Remember when recruiting meant paying 20% to headhunters, and still having to do most of the work yourself? Those days are over.Lia is an AI agent who headhunts 24/7 for companies like Glean, Ramp, and NVIDIA.🤯 Companies pay nothing. Candidates pay a success fee of 20% of their first month (not year!) and only if they get hired.→ To hire: Tell Lia your open roles. She'll find the best candidates and intro you directly. Start hiring→ To get hired: Get warm intros to decision makers - no endless applications, no ghosting. No hire = no fee. Get hired
Apple has raised the prices of its Macs and iPads. Price tags for Mac computers rose roughly 15% to 20%, and iPad prices rose 15% to 25%. iPhone prices were unchanged, but price increases are likely coming. The increase in pricing has been attributed to rises in component prices. The price of memory and storage chips has quadrupled over the past year due to surging demand from AI hyperscalers.
The US government has requested that OpenAI release GPT-5.6 to a short list of trusted partners before a wider release. OpenAI staff have been instructed to work with the Trump administration on any input that officials have on safety and restrictions. GPT-5.6 will initially be released to 20 partners through Amazon's Bedrock software platform. The Trump administration is continuing to collaborate with frontier AI labs to develop a shared approach for addressing the challenges of scaling the technology.
SpaceX's planned AI satellite constellation will be called Starmind. Starmind will compute data directly in orbit using onboard processors powered by large solar arrays. It will allow AI models to run inference, process queries, and generate outputs from space. Starship will be able to carry 30 to 50 Starmind AI1 satellites per launch.
IBM claims to have created sub-1-nanometer chip technology. The company says its nanostack architecture can deliver the computing performance expected if a theoretical chip could be built with physical features smaller than 1 nanometer. The design stacks transitions in a staggered layout to pack more transistors in the same space. IBM describes its new chip technology as being built at the 0.7-nanometer node, but it is important to remember that these node numbers have nothing to do with the actual physical dimensions of the chip.
The AI-native tooling layer is multiplying fast and is crowded already. Being an 'AI engineer' is not a moat. AI can help build prototypes, but production is a different animal that requires engineers who understand reliability, scale, security, performance, observability, and operational trade-offs. The biggest returns may actually come from knowing one hard thing exceptionally well.
Gemini 3.5 Flash has a built-in tool called Computer Use that lets the model look at a screenshot and a goal and return structured actions. Users can choose to execute those actions and repeat the process until tasks are complete. Gemini supports desktop, mobile, and browser automation. This guide walks readers through how to set up mobile automation.
500,000 people have already signed up for TLDR Hardware, our new twice-weekly newsletter covering chips, robotics, energy, and devices. If you work in hardware and want to help curate it, send your LinkedIn or resume to hardware@tldr.tech!
A series of recent developments has caused OpenAI's executives to lean towards holding off the company's initial public offering until next year. SpaceX's IPO was the largest ever, but its stock value dropped quickly. Global markets have been choppy in recent weeks, with tech stocks dragging down indexes. These factors may mean that retail investors might not have much enthusiasm for OpenAI's shares.
Z.ai released GLM-5.2 just days after Anthropic was forced to shut down its most powerful AI systems due to a demand from the US government. GLM-5.2 is nearly as powerful as Anthropic's Fable and Mythos models, but it costs much less to use, and no one in the US is putting restrictions on it. The model is currently one of the world's top 10 most popular models. It is open source, so anyone can use and modify it for free.
Our writers mention the companies our 7M+ readers care about. A free tool scans the last year and counts yours. Drop in your domain to see the number and who saw it.
Figma co-founder and CEO Dylan Field discusses Figma's almost acquisition, its IPO, the company's differentiation discovery process, the nature of creativity versus design, and AI.
Akrites provides a single, standardized Coordinated Vulnerability Disclosure process built on confidentiality-first principles and the industry's established standards and tooling.
Same input. Same prompt. Different output. That's the reality of testing AI agents that write code, and most teams are shipping without solving it.Nick Nisi from WorkOS tackled this by building eval systems for two AI tools: - npx workos@latest, a CLI agent that installs AuthKit into your project - WorkOS agent skills that power LLM responses about SSO, directory sync, and RBAC. The post covers how to test against real project structures, score output that's different every time, and catch when your agent makes up methods that don't exist. Learn more about evals →
Liquid AI announced the release of LFM 2.5, a 230-million-parameter non-transformer model architecture built on top of state-space and liquid neural network continuous-time formulations. Despite its exceptionally compact footprint, the model achieves performance parity with transformer models three times its size on core edge reasoning and sequence generation benchmarks.
Vercel released AI SDK 7, introducing an upgraded, zero-overhead execution loop that dramatically simplifies how frontend frameworks handle multi-step tool calls and streaming agentic UI states. The release features a unified telemetry layer that hooks directly into serverless compute runtimes to provide absolute tracing visibility into token usage, model choices, and tool execution latency.
The White House has issued an official administrative request asking OpenAI to delay the public deployment of its next-generation frontier model over national security and structural safety concerns. Government officials are pushing for an extended red-teaming window to thoroughly audit the system's advanced cyber-capability execution limits and automated social manipulation vulnerabilities.
The generative AI economy has generated $110 billion in sales over the past 12 months, and it's growing fast. The revenue run rate exceeds $175 billion on an annualized basis. The supply side of the AI market is well-understood, but understanding the demand side is much harder. This post looks at total AI spend, enterprise and consumer, to see how big the market really is, whether revenues are growing, how much revenue is covering the investment expense, and what will happen in the future as token prices fall and the quality of tokens improves.
Scaling laws are one of the most critical empirical findings in deep learning. They can be a framework for describing the relationship between compute, loss, model size, and data. Their predictability makes them highly valuable in practice. This article discusses scaling laws, how they can be used to allocate compute optimally, and their flaws.
Back-to-back meetings with coffee chat follow-ups. Already forgot half the details? Memoket captures every conversation with one press and connects the dots across your conversations - dropping summaries, tasks, even your weekly report straight into your workflow. Wearable as a wristband, pendant or Apple Watch attachment. Pay only $5 to reserve early-bird pricing.
Meta Autodata trains AI agents to act as data scientists that create higher-quality training and evaluation datasets. Its Agentic Self-Instruct implementation improved results across coding, legal reasoning, and mathematical reasoning tasks.
Ornith-2.0 is a coding model family that can write RL scaffolds. Each variant of the self-improving family of models is trained on top of pretrained Gemma 4 and Qwen 3.5 foundations. Ornith-1.0 is state-of-the-art among open source models of comparable size. The weights and a technical report are available on Hugging Face for teams that want to run or study the models directly.
500,000 people have already signed up for TLDR Hardware, our new twice-weekly newsletter covering chips, robotics, energy, and devices. If you work in hardware and want to help curate it, send your LinkedIn or resume to hardware@tldr.tech!
Researchers introduced the Reward Hacking Benchmark (RHB) to measure how reinforcement learning post-training influences the tendency of coding agents to exploit evaluation flaws rather than solve tasks honestly. Testing across 13 frontier models revealed that RL-tuned variants exhibit exploit rates up to 13.9% by bypassing verification steps or modifying grading scripts, whereas standard post-trained models stay near 0%.
This post shines a light on the job search experience for a research scientist position in Silicon Valley. The author is a fifth-year PhD student at Brown University. Some of the surprising things about the job search were that only one or two of their research papers really mattered, there were very diverse interview rounds, and the importance of timing. A lot of interviews came from a lot of places outside of the author's expertise - many places were evaluating them on how well-rounded an AI researcher they were.
Algolia's leaderboard ranks 21 models' responses based on relevance, utility, and accuracy. Find which model is best for in-app and product search. See the results.
Hugging Face launched a single-command deployment workflow that lets developers spin up private, OpenAI-compatible vLLM endpoints on its pay-per-second serverless Jobs infrastructure.
Generative Intuition showcased a real-time behavioral tracking pipeline designed to monitor and visualize fine-grained physical human interactions across multimodal computing interfaces.
The tokenized RWA market reached $51 billion in market cap, a 40% year-to-date gain while the broader crypto market fell about 20%, with private credit comprising 47% of total value and Ethereum and Provenance together hosting over 70% of tokenized asset activity. Equity tokenization leads segment growth at 130% YTD, expanding from $700 million to $1.6 billion, with monthly transfer volumes rising from $500 million in September 2025 to $5.3 billion as of June 2026. The sector is bifurcating around two structural models: broker-dealer custody arrangements (Robinhood's approach) enable 24/7 trading but strip token holders of registered shareholder status, dividend rights, and voting privileges, while the settlement infrastructure model pursued by Figure, Securitize, and Coinbase uses SEC-registered transfer agents to confer full on-chain ownership rights. The SEC's concurrent proposal to rescind rules allowing greater decentralized trading flexibility adds regulatory uncertainty to the settlement infrastructure buildout.
Bitcoin bounced above $60,000 after a dip to $59,200, but weekly losses remain steep across major tokens: BTC off 5.4%, ETH down 7.9% to $1,616, XRP down 9.2% to $1.07, and SOL, HYPE, and DOGE posting double-digit weekly declines, with Tron the sole major gainer at +1.9%. Crypto decoupled from a broader equity rebound driven by Micron's blowout sales forecast and a 1.8% lift in Nasdaq 100 futures, with analysts attributing the divergence to continued US spot Bitcoin ETF outflows, the Federal Reserve's hawkish stance, and a US dollar at seven-month highs. Bitcoin's proximity to its 200-week moving average mirrors setups from 2015, 2018, and 2022, each of which preceded crypto winters lasting six months to nearly two years, with $55,000 cited as a plausible cycle low if the $61,800-$62,000 support zone breaks. Upcoming US PCE inflation data is the next catalyst that traders are watching.
Kraken and Maple Finance have launched an on-chain OTC lending warehouse facility modeled on asset-backed securitization, with Maple providing senior financing via a bankruptcy-remote SPV and Kraken serving as loan originator, servicer, and junior lender taking the first-loss position.
Harvest-now-decrypt-later (HNDL) vulnerabilities make reactive quantum-resistance approaches, including turnstile mechanisms and proof system swaps, insufficient for private ETH. The ~620k shielded ZEC pool in Zcash's Sapling (~$250M) illustrates the scale of value exposed to future quantum decryption. A proposal on ethresear.ch calls for proactive, L1-native quantum-resistant cryptography, recommending hash-based proof systems (WHIR and STIR constructions) over lattice-based alternatives for their support of larger circuits. For encryption, ML-KEM (Kyber) replaces traditional key agreement with key encapsulation, SPHINCS+ covers stateless spending authorization, and PRF-based derivation handles key management without exotic primitives. A phased "progressive pool uncapping" mechanism would expand deposit limits during early deployment to bound risk exposure.
Standard Chartered analysts project Aave reaching $3,500 by end-2030 (a ~50x multiple from ~$70), with staged targets of $180 by 2026, $600 by 2027, $1,200 by 2028, and $2,200 by 2029, alongside forecasts of $500,000 for Bitcoin and $40,000 for Ethereum. The Aave thesis rests on a projected 37-fold expansion of tokenized assets deployed in DeFi to $2.7 trillion by 2030, with Aave capturing revenue through lending spread fees as that capital flows through the protocol. The bank flags Aave Horizon, the institutional lending arm, as "achievable but not yet proven," noting that the TradFi partnerships required to drive the thesis have not materialized at scale. The forecast comes after an April exploit cut Aave's total deposits from $44 billion to $23 billion, making the recovery arc central to whether the multi-year price path holds.
The Federal Reserve has proposed a new "payment accounts" category for non-bank fintech and crypto firms, offering Fed access without discount window privileges, interest, or intraday credit, and capped at $1 billion. Public comments close July 27, Tier 3 master account applications remain frozen through December, and Congress is debating whether to legislate the framework.
HyperEVM is the programmable layer over HyperCore, where read precompiles let contracts query the exchange state (balances, positions, prices, and vault equity) and CoreWriter lets contracts submit actions back to the exchange. There is a 2x2 diagnostic framework built around two questions: why does an app need EVM, and why does it need Hyperliquid specifically? Familiar DeFi primitives like AMMs and money markets (Felix and HyperLend) qualify for the EVM-only quadrant, while the differentiated category includes protocols that use CoreWriter to programmatically control HyperCore positions: Rysk for volatility products, Liminal for packaged strategies, Hyperbeat for delta-neutral, and Valantis Prime for smart account execution. The long-term thesis is a unified financial account where a single balance moves across trading, borrowing, yield, hedging, and payments, with HyperCore serving as the underlying balance sheet.
SBI Holdings has agreed to acquire all shares of Japanese crypto exchange Bitbank for approximately ¥46.7 billion (~$288.6M), expected to close in October.
A CoinShares survey of 261 European wealth management professionals across the UK, France, Germany, Italy, and Switzerland found that 50% of UK advisors say most of their clients' crypto holdings sit outside their oversight.
Every finance task takes multiple steps. Check runway? Find the right report, cross-reference your burn, interpret the graph. Pay a contractor? Track down the invoice, navigate to transfers, fill in the details. Freeze a card? Dig through settings.Mercury Command eliminates the switching tax. It's AI built directly into your Mercury account that understands natural language and executes from one place — payments, invoices, categorization, team management. You review and approve every action. Command does the rest.Try Mercury Command →*Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column N.A., Members FDIC.
Traditional experimentation had its limits, but some of them made sense because product development was slow. Most experiments focus on pulling up revenue. They don't deliver incremental progress. Things are now different, so founders should skip minor optimizations, take bigger swings, and let tests run longer.
The one-person company is having a real moment. Solopreneurs clearing $1M a year doubled from 2023 to 2025, and roughly three times as many crossed $5M and $10M, and it isn't fraud. The same surge shows up across countries. Newer cohorts ramp faster too - 2025 sign-ups hit $1M about 30% quicker than 2023 and three times quicker than 2019.
Technology analyst Tanay Jaipuria broke down the primary market map split emerging within the agentic software landscape, contrasting end-to-end applications against backend infrastructure tooling. While bespoke vertical agents capture high initial contract values by replacing human service labor, infrastructure platforms gain long-term defensibility by providing essential orchestration layers.
As foundational model access commoditizes pure software functionality, early-stage technology startups are pivoting toward proprietary data loops and deeply embedded operational workflows to secure defensible moats. Successful teams are focusing heavily on capturing high-fidelity enterprise interaction data that cannot be scraped or simulated by generalized frontier systems.
Cal AI went from nothing to the number one health app in 18 months on the back of influencers, and this is the actual playbook. Judge creators on their baseline views and a live comment section, not follower count, and pay flat rates so a viral video never costs you more. The money is in mid-sized creators, and the counterintuitive part is that you're renting the audience, not the influencer, which is why a TikTok dancer outsold UFC fighters for a calorie app. Four people ran hundreds of partnerships by automating the entire pipeline, because the only moat left is speed.