AWS + OpenAI 🤝, Claude creative tools 🧑‍🎨, Nemotro Nano Omni 🤖

OpenAI CEO Sam Altman and AWS CEO Matt Garman recently sat down for an interview where they talked about Bedrock Managed Agents powered by OpenAI ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ 

Sign Up |Advertise|View Online TLDR Together With TLDR AI 2026-04-29 Webinar: How to Design Pricing for AI Agents With Metronome and Kyle Poyar (Sponsor) Decoy effects, good‑better‑best tiers, and $9.99 vs. $8.97: every convention in SaaS pricing was designed for human buyers, but the buyer is changing. What happens when AI agents are the ones building the shortlist and making comparisons? They don't respond to the same nudges and hacks - and they can research all day, if needed. This Metronome webinar with Growth Unhinged's Kyle Poyar tackles agent-native pricing strategy and the billing infrastructure needed to support it. Drawing on recent moves from Clay, Figma, and PostHog, it'll explore why many leading companies have committed to two-track billing amidst the shift to agentic billing.  Save your seat 🚀 Headlines & Launches An Interview with OpenAI CEO Sam Altman and AWS CEO Matt Garman About Bedrock Managed Agents (50 minute read) OpenAI CEO Sam Altman and AWS CEO Matt Garman recently sat down for an interview where they talked about Bedrock Managed Agents powered by OpenAI and how that would fit in with OpenAI's deal with Microsoft giving Azure exclusive access to OpenAI models. Microsoft and OpenAI have since amended their agreement to allow OpenAI to serve its products on other cloud providers. Azure's exclusivity was actively damaging Microsoft's investment in OpenAI, so Microsoft needed to do something, even if it meant diminishing Azure's differentiation. OpenAI also released Microsoft from the AGI clause, which means the agreement between the two companies will run until 2032 even if OpenAI achieves AGI. Claude Connectors for Creative Tools (4 minute read) Anthropic introduced connectors that integrate Claude with major creative software like Adobe, Blender, and Autodesk, enabling natural-language workflows, automation, and cross-tool pipelines for design, 3D, and audio production. 🧠 Deep Dives & Analysis Can agents replace the search stack? (7 minute read) Giving agents basic search tools results in better quality answers. Encouraging the agent to explore further creates more improvements. Agentic search models can figure out how to surface relevant results, but the training data still dictates the models' sweet spots. Large language models can't evaluate what they don't know, and there's still no real way to compensate for their lack of knowledge. Opus 4.7's New Tokenizer: What It Actually Costs (6 minute read) Anthropic improved Claude Opus 4.7's understanding of inputs with a new tokenizer. While the model price hasn't changed, the same inputs now cost more than previous models. Costs have increased between 12% to 27%, except for short prompts, which have actually become more cost-efficient. OpenAI's Q4 2026 IPO Might not Happen (9 minute read) OpenAI's CFO allegedly doesn't report to the CEO and reportedly has been excluded from financial discussions involving server procurement. The company has committed to enormous infrastructure obligations, but the CFO and board are unsure whether those commitments can be funded. These factors, along with the compressed IPO timeline, make it unlikely that OpenAI will be able to IPO this year. 🧑‍💻 Engineering & Research The Trust Problem With AI Agents (Sponsor) AI agents are getting more capable, but developers are trusting them less. The problem isn't the technology, but the design of the tools they're using. Here's why human-in-the-loop actually matters, and what it should look like. Join the waitlist and follow the 5-part series. Introducing NVIDIA Nemotron 3 Nano Omni: Long-Context Multimodal Intelligence for Documents, Audio, and Video Agents (11 minute read) NVIDIA's Nemotron 3 Nano Omni is a new multimodal model for document, audio, and video analysis, achieving best-in-class accuracy on benchmarks like MMlongbench-Doc and VoiceBench. It utilizes a hybrid Mamba-Transformer architecture integrated with specialized encoders for vision and audio, allowing efficient processing across long multimodal contexts. This model significantly improves throughput and reasoning speed for real-world applications such as document analysis, automatic speech recognition, and video understanding. Laguna XS.2 and M.1: A Deeper Dive (20 minute read) Laguna XS.2 and Laguna M.1 are agentic coding models built for long-horizon work. Laguna M.1 is the foundation for the Laguna family of models. Laguna XS.2 is a much smaller model but is still remarkably capable for its size. Both models are free to use for a limited time via Poolside's API and on OpenRouter. The weights for Laguna XS.2 have been released under an Apache 2.0 license. The Recurrent Transformer: Greater Effective Depth and Efficient Decoding (5 minute read) The Recurrent Transformer is a novel architecture designed to overcome the "temporal shallowness" of standard Transformers. By introducing layerwise recurrence, the model uses a combination of temporary and persistent key-value pairs to allow each layer to maintain a continuously updating internal memory. This mechanism enables the model to perform complex, iterative reasoning across a sequence without requiring massive layer stacks. 🎁 Miscellaneous AI Worries Have Returned to Wall Street. Now Come Earnings (9 minute read) Shares of several firms tied to OpenAI dropped on Tuesday after news that OpenAI had missed its own targets for revenue and users surfaced. Investors are worried that OpenAI's massive investment in AI may not produce the blockbuster profits many expect. OpenAI critics have said that certain financing arrangements the company has made are circular in nature, with the company's partners providing funding and the company spending money on computing with that partner. The company has defended its financial footing and says its leaders are aligned on securing computing resources. Meta's new AI model shows early promise, but investors want to see Zuckerberg's strategy (5 minute read) Meta's new AI model, Muse Spark, signals a shift in its AI strategy, moving from open-source to paid access, aiming to boost its ad business and compete in the growing AI market. Analysts are cautiously optimistic, noting Meta's internal testing and investments, despite trailing top performers like Anthropic's Claude. Meta's workforce reduction and strategic hires reflect its intensified focus on AI advancements and infrastructure. ⚡ Quick Links Why multi-agent systems work in demos but break in production (Sponsor) As agentic systems grow, coordination, integrations, testing, and data handling become architectural problems. Explore how neuro-san is designed to handle these constraints so multi-agent systems remain scalable and reliable. Read post. Ex-Twitter CEO's AI Startup Raises Funds at $2 Billion Valuation (5 minute read) Parallel Web Systems, a startup founded by Parag Agrawal, the former chief executive of Twitter, has developed a platform for AI agents to search the web. ElevenLabs launches Agent Templates for faster bootstrapping (2 minute read) Agent Templates on the ElevenAgents platform enables quick deployment of AI agents with pre-built frameworks. Become a curator for TLDR AI (3-5 hrs/week) TLDR is looking for an engineer/researcher at a major AI lab or startup to help write for 1M+ subscribers. Our curators have been invited to Google I/O and OpenAI DevDay, scouted for Tier 1 VCs, and get early access to unreleased TLDR products. Learn more. Google Grants DoD Broad AI Access (1 minute read) Google agreed to provide the US Department of Defense access to its AI on classified networks for broad lawful uses. Meta Sapiens2 Human-Centric Vision Models (GitHub Repo) Meta Sapiens2 is a family of high-resolution transformers trained on 1B human images that achieves strong results across tasks like pose estimation, segmentation, and surface prediction. Love TLDR? Tell your friends and get rewards! Share your referral link below with friends to get free TLDR swag! https://refer.tldr.tech/0b6a6dc1/2 Track your referrals here. Want to advertise in TLDR? 📰 If your company is interested in reaching an audience of AI professionals and decision makers, you may want to advertise with us. Want to work at TLDR? 💼 Apply here, create your own role or send a friend's resume to jobs@tldr.tech and get $1k if we hire them! TLDR is one of Inc.'s Best Bootstrapped businesses of 2025. If you have any comments or feedback, just respond to this email! Thanks for reading, Andrew Tan, Ali Aminian, & Jacob Turner Manage your subscriptions to our other newsletters on tech, startups, and programming. Or if TLDR AI isn't for you, please unsubscribe.