I never thought I’d find a Tatiana Maslany role I enjoyed as much as Orphan Black, yet here we are. Maximum Pleasure Guaranteed is Apple TV’s latest thriller, following on from recent huge releases like Margo’s Got Money Troubles and Widow’s Bay, and it’s another fantastic release from the streaming service.

Here, Maslany plays newly divorced mother Paula, who has been spending time with Trevor (Brandon Flynn), a camboy. Since Paula’s ex-husband has main custody of their daughter, she spends lonely nights talking with him, which inevitably leads to more.

When agentic workflows fail, developers often assume the problem lies in the underlying model’s reasoning abilities. In reality, the limited information provided by the retrieval interface is often the primary limiting factor.

Researchers at multiple universities propose a technique called direct corpus interaction (DCI) that lets agents bypass embedding models entirely, searching raw corpora directly using standard command-line tools.

The limits of classic retrieval

In classic retrieval systems such as RAG, documents are chunked, converted into vector representations (or embeddings), and indexed offline in a vector database. When an AI system processes a query, a retriever filters the entire database to return a ranked “top-k” list of document snippets that match the query. All evidence must pass through this scoring mechanism before any downstream reasoning occurs.

But modern agentic applications demand much more. “Dense retrieval is very useful for broad semantic recall, but when an agent has to solve a multi-step task, it often needs to search for exact strings, numbers, versions, error codes, file paths, or sparse combinations of clues,” the authors of the DCI paper said in comments provided to VentureBeat. “These long-tail details are precisely where semantic similarity can be brittle.”

Unlike static search, agents must also revise their search plans dynamically after observing partial or localized evidence. Exact lexical constraints and multi-step hypothesis refinement are difficult to execute with semantic retrievers. Because the retriever compresses access into a single step, any critical evidence filtered out by the similarity search cannot be recovered later, no matter how advanced the agent’s downstream reasoning capabilities are. As the authors explain, current retrieval pipelines can become a bottleneck because “they decide too early what the agent is allowed to see.”

Direct corpus interaction

This direct access addresses a core problem in enterprise environments: data staleness. Embedding indexes are always a snapshot of a specific moment in time, taking considerable compute and time to build and maintain.

“In many enterprise settings, the data is not a stable document collection. It is daily financial reports, live logs, tickets, code commits, configuration files, incident timelines, and internal documents that keep changing,” the authors said. DCI lets the agent reason over the current state of the workspace rather than yesterday’s vector index.

The agent operates in a terminal-like environment where its observations are raw tool outputs such as file paths, matched text spans, and surrounding lines. The core tools provided by DCI are few but highly expressive. Agents use commands like “find” and “glob” to navigate directory structures and locate files. For exact matching, they use “grep” and “rg” to locate specific keywords, regex patterns, and exact strings. When local inspection is needed, tools like “head,” “tail,” “sed,” “cat,” and lightweight Python scripts allow the agent to peek at the context surrounding a match or read specific file sections.

The agent can combine these tools via shell pipelines to execute complex search logic in a single step. An agent can pipe commands to enforce strict lexical constraints, such as searching a file for one term and piping the output to search for a second term. It can combine multiple weak clues across a corpus by finding a specific file type, searching for a keyword like “report,” and filtering for a year like “2024.” It can also immediately verify a hypothesis by inspecting the exact lines around a keyword match.

DCI delegates semantic interpretation directly to the agent instead of relying on embedding-based similarity search. The agent can formulate hypotheses, test exact lexical patterns, and extract detailed information that a traditional semantic retriever might miss.

The researchers propose two versions of this system. DCI-Agent-Lite is designed as a lightweight, low-cost setup built on the GPT-5.4 nano model and restricted purely to raw terminal interactions like bash commands and basic file reads. Because reading raw files can quickly fill up a smaller model’s memory, this version relies on lightweight runtime context-management strategies to sustain long-horizon exploration.

DCI-Agent-CC is the higher-performance version, designed for teams with more compute budget. It runs on Claude Code powered by Claude Sonnet 4.6. Claude Code provides stronger prompting, more robust tool orchestration, and superior built-in context handling, which improves the agent’s stability during complex, multi-step searches across heterogeneous datasets.

DCI in action

The researchers tested both versions of DCI across agentic search benchmarks like BrowseComp-Plus, knowledge-intensive QA with single-hop and multi-hop reasoning, and information retrieval ranking in tasks requiring domain-specific reasoning and scientific fact-checking.

They tested DCI against three baselines. The first included open-weight retrieval agents such as Search-R1 and proprietary agents powered by frontier models like GPT-5 and Claude Sonnet 4.6, paired with standard retrievers. The second baseline included classical sparse retrievers like BM25 and dense retrievers like OpenAI’s text-embedding-3-large and Qwen3-Embedding-8B. The third baseline consisted of high-performing reasoning-oriented re-rankers like ReasonRank-32B and Rank-R1.

DCI systematically outperformed the baselines, according to the researchers. On the complex BrowseComp-Plus benchmark, swapping a traditional Qwen3 semantic retriever for DCI on a Claude Sonnet 4.6 backbone improved accuracy from 69.0% to 80.0% while reducing the API cost from $1,440 to $1,016. The return on investment for lightweight agents was also noticeable. DCI-Agent-Lite with GPT-5.4 nano competed with the OpenAI o3 model using traditional retrieval while cutting costs by more than $600.

On multi-hop QA benchmarks, DCI-Agent-CC reached an 83.0% average accuracy, improving on the strongest open-weight retrieval baseline by 30.7 points, according to the researchers.

The data shows that DCI has lower overall document recall than dense embedding models, but once it finds a relevant document, it extracts substantially more value from it.

“If an enterprise AI lead asked where DCI is most clearly useful, I would point to tasks that require exact evidence localization in a dynamic workspace: debugging production incidents, searching large codebases, analyzing logs, compliance investigation, audit trails, or multi-document root-cause analysis,” the researchers note.

In one complex deep-research task, the agent had to identify a specific soccer match based on 12 interlocking clues, including exact attendance, yellow cards, and player birth dates. A traditional retriever would fail by surfacing short, disconnected snippets. Instead, the DCI agent explored the file directory, read specific lines of a 1990 England versus Belgium match report to verify the exact number of substitutions, pulled a specific quote from an interview file, and verified the exact birth dates of two players by peeking into their Wikipedia text files. By chaining these simple commands, DCI ensures that no evidence is permanently lost behind a flawed semantic search algorithm.

Limits and practical implementation of DCI

DCI has a clear operating envelope where it scales excellently in search depth but struggles with search breadth. When the experimental corpus was expanded from 100,000 to 400,000 documents, the system’s accuracy dropped significantly and the average number of tool calls rose. While DCI is powerful once a promising document is found, the cost of locating that initial useful anchor document grows sharply as the size of the candidate space increases.

DCI also has lower broad document recall compared to dense embedding models. It trades exhaustive recall for high-resolution, local precision. If an enterprise workflow strictly requires finding every single relevant document across a massive dataset, DCI may not be the right tool.

Granting an agent expressive tools like an unrestricted bash shell increases latency and compute costs due to the high volume of iterative tool calls required to complete a search. It also creates significant context-management and security challenges for IT departments.

“Tool calls can return large outputs; long trajectories can fill the context window; and raw terminal access requires sandboxing, permission control, and careful engineering,” the authors said. To manage the context window, the researchers found that moderate truncation and compaction help the agent sustain longer searches, whereas overly aggressive summarization tends to discard useful evidence.

Because of these operational realities, DCI is not meant to be a mandatory replacement for existing vector infrastructure. Instead, it serves as a complementary one.

“For orchestration engineers and data architects, our view is that the most practical near-term deployment pattern is hybrid,” the authors said. Semantic retrieval can still provide high-recall candidate discovery when a user’s intent is broad or underspecified. “DCI can then operate as a precision and verification layer: the agent can search within the retrieved documents, expand from them into neighboring files, check exact constraints, and combine weak signals across documents.”

The researchers have released the code for DCI under the permissive MIT license.

“Longer term, DCI changes how we think about enterprise data. Data will not only need to be stored for humans or indexed for search engines; it will need to be organized for agents that can inspect, compare, grep, trace, and verify,” the authors conclude. “File names, timestamps, stable identifiers, metadata, version history, and machine-readable structure become part of the retrieval interface.”

The new standalone app sits on top of Facebook Groups, adds an AI tab called “Ask” and an admin assistant, and arrives the same month Mark Zuckerberg told staff he and Chris Cox had discussed whether they could build 50 new apps.

Meta has released a new standalone app called Forum, without a launch event, a blog post, or much of a press push at all.

The product is built on top of Facebook Groups and is listed in the App Store as “a dedicated space built for deeper discussions, real answers and communities you care about,” phrasing that does most of the work of telling you what Meta wants it to be: a Reddit, with a Facebook account attached.

The launch was first spotted by the social-media consultant Matt Navarra and reported by TechCrunch, with independent confirmation from Engadget and MacRumors. A Meta spokesperson told Engadget the product was still in testing:

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

“We test lots of new products publicly to see what people find interesting and useful to their experiences across our apps.”

Users sign in with their existing Facebook account, at which point their groups, profile and activity carry over. Posts can be made under a nickname, in the same way they can in the main Facebook app, although group administrators can still see real identities.

Anything posted on Forum shows up in the corresponding group on Facebook, and vice versa, so the product is less a separate network than a separate front door into the one Meta already has.

The feed is the differentiator. Where Facebook’s main timeline mixes friends, Pages, algorithmic suggestions and ads, Forum’s surfaces only conversations from groups a user is in, with prompts to discover others. Two AI features sit on top.

The first, “Ask,” lets users put a question to the app and receive an answer compiled from discussions across groups, sparing them the chore of searching one at a time.

The second is an admin assistant designed to help moderators run groups and handle moderation. Both are pitched as time-savers rather than as the product’s defining feature.

This is not Meta’s first standalone Groups app. The company shipped one in November 2014 and shut it down in 2017, for reasons it never made especially clear at the time.

The Reddit framing is also conspicuous given timing: Reddit went public in March 2024, has spent the past two years licensing its data to AI companies for training, and remains the closest thing the consumer internet has to a working community-discussion product at scale. A Facebook-flavoured competitor with an AI answer tab is, at minimum, a recognisable shape.

Forum is the second new Meta app in roughly a month. In late April, the company began testing Instants, a standalone Instagram companion for disappearing photos that borrows visibly from BeReal and Snapchat. Meta Edits, last year’s CapCut-shaped video editor, sits in roughly the same lineage.

That cadence is not an accident. The Wall Street Journal reported earlier this month that Zuckerberg told staff in an internal Q&A that AI-driven efficiency was allowing Meta to build more products with smaller teams, and that he and chief product officer Chris Cox had discussed whether the company could ship 50 new apps.

He then talked himself partway back down: “Like, yeah probably. But we probably should start by doing a few before we just, like, ramp up trying to do 50 all at once.”

Forum is one of the few. Whether users want a separate app for their Facebook groups, with an AI tab attached, is the question the spokesperson’s testing line was designed not to answer. The 2014 Groups app suggests one historical data point. The fate of Instants, Edits, and whatever ships next will produce the rest.

A navigation issue affecting the My Pixel app has left a number of Pixel owners unable to access key sections of the preinstalled Google app, with the bottom tab bar disappearing entirely from the interface without warning.

The missing bar ordinarily carries four tabs covering Home, Tips, Support, and Store, giving users a single access point for feature guidance, troubleshooting help, and direct links to Google’s hardware shop from within the device itself.

Reports from affected users on Reddit confirm the tabs have vanished from their devices, with at least one account suggesting the issue followed a dismissal of a pop-up notification that appeared when the user attempted to navigate to the Google Store section of the app.

That sequence points toward a possible server-side or account-level trigger rather than a local software fault, though Google has not acknowledged the issue publicly or provided any indication of whether a fix is in progress.

The timing is notable given that My Pixel serves as one of the primary in-app channels through which Google surfaces Pixel-specific feature tips and direct hardware purchasing options, making a navigation failure more disruptive than a cosmetic glitch would suggest.

For users who rely on the app’s support tab as their first point of contact when troubleshooting device issues, the disappearance of the tab bar effectively removes that route entirely until the problem is resolved, leaving the Google Support website as the most direct alternative in the interim.

Google has not confirmed whether the issue affects specific Pixel models, specific Android versions, or a broader cross-section of the Pixel install base, and the scope of the problem remains unclear based on available reports.

That silence extends to any fix or workaround, with Google yet to set out a timeline for restoring full navigation functionality to affected devices or clarify whether the issue traces back to a server-side change, an app update, or something else entirely.

The Texas Attorney General has sued Meta over allegations that the company’s WhatsApp messenger, used by more than 3 billion people, doesn’t provide the end-to-end encryption (E2EE) it has long claimed.

Since at least 2016, Meta (then named Facebook) has said WhatsApp provides robust end-to-end encryption, meaning that messages are encrypted on a sender’s device with keys that are available only to the receiver’s. By definition, E2EE means that no one else—including the platform itself—can read the plaintext messages.

In sworn testimony before two US Senate committees in 2018, CEO Mark Zuckerberg said Meta does “not see any of the content in WhatsApp; it is fully encrypted” and that “Facebook systems do not see the content of messages being transferred over WhatsApp.” The engine for this E2EE is the Signal protocol, an open source code base that multiple third-party experts have said lives up to its promises.

In a complaint filed Thursday, Texas AG attorneys said Meta’s claims are false and that the company can and does read the unencrypted contents of WhatsApp messages. They said they are filing the action to “prevent WhatsApp and Meta from continuing to willfully deceive [Texans] by misrepresenting that their private communications were just that—private and inaccessible even to WhatsApp and Meta—when, in fact, WhatsApp and Meta have access to all WhatsApp users’ communications in their entirety.”

“The gravity of Meta’s and WhatsApp’s violation of users’ privacy and trust cannot be overstated,” the attorneys wrote. “All users were entitled to believe their communications were private when WhatsApp and Meta unequivocally and repeatedly promised that no one—not even WhatsApp and Meta—can access their messages.”

In an email, Meta called the allegations “baseless” and vowed to fight the lawsuit in court.

He said, she said

The sole factual evidence cited for the claims is an article published last month by Bloomberg. It reported that the US Commerce Department’s Bureau of Industry and Security had abruptly closed an investigation into allegations that Meta could access encrypted WhatsApp messages shortly after one of the department’s agents sent an email outlining the probe’s preliminary findings.

The search engine’s definitions have been replaced with AI Overviews.

Most GPS watches make you choose between rugged durability and genuinely useful health tracking, which is exactly why the Garmin Instinct 3 Solar stands out.

The Garmin Instinct 3 Solar makes no such compromise, and it is currently down from £299.99 to £209.99 at Argos, saving you £90 in the process.