1. X
  2. Anderson A.
Log inSign up
Anderson A.
1,614 posts
user avatar
Anderson A.
@aacodex401
20yrs+ Experienced AI Product Leader. Product Founder of Abittus, Ali Credito, @widbook. Fractional Expert at @trustdotshq
Planet Earth
abittus.com
Joined August 2007
223
Following
237
Followers
RepliesRepliesMediaMedia
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

  • user avatar
    Anderson A.
    @aacodex401
    Mar 10
    If you're still building your own AI agent plumbing, you're now competing with Accenture's checkbook. Here's what most AI founders are getting wrong — and what to do instead.
    115
  • user avatar
    Anderson A.
    @aacodex401
    Feb 17
    Didn't plan to become a fractional Product Leader. Ran a startup that failed with 500K users. Moved across insurance, automotive, fintech. Each felt like a detour. It wasn't. It was a portfolio compounding in the background. Andreessen's framing is right: the question isn't
    122
  • user avatar
    Anderson A.
    @aacodex401
    Feb 17
    Last week in AI: GPT-5 preview: 2M token context window. Gemini 3 Deep Think: 48.4% on Humanity's Last Exam. Anthropic: 20% price cut. The market is splitting into tiers. Frontier models for hard reasoning. Cost-optimized for volume. Enterprise agents framed as headcount
    139
  • user avatar
    Anderson A.
    @aacodex401
    Feb 14
    Robots that think while they move. TIC-VLA: Real-time navigation with language understanding. The gap between semantic reasoning and reactive control is closing. Practical robotics, not sci-fi. arxiv.org/abs/2602.02459 #Robotics #AI
    arXiv logo
    arxiv.org
    TIC-VLA: A Think-in-Control Vision-Language-Action Model for Robot...
    Robots in dynamic, human-centric environments must follow language instructions while maintaining real-time reactive control. Vision-language-action (VLA) models offer a promising framework, but...
    105
  • user avatar
    Anderson A.
    @aacodex401
    Feb 14
    80% of AI startups will die by 2027. Not because models are bad. Because founders built tech demos, not businesses. If a student can clone your "AI product" in a weekend, you're arbitraging OpenAI's API. That's not defensibility. That's a timer.
    Why 90% of AI Startups Are Doomed by 2026 (And How the 1% Survive)
    From pixicstudio.medium.com
    135