AcceleratingBiz®

Growth CEO’s AI Briefing · Edition #130

Crowded Frontier

Tuesday, October 13, 2026 · 2 Developments · 2 Perspectives · 1 Case Study · 2 Resources

Designed by AcceleratingBiz®. Curated, synthesized, and produced by Claude agents. AI-assisted production means occasional errors may appear.

Two new frontier models arrived in two days, both priced to match the leader rather than undercut it. That shifts the buying question from which lab is best to which terms you can live with.

  • Muse Wonder. Meta's first frontier models match Claude Opus 5.5's price, and the stronger one comes only in plans that train on your data by default.
  • Gemini 4 Argon. Google's new flagship is half price until November 30, then rises to the same $4 and $20 per million tokens.
  • Karpathy's funnel. Andrej Karpathy estimates only a sliver of users see what frontier AI can do, likely including your workforce.
  • Moats test. Dan Hockenmaier argues only four advantages compound, and AI has not removed the cost of relearning a workflow.
  • Epic's mandate. An order for no hand-written code on Epic Games' Disney project shows the gap between a directive and team buy-in.
  • Marketing reality. MarTech finds most marketers still copy and paste AI output, and fewer can prove ROI than a year ago.
  • Gemini citations. A free tool measures how often Gemini cites your page across repeated runs.

At equal prices, your edge is how well teams use models and what contracts allow.

Developments

Product & Technology

Meta's Muse Wonder Reaches the Frontier at Opus Prices (October 12, 2026)

On October 12, Meta released Muse Wonder and Muse Wonder Full, its first models aimed at the top tier, built for agents with memory that carries across conversations. On the independent Artificial Analysis index, Wonder Full scores 56 and Wonder 52, behind Claude Opus 5.5 at 58. Wonder costs $4 per million input tokens and $20 per million output tokens on Meta's API, the same as Opus 5.5. Wonder Full is sold only in Muse Code plans at $50 or $200 a month, which default to a mode that lets Meta train on what users submit.

So what

A fourth frontier lab adds leverage in model negotiations, but nothing here is a reason to switch on price or quality. The exposure is data: anyone on your team trying Muse Code should turn off Contributor mode before pasting code or client files, since the opt-out covers only future submissions. Add training defaults to your AI tool approval checklist. Revisit if Wonder Full reaches the API, where Meta says its prices will sit below rivals' top models.

Hype check

Meta's own chart puts Wonder Full ahead on 7 of 14 benchmarks, mostly long coding and legal tasks, while independent scoring ranks it second. One developer's launch-day agent test ran six hours and about $96 without clearing an early game milestone; that is one run, not a benchmark.

Confidence: Established fact — launch, prices and plan terms come from Meta and VentureBeat; relative quality rests on early benchmarks.

Go deeper Meta · The Neuron

Product & Technology

Google Ships Gemini 4 Argon, Half Price Until November 30 (October 13, 2026)

Google DeepMind released Gemini 4 Argon on October 13, its first major flagship since Gemini 3 Pro last November. It is available through the Gemini API, Gemini for business and the Gemini app for paid subscribers. Introductory pricing is $2 per million input tokens and $10 per million output tokens until November 30, when it rises to $4 and $20. Every's reviewers, after a week of testing, call it a breakthrough for Google-only shops but say its desktop app lags the model.

So what

For companies standardized on Google Workspace, the gap with Claude and ChatGPT just narrowed, which may ease pressure to approve a second vendor. Because the discount ends November 30, model any evaluation on the $4 and $20 list price, not the launch price. Ask your team to run Argon on two or three recurring tasks this month and compare cost per completed task. Treat Google's benchmark claims, including first place on the Vals Index, as its own until outsiders confirm them.

Confidence: Established fact — release, availability and pricing come from Google's announcement; quality claims rest on Google and Every's early testing.

Go deeper Google · Every

Perspectives

AI Strategy & Transformation

Karpathy's "Sharp Funnel": Few Users See What Frontier AI Can Do (October 11, 2026)

In an October 11 post on X, OpenAI co-founder Andrej Karpathy described AI adoption as a narrowing funnel rather than two camps. By his own napkin math, about 6 billion people have barely used chatbots, 1 to 2 billion use free tools as a better search engine, about 20 million professionals use frontier models for weeks-long work, and about 5,000 lab insiders see the most advanced systems. He warned that "human review and comprehension are starting to fall behind."

So what

The same funnel likely runs through your company: a few power users on paid frontier tools and many employees judging AI by free-tier answers. Leaders who calibrate on the free experience will underinvest, and those who calibrate on demos will overpromise. Find out how many of your people use paid frontier tools on real work each week, and have your strongest users show their workflows to the rest. Treat Karpathy's figures as illustration, not data.

Confidence: Informed perspective — a respected researcher's estimate that he calls rough; figures confirmed via Benzinga's report of the post.

Go deeper Andrej Karpathy on X · The Rundown AI

Business

Dan Hockenmaier: Only Four Advantages Compound, and AI Has Not Erased Switching Costs (October 13, 2026)

In an October 13 guest post for Lenny's Newsletter, Dan Hockenmaier argues that only four moats compound over time: switching costs, network effects, scale economies and data effects. Speed, distribution and brand are fuel that gets a company moving, he says, not lasting defenses. He grants that AI makes migration cheaper but argues the larger switching cost is the "workflow tax" of relearning a product and rebuilding how a team works. His test is what the team actually builds every day, not what the strategy deck says.

So what

This is a useful counterweight to recent arguments that AI makes leaving a vendor cheap, including Michael Nygard's case in our October 9 edition. Name which of the four advantages your company is building and point to the roadmap work that strengthens it. If the honest answer is speed or brand, that is fuel, not a moat. For SaaS companies, deeper workflow embedding and integrations are the most direct defense.

Confidence: Informed perspective — a framework from an experienced marketplace operator; the measurement step sits behind a paywall.

Go deeper Lenny's Newsletter · Superhuman AI

Case Studies

People & Organization

Epic Games Orders "No More Hand-Written Code" on Its Disney Game (October 12, 2026)

Bloomberg reported on October 12 that a director on Epic Games' Disney project told developers the game would be made "AI first," with no hand-written code and all art produced through AI workflows. The game, rumored for November, has slipped to March, though Engadget notes the delay came before the directive. Some developers reportedly hoped the order was not serious, and an Epic spokesperson described the approach as "AI-accelerated development, led by Epic developers." A GDC survey found 52% of game developers now call generative AI bad for their industry, up from 30% in 2025.

So what

Blanket mandates send a clear signal but can cost the trust of the people expected to carry them out. If you are pushing AI adoption, set targets by workflow and outcome rather than banning the old way of working. Pair any mandate with a named owner for quality, since rework from AI output lands on the same team. Track team sentiment alongside throughput.

Confidence: Watch this — based on Bloomberg's sources and an internal message; Epic has not said how widely the rule applies.

Go deeper Bloomberg · The Neuron

Resources

Go-to-Market

MarTech's Q4 Read: Marketers Use AI Widely but Still Copy and Paste (October 12, 2026)

MarTech's Q4 2026 state-of-AI review says more than 90% of marketers report using AI, mostly for content, but most still work in standalone chatbots and paste results into other tools. AI features built into each marketing app tend to create "AI-powered silos" rather than connected workflows. Marketers report that reviewing high volumes of AI output adds work, and fewer are confident they can prove AI ROI than in 2025.

So what

Use this as a checklist for your next marketing review. Ask your CMO which AI workflows run end to end without manual copying, what share of output gets reworked, and how ROI is measured net of review time and integration cost. Fix data quality before scaling volume, since AI amplifies bad inputs. More content is not the goal if customers did not ask for it.

Confidence: Informed perspective — an editor's synthesis of industry reports and interviews; the article does not name its surveys.

Go deeper MarTech · TLDR Marketing

Go-to-Market

A Free Checker Shows How Often Gemini Cites Your Page (October 12, 2026)

Suganthan Mohanadasan released a free Gemini Grounding Gap Checker that asks Gemini the same question five times with Google Search on and counts which sources and claims recur. It compares those claims with your page's raw HTML, labels each as stated, touched or missing, and suggests edits. His tests show why repeat runs matter: one page went from cited in 3 of 5 runs to 1 of 5 an hour later. In his September tests, publishers blocking Google-Extended in robots.txt were cited in none of 10 runs.

So what

Yesterday's Claude citation study showed that page format matters; this tool lets you measure Gemini on your own buyer questions. Have marketing run your top five category questions and fix the claims marked missing on key pages. Check that your robots.txt does not block Google-Extended unless that is a deliberate choice. Results reflect Gemini's API, not AI Overviews or ChatGPT.

Confidence: Informed perspective — a practitioner's tool with small-sample tests, open about its limits.

Go deeper Suganthan Mohanadasan · TLDR Marketing

Sources in this edition: The Neuron, The Rundown AI, Superhuman AI, The Deep View, Every, Lenny's Newsletter, TLDR AI, TLDR Marketing, Daily AI, AINews

← Latest Editions and Search