Algomizer | LLM Optimization

We are grateful to be included in the a16z GEO thesis and to see them shedding light on just how critical this layer of discovery has become. LLMs now mediate an increasing share of discovery and decision moments. In these environments, visibility is finite. A small number of outputs guide the user’s next action. Absence at this layer does not reduce visibility, it *removes* the brand from consideration entirely for that interaction. Discovery today increasingly means being selected by the model at the right moment, in the right context. Understanding how that selection happens, and how to influence it, is becoming essential for any brand that wants to stay competitive as LLM-driven behavior accelerates.

Anthropic's shipping pace is insane right now. In a few weeks they've pushed Claude Code channels, persistent memory, autonomous cron jobs, a security tool that finds and fixes bugs on its own, 1M context, new models, MCP integrations, and now you can text your agent from Telegram or Discord. They're building something you don't open anymore. It just runs and you ping it from your phone when you need to.

We just released Claude Code channels, which allows you to control your Claude Code session through select MCPs, starting with Telegram and Discord. Use this to message Claude Code directly from your phone.

Most robotics teams split the thinking and the moving into two separate systems that talk to each other. Tesla is running both as one model. That matters because in the real world, deciding to step over something and controlling how your foot lands can't have a delay between them. Safety falls apart when your planning runs at a different speed than your motor control. And Tesla has years of weird edge cases from self-driving that taught them this. No other humanoid robotics team has that dataset.

Ashok Elluswamy, Tesla's AI lead, during a GTC discussion, highlighting the fundamental similarity in AI approaches for self-driving cars and humanoid robots: - Hierarchical decision making is useful, but it has to be done as part of the same decision-making process as lower-level controls. - We haven't seen the long tail of humanoid robotics, but Tesla has seen the long tail of self-driving, where high and low-level decisions have to be jointly made at a pretty high framerate. - Optimus's architecture is designed in a similar way, where there's a hierarchy but it's all running as part of the same model and the latencies involved in decision making are well modeled. - This architecture will scale quite well with humanoid robots. - The distinction of the decision-making levels is only in the developer's mind. For the model, it's a continuous space of decision making, where there are dials available to make them more fine or coarse. - Humanoids have more sensor modalities and higher degrees of freedom compared to self-driving, but the fundamental constraints remain the same: you need to make real-time decisions. There's obviously a hierarchy to these control signal outputs, but the lowest frequency cannot be too low, because the safety of the robot cannot depend upon things running at very low frequencies.

Most vibe coding tools stop at the frontend. You get a nice-looking UI with no backend, no database, no auth. Then you spend hours wiring everything together yourself. This goes from prompt to full-stack deployed app with live data, authentication, and multiplayer. That's a different game.

We’re launching a brand new, full-stack vibe coding experience in @GoogleAIStudio, made possible by integrations with the @Antigravity coding agent and @Firebase backends. This unlocks: — Full-stack multiplayer experiences: Create complex, multiplayer apps with fully-featured UIs and backends directly within AI Studio — Connection to real-world services: Build applications that connect to live data sources, databases, or payment processors and the Antigravity agent will securely store your API credentials for you — A smarter agent that works even when you don't: By maintaining a deeper understanding of your project structure and chat history, the agent can execute multi-step code edits from simpler prompts. It also remembers where you left off and completes your tasks while you’re away, so you can seamlessly resume your builds from anywhere — Configuration of database connections and authentication flows: Add Firebase integration to provision Cloud Firestore for databases and Firebase authentication for secure sign-in This demo displays what can be built in the new vibe coding experience in AI Studio. Geoseeker is a full-stack application that manages real-time multiplayer states, compass-based logic, and an external API integration with @GoogleMaps 🕹️

this is misleading without context. when US companies spend billions on AI, most of that money goes to chips and hardware made in Taiwan and South Korea. GDP counts domestic production, so the investment shows up on one line but gets cancelled out by imports on another. AI isn't failing. The economic payoff is just landing in Asia, not America...

"Massive investment in AI contributed basically zero to US economic growth last year," per Goldman Sachs

Bezos recently became co-CEO of Project Prometheus, an AI startup building models that simulate the physical world. Raised $6.2B. The $100B fund is how he puts that tech to work. Buy manufacturing companies in aerospace, chipmaking, defense. Plug in the AI. Each acquisition becomes a customer for Prometheus and a training ground for the models. Same playbook he ran at Amazon with logistics. Own the infrastructure, use it yourself, then sell it to everyone else.

BREAKING: Jeff Bezos is reportedly in talks to raise $100B for a new fund aimed at acquiring manufacturing firms and automating them with AI, per WSJ.

So JPMorgan recently cut the value of tech loans they hold as collateral for private credit funds. Their own risk models are projecting a 3-5% jump in tech-loan defaults through 2027. And when JPMorgan marks deals down, other lenders start feeling the pressure to do the same. Now private credit is illiquid. You can't short it the way you'd short a stock. So what did Goldman and JPMorgan do? They put together baskets of publicly traded companies that have exposure to the space, think BDCs and alt managers, and gave hedge funds a way to bet against private credit through liquid proxies. The banks built the lending machine. Now they're selling the hedge against it.

BREAKING: Goldman Sachs and JPMorgan are offering hedge fund clients ways to short the $1.8 trillion private credit market.

It holds across every region at the final decision stage too + North America: 17.2% + APAC: 19.2% + EMEA: 15.8% And 97.8% of the citations driving that influence come from content you don't own. AI surfaces what third parties say about you, review sites, comparison pages, niche publications in your category. Your own website barely factors in. The brands showing up in those answers are getting to buyers before sales ever does. Want to see where your brand stands? Get your free visibility report 👇

AI assistants are now the #1 source influencing B2B vendor shortlists. @G2dotcom surveyed 1,169 B2B Tech decision makers in April 2025. Here's exactly where buyers are going when they start evaluating vendors: AI assistants: 17.1% Software review sites: 15.1% Vendor websites: 12.8% Peers and colleagues: 8.9% Vendor salespeople: 8.8% AI assistants ranked above your website, your sales team, and the analyst firms you pay to get featured in. The buyer's first filter is now a prompt. And buyers who find you through AI convert 40% better because they arrive already knowing what they want.

It holds across every region at the final decision stage too + North America: 17.2% + APAC: 19.2% + EMEA: 15.8% And 97.8% of the citations driving that influence come from content you don't own. AI surfaces what third parties say about you, review sites, comparison pages, niche publications in your category. Your own website barely factors in. The brands showing up in those answers are getting to buyers before sales ever does. Want to see where your brand stands? Get your free visibility report 👇

AI assistants are now the #1 source influencing B2B vendor shortlists. @G2dotcom surveyed 1,169 B2B Tech decision makers in April 2025. Here's exactly where buyers are going when they start evaluating vendors: AI assistants: 17.1% Software review sites: 15.1% Vendor websites: 12.8% Peers and colleagues: 8.9% Vendor salespeople: 8.8% AI assistants ranked above your website, your sales team, and the analyst firms you pay to get featured in. The buyer's first filter is now a prompt. And buyers who find you through AI convert 40% better because they arrive already knowing what they want.

Three consecutive weeks of inflows at that scale is probably signalling positioning. 75% of flows go through Bitcoin ETFs, driven by institutional allocation. Recovery of prior outflows helps stabilize the market, and the AUM growth during uncertainty fits crypto acting as a liquid global asset. Bitcoin leading while Ethereum stays flat matches a more conservative allocation shift. this is early re-accumulation driven by institutions.

Crypto market momentum is accelerating: Crypto funds recorded +$1.06 billion in inflows last week, the highest since the 3rd week of January. This marks the 3rd consecutive weekly intake, bringing the total to +$2.8 billion. This now recovers most of the -$3.9 billion in outflows from the prior 5-week selling streak. Bitcoin ETFs led with +$793 million, making up 75% of total inflows, bringing the 3-week inflows to $2.2 billion. Ethereum saw +$315 million, bringing year-to-date net flows close to zero. Since the start of the Iran War, total crypto ETF assets under management have risen +$12 billion, or +9.4%, to $140 billion. Investors are returning to crypto despite the geopolitical turmoil.

I mean, the scale of disruption you’re pointing at is real. If a region that central to energy gets destabilized, it pushes through everything. Prices, supply chains, basic goods. Capital would shift pretty quickly into essentials. Energy, food, infrastructure. Funding for tech tightens because uncertainty just kills risk appetite. And it definitely forces a faster move toward energy independence. Not because anyone planned it cleanly, just because there’s no choice once supply becomes unstable. It’s the kind of situation where everything starts repricing at once, and that pressure spreads globally.

I'm going to make some obvious points. (1) Blowing up all the oil infrastructure in the Middle East is an insane idea, and may well result in a global economic crash and humanitarian crisis unrivaled in the lives of those now living. We're talking about the price of everything everywhere rising, from food to gas, at a moment when inflation was already high. All of that will be laid at the feet of the authors of this war. (2) The antebellum status quo of Feb 27, 2026 was just not that bad, but we're unlikely to return to it. Expect indefinite, long-term, ongoing disruptions to everything out of the Middle East. (3) Also assume tech financing crashes for the indefinite future. The genius plan to get the Gulf states caught in the crossfire has incinerated much of the funding for LPs, for datacenters, and for IPOs. Anyone in tech who supported this war may soon learn the meaning of "force majeure" as funding gets yanked. (4) Many capital allocators will instead be allocating much further down Maslow's hierarchy of needs, towards useful basic things like food and energy. (5) It's fortunate that all those progressives yelled about the "climate crisis." Yes, their reasoning about timelines was wrong, and much of the money was wasted in graft, but the result was right: we all need energy independence from the Middle East, pronto. It's also fortunate that Elon and China autistically took climate seriously. Now they're going to need to ship a billion solar panels, electric vehicles, batteries, nuclear power plants, and the like to get everyone off oil, immediately. (6) It's not just an oil and gas problem, of course. It's also a fertilizer problem, and a chemical precursor problem. Maybe some new sources will come online at the new prices, but it takes time to dial stuff up, particularly at this scale, so shortages are almost a certainty. That said, China has actually scaled up coal-to-chemicals[a,c] (C2C), and there's also something more sci-fi called Power-to-X[b] which turns arbitrary power + water + air into hydrocarbons. But all of that will need to get accelerated. I have a background in chemical engineering so may start funding things in this area. (7) Ultimately, this war is going to result in tremendous blame for anyone associated with it. It's a no-win scenario to blow up this much infrastructure for so many people. Simply not worth it for whatever objective they thought they were going to attain. But unless you're actually in a position to stop the madness, the pragmatic thing to do is: scramble to mitigate the fallout to yourself, your business, and your people. [a]: reuters.com/business/energ… [b]: alfalaval.com/industries/ene… [c]: reuters.com/sustainability…

If a model can run multiple self-improvement loops without human tuning, that truly changes how progress happens. You set direction, it handles iteration, and then the pace just accelerates from there. Also the hardware point is very interesting. If this runs on lighter setup, way more teams can actually experiment with it. And once it starts contributing to its own research, that feedback loop compounds pretty quickly. This is basically what recursive improvement looks like when it starts working for real.

MiniMax-M2.7 just landed in MiniMax Agent. The model helped build itself. Now it's here to build for you. ↓ Try Now: agent.minimax.io

"peanut butter" raises means spreading thin, uniform increases across everyone instead of meaningful performance-based pay. nearly half of companies doing this mirrors exactly what happened in 2008 when companies were conserving cash without pulling the trigger on layoffs. it's a pretty telling signal about where corporate confidence actually sits right now, regardless of what earnings calls are saying.

Nearly half of companies are opting for weak “peanut butter” pay raises, mirroring the 2008 recession trend, per FORTUNE.

Iran striking Ras Laffan hours after the Israeli strikes on their gas infrastructure tells you the energy asset phase of this conflict has started. the "legitimate targets" declaration is what changes the calculus. Saudi Aramco, UAE facilities, everything in the region is now in a different risk category than it was yesterday. the LNG angle hits Europe harder than most people are tracking. Europe rebuilt its entire energy strategy around LNG after 2022. a sustained disruption to 20% of global supply lands very differently in Berlin and Paris than it does in Washington.

BREAKING: Qatar reports "extensive damage" from an Iranian missile strike at Ras Laffan, the world's largest LNG facility. Details include: 1. Ras Laffan accounts for ~20% of global LNG supply 2. The plant was reportedly lit on fire as a result of the strike 3. The attack comes just hours after Israeli strikes on Iran's largest natural gas plant 4. Iran warned that a number of energy assets across the Gulf are now “legitimate targets” Natural gas prices are surging on the news.

Your brand might be ranking on Google and completely invisible in AI answers. The reason usually has nothing to do with content volume or SEO fundamentals. It comes down to how AI systems actually cite and recommend brands, and most companies have no visibility into that process. There are three types of citations that actually drive business outcomes: i. Direct recommendations: an LLM suggests your solution unprompted, during a planning or research session ii. Comparative mentions: your brand shows up in side-by-side evaluations when someone asks "what's the best tool for X" iii. Implementation guidance: your product gets woven into step-by-step workflows when someone asks how to do something Each one requires a different approach: + Direct recommendations come from strong brand-problem associations across trusted third-party sources. + Comparative mentions come from clear differentiation signals in category-level content. Implementation guidance comes from structured, detailed documentation that LLMs can parse and reference. We track citation frequency, context quality, and recommendation strength. Those metrics tell you whether AI systems are positioning you as the answer or just listing you as an option. Most companies are optimizing for search engines while their buyers are already getting answers from ChatGPT, Claude, and Perplexity. The discovery channel shifted. The optimization strategy needs to follow. Want to see where your brand stands in AI answers? Get your free visibility report here. (link in the comments)

Your SAAS might be ranking on Google and completely invisible in AI answers. The reason usually has nothing to do with content volume or SEO fundamentals. It comes down to how AI systems actually cite and recommend brands, and most companies have no visibility into that process. There are three types of citations that actually drive business outcomes: i. Direct recommendations: an LLM suggests your solution unprompted, during a planning or research session ii. Comparative mentions: your brand shows up in side-by-side evaluations when someone asks "what's the best tool for X" iii. Implementation guidance: your product gets woven into step-by-step workflows when someone asks how to do something Each one requires a different approach: + Direct recommendations come from strong brand-problem associations across trusted third-party sources. + Comparative mentions come from clear differentiation signals in category-level content. Implementation guidance comes from structured, detailed documentation that LLMs can parse and reference. We track citation frequency, context quality, and recommendation strength. Those metrics tell you whether AI systems are positioning you as the answer or just listing you as an option. Most companies are optimizing for search engines while their buyers are already getting answers from ChatGPT, Claude, and Perplexity. The discovery channel shifted. The optimization strategy needs to follow. Want to see where your brand stands in AI answers? Get your free visibility report here. (link in the comments)

GPT-5.4 mini is genuinely impressive. 88% on GPQA Diamond. That's expert-level scientific reasoning on a mini model. A year ago that number would've been flagship territory. And it's scoring 54.4% on software engineering benchmarks, 72.1% on computer use, 57.7% on tool calling. All while being 2x faster than the previous mini. The part that actually matters for anyone building with AI is that the gap between mini and flagship keeps shrinking. What required a full GPT-5.4 call a few months ago can now run on mini at a fraction of the cost. That changes the economics of a lot of products pretty quickly.

We’re introducing GPT-5.4 mini and nano, our most capable small models yet. GPT-5.4 mini is more than 2x faster than GPT-5 mini. Optimized for coding, computer use, multimodal understanding, and subagents. For lighter-weight tasks, GPT-5.4 nano is our smallest and cheapest version of GPT-5.4. openai.com/index/introduc…