How one developer finally got a local LLM running with Home Assistant in an hour

A developer who had been struggling for over a year to run a local AI model succeeded in just one hour after receiving advice from the DEV Community. Using llama.cpp and the Qwen 3.8 27B model on an RTX 4090 and RTX 4070 Ti GPU pair, they built a working local AI agent connected to Home Assistant. The setup allowed the model to modify smart home dashboards autonomously using prompts and screenshots, with built-in vision support requiring no extra configuration. Community discussion also highlighted recurring issues with OpenWebUI, including excessive CPU usage bugs fixable via Docker environment variables, and noted that llama.cpp ships its own built-in web UI as an alternative. Several users shared similar local AI stacks combining llama.cpp, SearXNG, and MCP servers as a self-hosted equivalent to commercial models like Claude.

Excel's XLOOKUP Explained: How It Differs From VLOOKUP and Why It's Better

XLOOKUP is a modern Excel function that retrieves data by taking three arguments: the value to find, the column to search in, and the column to return results from. Unlike VLOOKUP, which requires selecting an entire data range and counting columns from left to right, XLOOKUP lets users reference individual columns independently. This means the return column can be positioned anywhere — to the left or right of the search column — without breaking the formula. Adding new columns to a dataset also does not disrupt an XLOOKUP formula, making it more robust for dynamic spreadsheets. The function is considered a significant upgrade over VLOOKUP for everyday data lookup tasks in Excel.

Bitcoin's surging price faces 1 key level that could signal if the bear market is really over

Your day-ahead look for Aug. 25, 2026

India Shifts Focus to Building Its Own Military Technology Ecosystem

India's defence sector is undergoing a fundamental transformation in how military technology is developed and produced domestically. The change goes beyond individual weapons systems like missiles or fighter jets, focusing instead on reshaping the broader defence-industrial base. This shift signals a long-term strategic move toward self-reliance in military capabilities. The evolving ecosystem reflects India's ambition to reduce dependence on foreign defence suppliers and build indigenous technological depth.

AI Agent Engineering in 2026 Shifts Focus to Reliability and Orchestration Layers

A weekly digest covering AI agent developments from August 18–25, 2026 highlights a growing industry focus on reliability over raw model capability. Research from Thinkingbox revealed that a single successful agent run can mask significant performance collapse across repeated attempts, with pass@1 scores hiding failures at pass@20. Tools like LangSmith's Tuned Evaluators aim to address this by grading the full volume of production traffic at low cost. Multiple frameworks — including AgentWeave, Spine-Branch Coordination, and AutoSaddler — reflect a harness-first engineering mindset, treating the orchestration layer around models as the primary source of performance gains. AgentWeave alone reportedly cut tool exposure by 70% and reduced latency by 51%, underscoring the practical value of routing and coordination improvements over model-level changes.

Why LLMs Need Dedicated Tools, Not Guesswork, for Medical Scheduling

Large language models like Claude and GPT-4 struggle with deterministic tasks such as medication scheduling, often producing incorrect dates when constraints like preferred days of the week are introduced. Because LLMs are probabilistic by nature, errors in dose calculations carry serious liability risks in healthcare settings. To address this, the Injection Day Alignment MCP server provides three structured tools — schedule generation, shift-offset calculation, and compliance verification — that handle the underlying arithmetic reliably. These primitives use ISO 8601 date formats to eliminate timezone ambiguity and support complex constraints such as aligning injection days to avoid pharmacy closures. The broader argument is that AI agents operating in sensitive domains require precise, deterministic interfaces rather than relying on models to perform critical logic internally.

US diplomats heading back to Middle East as immediate Iran war risk recedes: Report

Woman Blocks Bulldozer Amid Violent Land Dispute in Thane

A tense land dispute in Thane, Maharashtra turned violent when a private builder deployed a bulldozer in an attempt to clear the contested property. A woman was seen physically embracing the bulldozer in a bid to halt the demolition work. The incident highlights the ongoing conflict between the builder and those occupying the land. The confrontation was captured on video, which has since drawn widespread attention online.

How to Run Effective Ad Hoc Testing Sessions Without Losing Structure

Ad hoc testing lets developers and testers explore software freely without predefined test cases, making it useful for uncovering unexpected behaviour in seemingly stable features. Testers are advised to focus on one feature at a time, complete normal user journeys first, then gradually introduce edge cases using invalid, duplicate, or boundary-value data. Each session should be kept to around 30 to 45 minutes to maintain focus and yield more actionable observations. Despite its informal nature, all meaningful findings should be documented in enough detail for teammates to reproduce the issues. High-risk, repeatable discoveries should be converted into regression test cases to strengthen the overall QA process.

Developer Discovers RAG System's Full-Text Search Was Broken After First-Ever Benchmark

A developer building Langhuan, an open-source knowledge base for AI agents, admitted to never benchmarking its retrieval system despite multiple builds and deployments. Last week, they ran their first formal evaluation using 200 queries sampled from the MIRACL-zh dataset, measuring recall, MRR, and nDCG across vector, full-text, and hybrid search configurations. The benchmarking immediately exposed two critical bugs: a missing foreign key constraint that crashed fresh SQLite instances, and a vector search extension that had never been linked into the production binary. Further results revealed that full-text search (FTS) had zero recall across all 200 queries, while hybrid search scores were identical to vector-only — meaning one of the two retrieval channels had silently never functioned. The developer noted that existing unit and integration tests had missed all these issues because none of them simulated the behavior of a real, first-time user.

Best Car Phone Mounts of 2026: Top Picks After Real Road Trip Testing

A Wired reviewer tested approximately a dozen car phone mounts during road trips this summer to identify the top performers. The evaluation focused on two key criteria: how securely each mount holds a phone and whether it obstructs the driver's line of sight. Among the models tested, picks from Belkin, Andery, and Andobil emerged as standout recommendations. The review aimed to help drivers find reliable hands-free phone solutions for safer driving.

Developer Builds Token Fuel Gauge for Claude Code to Prevent Silent Quality Drops

A developer running an autonomous business setup on Claude Code's MAX plan discovered that token consumption within its 5-hour usage blocks is invisible by default, with no on-screen indicator as the limit approaches. Output quality was found to degrade silently near the 800k output-token threshold, a problem the developer noticed only after reviewing usage data with a tool called ccusage. To address this, the developer built a shell script called token-budget-advisor.sh that aggregates token usage and session costs from two sources and displays a color-coded status — green, yellow, or red — directly in the terminal status line. The system is designed to eliminate the cognitive overhead of manually checking remaining capacity, drawing on the same design principle as a car's fuel gauge. Once integrated into a dashboard update script, the tool automatically surfaces budget state every time a terminal is opened, prompting timely decisions about when to defer heavy workloads to the next usage block.

Greenworks MaximusZ Electric Riding Mower Packs Five Motors for 54-Inch Cut

Greenworks has launched the MaximusZ, an electric riding mower featuring five motors — two for wheel drive and three powering a trio of cutting blades. The mower cuts a 54-inch-wide swath in a single pass, offering notable agility including smooth 360-degree turns. It enters a growing market of electric outdoor power equipment aimed at replacing gas-powered alternatives. While the MaximusZ impresses on performance, reviewers note the model comes with certain compromises that prospective buyers should consider.

Regulators Begin Scrutiny of AI Systems That Target Users at Vulnerable Moments

Digital platforms can now infer users' emotional and financial states in real time and deploy loan offers or persuasive prompts precisely when individuals are most vulnerable, raising serious questions about consent and manipulation. A peer-reviewed paper published in Frontiers in Psychology in April 2026 found that AI-driven design increasingly exploits inferred cognitive and emotional vulnerability to influence user decisions. The UK's Treasury Select Committee warned in January 2026 that British regulators were insufficiently managing AI risks, prompting the FCA's Mills Review to examine how Consumer Duty rules apply to AI personalisation. The EU AI Act's Article 5 prohibitions on AI systems exploiting socio-economic or psychological vulnerabilities became enforceable in February 2025, carrying fines of up to 35 million euros or 7% of global turnover. Meanwhile, the affective computing market — technologies that detect and respond to human emotions — is projected to grow from roughly $100 billion in 2025 to over $300 billion by the early 2030s, far outpacing the legal frameworks meant to govern it.

Illinois Prosecutors Shared Defendants' Data With ICE Without Warrants

County prosecutors in Illinois secretly provided defendants' personal information to federal immigration authorities. The data was handed over without criminal warrants, bypassing standard legal safeguards. The practice occurred without any public disclosure, keeping affected individuals unaware their information was being shared. No legislative oversight was in place to monitor or regulate the arrangement between local prosecutors and ICE.

India Seeks Alternative Oil Sources as US Pressure Mounts on Russia and Iran

India faces potential disruptions to its crude oil supply as US sanctions threaten imports from both Russia and Iran. Russia currently accounts for over 50% of India's crude oil basket, a share that grew amid the Middle East crisis. Secondary sanctions could force India to drastically reduce Iranian oil purchases, while tariffs of up to 100% could make Russian crude imports unviable. With two of its major suppliers under pressure, India is exploring alternative sources, including Venezuela, to fill the potential gap. The situation underscores India's vulnerability to geopolitical shifts affecting its energy security.

Bitcoin Rally Looks Structurally Sound as Short Squeeze Clears Futures Market

Bitcoin has experienced a significant short squeeze, forcing bearish traders who had bet against the price to exit their positions at a loss. Open interest in Bitcoin futures has collapsed as a result, indicating a large number of leveraged short positions have been liquidated. Funding rates, which reflect the cost of holding leveraged positions, have remained subdued throughout the move. Analysts note that falling open interest combined with low funding rates suggests the current rally is built on healthier market foundations rather than excessive speculation.

Manipur Ramps Up Pressure on Centre for NRC Before Census 2027

Protests in Manipur are intensifying over demands to complete the National Register of Citizens (NRC) before the upcoming Census 2027. The movement has gained political backing, with Chief Minister Yumnam Khemchand Singh publicly supporting the call. A delegation of BJP legislators from Manipur has travelled to New Delhi to press the Central Government on the issue. The push reflects growing concerns in the state about identifying and documenting residents ahead of the national headcount.

China vows to protect interests after US broadens Iran sanctions

The United States has expanded sanctions against Iran, warning that countries continuing to trade with Tehran will face isolation. China, which is a major buyer of Iranian oil, has responded by pledging to defend its own interests. The move reflects growing tension between Washington and Beijing over Iran-related trade ties. The US measures are aimed at pressuring Iran by cutting off its key revenue streams, particularly oil exports. China's pushback signals potential friction in US-China relations over the enforcement of these sanctions.

Operation Combat: How Japan PM Sanae Takaichi dealt with cockroach chaos