, and OpenClaw, with Cline offering it for free; third-party testing compared it with Claude Opus 5.5, and both completed a Three.js task on the first attempt. On October 21, Parallel Web Systems' search capability efficiency leaderboard ranked it first. On October 22, MiniMax announced that Inferact had become its API launch partner, and the model launched on Runware. On October 23, MiniMax said SGLang and Miles provided support on the first day of release.

Chelsea Finn proposes optimizing information organization to extend LLM context

On October 22, 2026, Chelsea Finn reposted the new paper Librarian Agents by Anikait Singh et al., and proposed optimizing information organization to improve downstream utility. She pointed out that agentic search can scale to information of arbitrary size but tends to miss relevant content; whereas optimizing information organization for downstream use can bring 30+% improvements even in OOD domains. The tweet includes links to the paper and code. There are currently no subsequent development facts.

Microsoft researchers prove the KLS conjecture

Microsoft Research researcher Xinzhi Zhang and co-authors proved the KLS conjecture proposed in 1995, giving a dimension-independent fixed bound; the previous best result only allowed the bound to grow slowly with dimension. According to Microsoft Research, the proof came from a human-guided agent search process, in which the team explored more than 100 candidate paths, which researchers evaluated, pruned, and redirected. An independent team also gave a complete proof around the same time, followed by Lean 4 formal verification. The authors state that previous unguided end-to-end attempts with multiple frontier models failed to achieve the goal. Another report states that AI agents explored about 100 research paths, the draft was about 140 pages, AI was responsible for exploring candidate methods, executing derivations, assisting in writing the manuscript, and performing internal proof checks, while the recursive method and error bounds that completed the proof are credited to the two researchers.

Cohere Releases Enterprise Retrieval Model Rerank 5

On October 21, 2026, Cohere released its enterprise retrieval model Rerank 5, available in two tiers: Pro and Fast. According to Cohere's product and research blog, the model is now available on the Cohere API, Model Vault, Microsoft Foundry, and Amazon SageMaker, with Pro priced at $0.05 per 1 million tokens and Fast at $0.02. Cohere states that Rerank 5 Pro leads all tested models on four public benchmarks—ViDoRe V3, nanoMTEB, BRIGHT, and BEIR—ranking first on 39 of 46 composite datasets. A new feature, Context Focus, can return the most relevant text snippets within a document in a single call; in official tests, removing half the context cost only 2 percentage points in answer quality. Follow-up reports add that both tiers support a 32K context window and more than 100 languages, and that upgrading requires no index rebuild; in search agent tests, paired with a default threshold of 0.1, it reduced tokens per task by 19% and raised accuracy from 46% to 57%.

English translation is being prepared.

Researchers at the University of Melbourne propose that generative AI is distorting archaeology and flattening Indigenous knowledge

Researchers at the University of Melbourne argue in an article on AIhub that generative AI is distorting archaeology and flattening Indigenous knowledge. The article states that GenAI scrapes and reuses Indigenous cultural and intellectual property (ICIP) on a large scale without consent, compressing local Indigenous knowledge tied to specific lands and communities into statistical patterns, thereby erasing its context. The authors say their recent research found that AI can often distinguish science from Indigenous consensus and long-circulating fringe narratives, but that this capability itself relies on training data collected or reproduced without community permission. The article also mentions that among the top results of a Google image search for "cultural heritage Australia" is a GenAI-fabricated image of a nonexistent giant stone carved head circulating. This judgment is currently reported by a single source, with no subsequent factual developments.

Voxel51 releases version 3.0 data platform

On October 21, 2026, Voxel51 released version 3.0 of its data platform, targeting robotics and physical AI scenarios. According to Robotics & Automation News, the platform integrates the search, organization, annotation, and evaluation of robot multi-camera, video, and sensor data into a single platform, and adds a new AI assistant, Voxel51 Agent, that can execute tasks across the development workflow. The platform supports semantic search, text queries, and metadata filtering, as well as synchronized viewing of multimodal time series, and provides 2D, 3D, and video annotation and Agentic Labeling automatic annotation. The report also states that Voxel51 has partnered with Nebius to run its workflows on Nebius AI Cloud, and offers a free self-service version. There are currently no subsequent development facts.

OpenAI launches locked chats feature for ChatGPT

OpenAI launches a locked chats feature for ChatGPT. According to the official description, this feature moves conversations out of regular history and search and into a protected space; on web and desktop it is unlocked with a PIN, and on supported mobile devices it can be unlocked with a PIN, Face ID, Touch ID, or device unlock; new messages in locked conversations do not update the memory used by regular conversations. The feature is rolling out starting October 21, 2026, to the Free, Go, Plus, Pro, Business, Enterprise, and Edu plans. Follow-up reports add that after users set a 4-digit PIN, they can hide designated conversations from the history list, pinned section, and ordinary search results, tucking them into a "Locked chats" section in the sidebar, covering web, iOS, Android, and the new macOS and Windows desktop apps; locking does not provide end-to-end encryption, nor does it change data retention and model improvement settings, and locked chats may still draw on memory already saved in regular chats; forgetting the PIN itself does not delete chats, but choosing to reset the PIN and confirming will permanently delete all locked chats in the account, memories saved only in locked chats, and files protected only by the old PIN, with no way to recover them.

English translation is being prepared.

Databricks and NVIDIA Bring RFD4-Proteina to Genesis Workbench

Databricks and NVIDIA plan to integrate NVIDIA BioNeMo's open-source protein design model RFD4-Proteina into Databricks' Genesis Workbench. According to a Databricks blog post dated October 21, 2026, the model will run on the generally available AI Runtime serverless NVIDIA GPU compute. RFD4-Proteina can simultaneously generate protein sequences and all-atom structures, and supports optional reward-guided search that steers candidate designs toward target properties such as thermostability and binding affinity. Genesis Workbench includes two built-in RFD4-Proteina workflows, Antibody Design and Vaccine Design, and can be extended through Genie Code and NVIDIA BioNeMo Agent Toolkit skills. No further progress has been disclosed at this time.

DeepSeek launches Search and Fetch APIs

On October 17, 2026, DeepSeek launched two APIs, Search and Fetch. According to TMTPost, these two interfaces are driven by a self-developed search engine, charging 5 yuan per 1,000 successful requests, with failed, no-result, or timed-out requests not billed; official billing starts on November 1; previously they were free to use in DeepSeek Harness, Claude Code, and Codex, each rate-limited to 5 requests per second. On October 18, DeepSeek issued an announcement stating that the DeepSeek API now supports web search and scraping, billed by number of requests at a unit price of 5 yuan per 1,000 successful requests; accessing the Search and Fetch interfaces via server-side tool calls in DeepSeek Harness, Claude Code, and Codex is currently not billed, and will be charged per request starting at 00:00 Beijing time on November 1, 2026. The announcement also explained that the Search interface returns structured results by query, and the Fetch interface scrapes a single web page and returns the title and body in Markdown; both are raw JSON APIs with the Base URL api.deepseek.com. Currently both interfaces are live, and billing has not yet started.