Ever wish you could chat with your CRM? You can with Lightfield. In fact, the average user asks it 400 questions a week. E.g., you can ask it to:

👉 Compare closed-won to closed-lost and find the difference.

👉 Prep you for upcoming calls.

👉 Draft win-back emails for every churned account that mentioned a missing feature you just shipped.

Lightfield updates itself based on every email, call, and meeting, automatically. Stop filling in fields and start selling again.

➡️ See why +5,000 companies signed up in 10 months - and why a16z just got behind Lightfield's $47M Series A. Book a demo

Apple is working on an AI server powered by its M-series Ultra chips. It is expected to be released in 2029. The enterprise server will come with either two or four M8 Ultra chips. The project, which received support from John Ternus when it began a year ago, back when Ternus led Apple's hardware engineering efforts, will be Apple's first server to hit the market in nearly two decades.

Snap's forthcoming Specs augmented reality glasses will launch later this year at $2,195. People will be able to try the glasses at select Verizon stores and purchase custom wireless data plans for them. There will be a higher-priced version at $2,395 that includes a carrying case with built-in cellular data connectivity.

Researchers have developed a technique for installing millions of human brain cells into mice. The scientists bred mice that were missing most of their cerebral cortexes and then replaced a large part of those missing mouse cells with lab-grown human brain cells. This resulted in mice with human neurons that could be altered to approximate human brain conditions. The technique could potentially create better scientific models for studying neurodegenerative disease.

OpenAI has revealed several more incidents involving its AI systems as part of its new framework for reporting misalignment. The company's AI systems hid mistakes, made up data, and moved miles onto the open internet without permission. The industry is currently in heated discussions about whether AI development needs to be slowed to address the technology's potential dangers. AI leaders have called for a pause in the technology's development to provide more time to build proper guardrails.

A recap with the wrong number. A customer quote that came out garbled. Wispr Flow Notetaker gets the hard words right, keeps every meeting searchable, and sends that context into Claude or ChatGPT. Available now for Mac and Windows, free to try.

Get Notetaker free | See how it works

Google has opened early access to Model Context Protocol for Google Home in English for Google Home Premium Advanced users in the US. Any agent that supports calling MCP tools can now interact with all of the devices and event history in the Google Home ecosystem. The MCP allows agents to analyze and answer questions about home activity, summarize camera footage across rooms, manage devices, and create smart home dashboards using everyday language. It contains safety protections, such as prohibiting sensitive actions like unlocking doors, but depending on the agent, using the MCP can result in unexpected or even undesired behavior.

Query optimizers continue to leave much to be desired despite an enormous body of research. A Postgres database should know everything about the stuff that lives in its tables, but the problem is enormously hard. What's not as hard is verifying whether a query plan an optimizer picks is good or not. Language models are particularly good at learning how to do tasks with easily verifiable outputs. A small, open-weights model can be post-trained via supervised fine-tuning and agentic reinforcement learning to produce Postgres query plans that beat Postgres' default plans.

While Google isn't required to break up its ad tech business, judges have mandated changes to the company's practices. The changes largely align with recommendations made by Google and the Justice Department to correct the company's monopolistic behavior. Google will make its ad pricing more transparent to marketers and rival ad businesses, and appoint a trustee to oversee the changes.

Code can't explain things like the history can. You can't understand a codebase from its current state alone. AI models have no commit history. Nobody wrote down why it does things, so researchers have to reconstruct its thought processes from whatever traces are available.

Skip the "What's for lunch?" Slacks. Get a pre-populated weekly meal calendar with your team's favorite lunch spots. It does everything – provides restaurants, sends reminders, and checks out orders automatically.

See how it works

100+ sessions, five tracks, top speakers, hands-on labs, and practical AI, security, and networking guidance. Register Now

AI is permanently changing the way mathematical research is done.

OpenAI's Sponsored Agents let people start a conversation with a business-sponsored agent after clicking an ad in ChatGPT.

TIN (Text INdex) is a fast, full-featured, and reliable full-text search extension for Postgres.

Meta's Muse is impressive, but there aren't that many companies with a worse tech brand when it comes to user trust.

Regulatory capture is when the government slowly gains more and more power to regulate an industry and eventually becomes the industry's protector.

Filings describe hyperscaler contracts but reveal little about private labs' customer agreements.

A look into the interconnected network of Chinese hacking-for-hire companies that steal and analyze confidential information and sell it to Chinese authorities.

Every Dreamforce presentation glosses over one question: where does your agent get its data? If it has to wait for your reps to update a legacy CRM, your future isn't actually agentic.

Lightfield is the CRM that updates itself from every email, call, and meeting. It uses hard evidence to answer your questions, like:

→ "What objections keep coming up across my calls?"

or

→ "Show me every deal where the champion has gone quiet and draft a re-engagement email for each one."

5,000+ companies run on it, including deployments with hundreds of users.

Book a demo to see what an agentic CRM really looks like

OpenAI has held early discussions with investors about a financing round that would precede the company's highly anticipated initial public offering expected next year. The new funding round, which could value the company at more than $1.2 trillion, has been delayed due to concerns about AI safety. OpenAI has reached more than one billion active users, with more than 200 million businesses using its models.

SpaceX plans to launch its 14th Starship mission as early as September 22. It will be the first time SpaceX attempts to launch the experimental vehicle into orbit. The super heavy lift rocket will be carrying 26 of the larger V3 Starlink satellites into orbit 275 kilometers above Earth. The upper stage will seek to complete six orbits around Earth before completing its mission after about 10 hours. SpaceX will not attempt a return of the upper stage to the launch site in Texas on this mission, as it plans to gather more data about heat shield performance.

Researchers at the Janelia Research Campus at the Howard Hughes Medical Institute in Virginia, in partnership with Google, recently published a new map of the adult male fruit fly's brain. The effort, which involved hundreds of scholars, resulted in a complete map of the fly's 166,000 neurons that can be trained to complete tasks like solving a Rubik's Cube or playing poker. Internet engineers and artists have made the model complete many seemingly impossible tasks for a fruit fly, like parallel parking a car, playing Doom, and trading cryptocurrencies.

Agility Robotics has debuted a humanoid robot engineered to work safely near humans without risking harm to their co-workers. The robot autonomously takes precautions when it detects a person at a distance, such as moving to avoid the person or standing still so the person can pass without getting closer. It can even choose to squat and assume a seated position if a person is getting too close. The robot will be available in early access within the first half of 2027 and become generally available by the end of the year.

AI disruption typically follows 3 archetypes. In this on-demand webinar, Gartner breaks them down and lays out a plan to turn the AI threat to your advantage. Get practical steps to rebuild your moat and explore the AI hub for 4,000 case studies and use cases. Visit the hub

Google has released two new speech-to-speech models similar to OpenAI's GPT-Live family. Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking are advanced live dialogue models with major upgrades in intelligence and parallel reasoning. They are more intuitive to collaborate with than previous models and can be used to execute complex tasks using voice prompts. Users can select a model and voice preset, enter an optional system prompt, and then start a voice conversation through their browser. The models can be interrupted while talking.

Until Fable-tier intelligence gets cheap, everyone building orchestrators is just dabbling. They aren't getting anywhere useful yet. The pace of model development will likely continue. However, resource limitations like compute and RAM have started to flatten the curve.

The US Senate has voted to block consideration of the Clarity Act, a piece of legislation that would create business-friendly rules for cryptocurrencies. The vote makes it unlikely the legislation will pass anytime soon, if ever. Many Republicans embraced the bill, but Democrats largely opposed it. The Democrats wanted stronger language to prevent public officials from using crypto to make money. President Trump generated $1.4 billion from a network of crypto businesses last year.

Mark Zuckerberg made a social media post on Tuesday, saying that calls to pause AI development industrywide were the wrong approach. He said that every lab has the responsibility and incentive to move at the pace required to train its models safely. He acknowledged the risks of AI, but said companies were already incentivized to keep users safe and slow progress on their own. Meta has been aiming to catch up in the AI race after falling behind Anthropic and OpenAI last year. It plans to release a new model, internally called Watermelon, in the coming weeks.

Picking an ERP? Don't guess. Get the free Software Advice guide comparing NetSuite, Fishbowl, and Global Shop Solutions on pricing, features, and verified user reviews — so you choose with confidence. Get your free guide

The last time OpenRouter users spent more on OpenAI than on Anthropic models was more than two and a half years ago.

Cloudflare's new Disallow AI Training setting lets users easily stay indexed for search while refusing to let crawlers train on their content.

TypeSafe AI's Jev model, available now in early access, is part of a new class of frontier models called System One Models built to make fast, structured decisions that software can use directly.

No company owns what comes next, and more people should be able to advance the edge of what we know.

Some decisions, where creativity adds almost no value, don't need to be reinterpreted on every generation.

Tailcat is like netcat, but over Tailscale's data plane and without Tailscale's control plane.

A PM posts in #product: "Can we get a churn dashboard, refreshed daily?" Viktor replies with a link to a web app it built and deployed. Engineering sends it a pull request to review and finance points it at overdue QuickBooks invoices.Viktor is an AI employee in Slack and Microsoft Teams. It connects to 3,200+ tools, runs scheduled tasks on its own, remembers how your company works, and asks for approval before acting."It's a proactive employee that learns what's important in your company and delivers without hesitation." Matt Lis, CEO, CodeYourBrandStart free. Up to $100 in credits →

iOS 27 marks the beginning of an iPhone divide: only Apple Intelligence-capable models will get Siri AI and other AI-powered features. The new Siri AI uses Google's Gemini models to answer questions. It can look through users' private data, but Apple says it doesn't store user data and that neither Apple nor Google can access it. This article provides a summary of the new capabilities Siri AI has as well as the new features available to all iPhones with iOS 27.

For 23 years, the single fact dominating every moment of every room Mark Zuckerberg has been in is the fact that he is Mark Zuckerberg. Today, his empire has 3.6 billion users, has a market capitalization of $1.6 trillion, and has averaged the addition of a zero to its valuation roughly once every US presidential term. It's not a normal situation for a human being to be in. This article takes a look at how Zuckerberg built his empire, the problems he faces at the top, Zuckerberg's various other ventures, and more.

It will be impossible to stop people, amateur or professional, from pushing a button to produce mathematics. The field shouldn't only be reserved for graduate students and the like. There's now more interest in maths than at any other time in history, and that's a good thing. This will bring an abundance of new maths. The only difference is that some questions will now be answerable via a cash injection.

The US has officially announced the existence of space control weapons in orbit. Pentagon leaders have become more open to discussing space warfare in recent years, so the deployment is not entirely unexpected. The announcement didn't include details about the weapons. The deployment of space technology often takes years, so the weapons program has likely spanned multiple presidential administrations.

Your agent's imagination impresses in a pilot. To deliver in prod, you need to connect it to reality. New Relic NOW, live on October 6th, shows you how to build the operational data system required to shrink MTTR down to minutes, shrink tool spend, and deploy AI safely at scale. Save your spot

There are at least two ways to have AI assist in programming: one is to use it to help understand and interact with the implementation, and the other aims to remove the need for that interaction altogether. These are different paths, so they require different tools and different approaches. The goal of someone using AI to further their understanding is not to become a vibecoder, and ceasing to read code isn't, in itself, progress.

Xcode has all the tools needed to develop, test, and distribute apps for Apple platforms. Xcode 27 features coding agents powered by any model. Users can customize Xcode to the way they work and manage all of their devices in Device Hub. The release includes updates to performance and testing tools and enhancements to localization.

Apple has designed its new Siri architecture to work with third-party AI models at a surprisingly deep level. One mechanism called Model Delegation allows Claude to appear as a Siri extension in the same way as the existing built-in ChatGPT extension. It appears that Claude hands tasks back to Siri if a request requires access to an Apple system feature. An inference provider option allows Apple's own server-side Siri model to be completely replaced by another model.

If all of the AI labs slowed down, they would collectively spend less on compute, and they'd also be able to charge frontier-model prices for a longer time. However, if they all got together in a room and agreed to slow down, that would look like an antitrust conspiracy. It is generally illegal for competitors to get together and agree to limit the output of their industry. If they all published papers about how important it is to slow down, and the government believes those papers, it might help them coordinate.

Salesforce built the CRM you type into. Lightfield is the agentic CRM that updates itself from every call and email, then prospects and drafts outreach for you. Skeptical? Book a demo

AI is reshaping how software gets planned, built, and run. Join Atlassian and Vercel's CEOs and leaders from DX, Lovable, and more on how they're adapting for the AI era. Reserve your spot.

This post looks at what Figma conventions do, where they live, and what they cost, and explores icons, subcomponents, and images in depth.

Agentic engineering is about making hidden constraints visible and cheap changes trustworthy.

LLM-generated text scores badly on Pangram even if it has been heavily edited.

@shadcn/lint is an agent-first linter for Tailwind design systems that works with existing design systems without any rewrite required.

Nvidia CEO Jensen Huang recently appeared in an interview where he discussed AI safety risks, what frontier labs should do about them, recursive self-improvement, AI job creation and data centers, why neo-clouds exist, AI's future, and more.

dbt Charts is a declarative language for dashboards that allows anyone with a data connection to generate a report in an afternoon.

Orkes Shift '26 is one day on running agentic systems in production: durable execution, retries, distributed failures, and human approvals. October 15, Convene Union Square, San Francisco. Early bird pricing ends soon and capacity is limited.The agenda:

• “Rebuilding a Monolith: Choreography to Orchestration”
• “Designing a Focused Agent Runtime on Conductor”
• “Stop Babysitting Your AI Coding Agents”
• Speakers from LinkedIn, Bumble, and the Orkes founders, plus a fireside and happy hour

Conductor runs 1B+ executions a day at Netflix, Tesla, and LinkedIn.Register here →

Anthropic's CEO recently released a 3,800-word essay calling for a global slowdown of AI development. He said that while the technology offers many benefits, it is advancing too quickly for researchers to continue safely. AI leaders are preparing for a day when an AI system can grow more advanced without the aid of human researchers. Some researchers believe that achieving recursive self-improvement will lead to AI systems spinning out of control.

OpenAI CEO Sam Altman has confirmed that the company will not be going public this year. He said it would be an ill-advised moment to go public due to current safety concerns. OpenAI had hired bankers and lawyers with the goal of going public this year. The company is now leaning toward 2027 due to the volatility of tech stocks and its own financial challenges.

The push by AI companies to solve mathematical problems as a benchmark is detrimental to the science of mathematics. The goals of AI companies and the mathematical community are severely misaligned. This is a symptom of broader alignment issues that impact other scientific and creative professions, as well as the whole of society.

Construction is labor-intensive. There has, unsurprisingly, long been interest in automating the process to reduce the amount of labor required. However, automation has only been successful in tasks that can be done off-site in factories and on-site tasks that are extremely factory-like. Almost nothing outside of these two categories has succeeded.

Agents can generate code. Getting it right for your system is the hard part. More MCPs solve access but not understanding. Join us for a FREE webinar on Sep 23 to see how leading teams save time and tokens by giving their agents a context layer.

Forward-deployed engineers have the hottest job in AI right now. While labs, startups, and PE firms are all hiring engineers to sit inside their customers' operations and solve their problems, almost none of them agree on what those engineers are supposed to accomplish. This article looks at what works and what fails and tries to predict where the industry is heading. It covers misunderstandings about the role, its history, what the goal of the role is, and more.

Large language models are able to write almost perfectly correct code. It is pretty straightforward to get a model to generate code and then let that code be checked by hidden tests. Checking the sloppiness of this code often requires human intuition and time, and is an extremely difficult task in general. This post illustrates why this is by going through possible ways of measuring slop.

The company behind Roblox announced several new features at its annual Roblox Developer Conference, including new game-creation tools, expanded NPC capabilities, and the ability to make games available across platforms. The company is also launching a dedicated card and wallet for creators. Roblox is rolling out its generative AI game creation feature more broadly to Serbia and Singapore. It is also gaining desktop access, and later this year, it will get a new prompt-based scene-generating feature.

While AI has the potential to produce a long line of technological miracles, like many technologies before it, it brings risk. These risks include losing control of AI systems, the misuse of the technology for cyberattacks and bioterrorism, and serious economic disruption. A race to the bottom spurred by commercial incentives makes these risks more acute. Anthropic's CEO has called for the industry to prioritize caution over speed and prudence over profit. Addressing these risks will require pacing the rate of capabilities advancement so that risk prevention has time to keep up.

Access models from 80+ providers with better prices, automatic routing and fallbacks, and no subscriptions. Build with the right model for every request. See how it works.

Ads get ignored on social media but not in TLDR! Reach developers, PMs, marketers, founders and other tech leaders where they actually pay attention. Learn more about sponsorship opportunities.

Most prompts are bad because people only add to them over time rather than removing unnecessary instructions.

The future will see more software, more people making it, and none of them typing.

The speed of the developer experience will matter a lot more when generating tokens stops becoming the bottleneck.

Big prompts are brittle and hard to debug and evaluate, but they are the best choice in some cases.

Humans who want to use AI for interpersonal interactions will need to work hard to compensate for the loss of trust it engenders.

Tesla recently made a post that simply wrote, 'Go for launch,' with a graphic showing a '10.1' date.

Devs finally have a challenger to GitHub code hosting with Origin.

Origin was built on Buildkite.

Buildkite is a launch partner for this new code ecosystem. Origin holds the code. Buildkite runs the build. Vercel ships it. Three layers, one path from idea to production.

We also run CI for OpenAI, Anthropic, Mistral, xAI, Airbnb, Nvidia, Canva, Uber, Meta, Pierre, Bun and more.

Build faster, more securely, more reliably, with the CI engineered for frontier scale, used by the frontier labs setting the pace.

Start with our 30-day all-access trial.Get building →

OpenAI's Agents API is now in public beta. The API brings the same harness and infrastructure that powers Codex to developers through a simple and flexible API. It allows developers to create useful agents with a powerful harness that manages context, uses tools efficiently, and coordinates subagents. The API provides infrastructure to keep agents running reliably for days, and environments where agents can work with files, run code, and save intermediate results.

Anthropic claims it has disrupted several plots to use its AI models to conduct research that could have helped develop biological weapons. The company says it still can't determine whether a query serves a legitimate or nefarious purpose, so it erred on the side of caution as the consequences of mistakes could be severe. Anthropic detected use from companies from blocked regions and parties that otherwise tried to obfuscate their purpose.

Unitree CEO Wang Xingxing recently unveiled details about his company's IPO in an interview that covers his business principles, Unitree's managerial structure, and how the company is changing its employees' lives. Unitree listed on the STAR Market on August 19, with its market capitalization briefly reaching 440 billion yuan after its IPO. Wang was briefly the richest person in China born in the 1990s.

Analysts predict that AI will soon deliver double-digit growth in advanced economies. Experts say it could push growth to something like 10% to 15% per year. The decade ahead could see economic growth of 30% per year and beyond. Predictions of double-digit growth are common among AI insiders.

Ahrefs Brand Radar tracks your brand across ChatGPT, AI Overviews, Gemini, Perplexity and Copilot — from 471M+ real search-backed prompts. Benchmark competitors, see what AI cites, add your own prompts.

Explore Brand Radar

This post follows the journey of a PostgreSQL query through all of the layers of a complex sharded system. It aims to help readers understand what goes into making a large-scale sharded PostgreSQL deployment appear to be a single server even when spanning thousands of servers. A real database has hundreds of tables - the example in the post keeps the schema simple with two tables spread across four shards.

Companies are experimenting with cheaper open models on inference providers. There is a trend toward more investment in model routing to route simpler requests to cheaper models. The industry is seeing more knowledge-sharing sessions on how to use AI models cost-effectively. More and more companies are setting per-developer limits on AI usage.

Centralized router platforms can offer customers better economics than local deployments. Outcomes pricing belongs to the infrastructure layer. The model layer could offer outcomes pricing, but it will have the same problems as the app layer. Router layer pricing is most politically aligned with customers, who just want to sell more compute and don't care which model is used.

The 'next thing' is always simpler than you think and very obvious in hindsight. OpenAI's addition of Computer Use is deceptively simple and obvious in hindsight. 'Computer use' is a very generic, unspecified thing. It requires general intelligence across many domains to get better at it.

Your vendor says "OWASP LLM Top 10 covered." That's a checklist, not a defense. Seven red flags every developer should catch before signing.

Astra will change the world one way or another.

The people building AI earnestly believe that it could kill us all by the end of the decade.

AI can deidentify you better than many adversarial bot programs.

Most changes that would have made agents better at codebases make agents better too.

Google has six months to implement the technical measures and finalize the processes for data access.

A storage solution is only part of what a team needs to collaborate around code.

Lightfield is the CRM that updates itself. Yesterday, it announced a $47M Series A led by Andreessen Horowitz.

Here is why thousands of companies switched to Lightfield: it captures every email, call, and meeting automatically. No manual data entry, ever.

Ask "Which deals have gone quiet this month?" or "Prep me for my call with Acme” and it answers from your actual conversations, not fields you never filled in.

Teach Lightfield how you sell and watch it go to work for you.

Skeptical? Grab time with the team to see it for yourself

Apple's first foldable iPhone, the iPhone Duo, starts at $1,999. The device expands to a 7.6-inch screen when unfolded, and a 5.4-inch screen when used as a conventional handset. It comes in two colors: Star White and Night Sky. More details on the device, as well as Apple's other recent releases, are available in the article.

The Apple Watch Series 12 and Ultra models have new Audio Intelligence tools that use ambient listening to take high-level notes of wearers' conversations throughout the day. The devices will also have a Live Recap function that can show a transcription of what was said in a live conversation in the last 15 seconds and be able to listen to important sounds, such as a crying baby, doorbells, or sirens. These features will not create or store recordings, and Apple will not be able to access the raw audio. They will be available to test in English later this year. The improved Siri is being rolled out on watchOS 27 in beta on September 14.

Modern cancer-vaccine design involves identifying mutations, predicting presentation, predicting immunogenicity, ranking targets, and designing RNA. However, biology is complicated, and whether a target makes a good vaccine candidate depends on all kinds of information. This makes cancer vaccines an unusually direct test of whether AI models can take what it has learned from biology and apply that knowledge to design a medicine for one particular human being.

It is unknown yet whether AI will lead to unprecedented growth or widespread unemployment. Anthropic's Economics team built a model to understand how AI might affect jobs, growth, and unemployment in the US in the coming years. The model found that AI will lead to adverse impacts on wages and job prospects for knowledge workers in scenarios where growth is faster than anything in economic history, and in those scenarios, the challenge will be making sure wealth is broadly shared. The answer will depend on how AI affects all the tasks that make up the economy, the new tasks it creates, and how fast AI takes on this work.

When approval fatigue turns sign-offs into rubber stamps, you need more than good judgment to keep AI secure. Gray Swan's eBook lays out 5 governance implications for AI security, along with 3 questions every board should be asking right now. Get your copy

PostHog's Replay Vision lets models view session recordings. This post looks at how PostHog built the tool and how it implemented a method for the model to actually see replays. It covers how PostHog built a custom rasterizer that runs hundreds of concurrent jobs across many pods. The custom rasterizer has so far rasterized more than 370 years of recordings into roughly 3.5 million videos since it was deployed in production in March.

Tailwind is joining Shopify, one of the first companies operating at scale to see the potential of Tailwind CSS and start building with it. Shopify has invested in making sure Tailwind is actively maintained and will continue to provide a stable long-term home for the project. Nothing will change with Tailwind CSS or any of Tailwind's other open source projects, but the project will no longer try to grow its business. All existing customers will maintain access to their products, but Tailwind has closed signups for new customers.

The price of AI is coming down. However, volume growth may not be large enough to counteract the effective decline in prices. The models driving volume increases are relatively cheap, standard models. These trends do not appear to be driven by the adoption of open-source models or Chinese models.

Automattic's board has put founder and CEO Matt Mullenweg on a leave of absence against his will. The company's Chief Financial Officer, Mark Davies, had conspired with other board members to put Mullenweg on a paid leave of absence. Davies is now acting as interim CEO. Mullenweg remains the leader of the WordPress project, and the company's teams, priorities, and work will continue as planned. It is unclear why the board took this action. Mullenweg remains on the board of directors.

The Granola MCP integration connects your meeting notes to your AI agent. Have your agent review meetings, update your CRM, and organize tasks in Linear. Free month with code: TLDR1MO

A system that improves itself needs guardrails and a sandbox.

Purchasers of open source work should be able to check whether a vendor contributes.

There are massive implications of a race towards superintelligence, and not enough of Congress is focusing on them.

While GPT-6 Astra and Claude Fable 5.1 seem to have the same rates, Fable's cache read is four times cheaper.

View Transitions and Fragment Refs are now stable in React 19.3.

OpenAI heard that some Millennium Prize problems had been solved using large language models and saw it as an opportunity to demonstrate the power of its latest model without regard for the teams who had been working on the problem already.