What 375 AI Builders Actually Ship

What 375 AI Builders Actually Ship

70% of production AI teams use open source models. 72.5% connect agents to databases, not chat interfaces. This is what 375 technical builders actually ship - & it looks nothing like Twitter AI.

Article content

70% of teams use open source models in some capacity. 48% describe their strategy as mostly open. 22% commit to only open. Just 11% stay purely proprietary.

Article content

Agents in the field are systems operators, not chat interfaces. We thought agents would mostly call APIs. Instead, 72.5% connect to databases. 61% to web search. 56% to memory systems & file systems. 47% to code interpreters.

The center of gravity is data & execution, not conversation. Sophisticated teams build MCPs to access their own internal systems (58%) & external APIs (54%).

Article content

Synthetic data powers evaluation more than training. 65% use synthetic data for eval generation versus 24% for fine-tuning. This points to a near-term surge in eval-data marketplaces, scenario libraries, & failure-mode corpora before synthetic training data scales up.

The timing reveals where the stack is heading. Teams need to verify correctness before they can scale production.

Article content

88% use automated methods for improving context. Yet it remains the #1 pain point in deploying AI products. This gap between tooling adoption & problem resolution points to a fundamental challenge.

The tools exist. The problem is harder than better retrieval or smarter chunking can solve.

Teams need systems that verify correctness before they can scale production. The tools exist. The problem is harder than better retrieval can solve.

Context remains the true challenge & the biggest opportunity for the next generation of AI infrastructure.

Explore the full interactive dataset here or read Lauren’s complete analysis.

Interesting to see the prevalence of database connections over chat interfaces in production AI. The reliance on open source models also stands out.

Like
Reply

Seeing those numbers reminds us how much real world AI work happens behind the scenes, with open source models and database linked agents driving true value for businesses.

Interesting to see the focus on database connections over chat interfaces for production AI. It reflects a pragmatic, data-driven approach to real-world applications.

This is embarrassing. Literally 9/10 of the comments on this post use the same prompt to generate a reply. 👇 1. Interesting to see what's really happening in production AI... 2. Interesting to see real-world AI deployments leaning so heavily on open source 3. Interesting to see open source and direct database connections  4. Interesting to see the database connection preference among production teams. 5. Interesting to see data that shows real-world AI usage differs so much from public perception. Great post btw, thanks for sharing!

Seeing what teams actually build versus the hype is always interesting. Focus on practical applications makes AI much more valuable. Good to see solid data backing that up.

To view or add a comment, sign in

More articles by Tomasz Tunguz

  • Private Equity : The New Distribution Channel for AI Startups

    Private equity firms have emerged as the newest distribution channel for AI startups. While public companies have…

    5 Comments
  • The Bacon & the Skillet: When Does the AI Market Congeal?

    The AI market today is bacon in a hot skillet. Everything is sizzling, moving, & changing at an incredible pace.

    25 Comments
  • The Scaling Wall Was A Mirage

    Two revelations this week have shaken the narrative in AI : Nvidia’s earnings & this tweet about Gemini. The AI…

    25 Comments
  • Teaching Local Models to Call Tools Like Claude

    Ten months ago, DeepSeek collapsed AI training costs by 90% using distillation - transferring knowledge from larger…

    29 Comments
  • Running Out of AI

    By Monday lunch, I had burned through my Claude code credits. I’d been warned ; damn the budget, full prompting ahead.

    34 Comments
  • Datadog: As Reliable as Your Golden Retriever

    Datadog is becoming a platform company, & its Q3 2025 results underscore how successful this transition is. If nothing…

    20 Comments
  • Are We Being Railroaded by AI?

    Just how much are we spending on AI? Compared to other massive infrastructure projects, AI is the sixth largest in US…

    7 Comments
  • Are We Being Railroaded by AI?

    Just how much are we spending on AI? Compared to other massive infrastructure projects, AI is the sixth largest in US…

    23 Comments
  • A 1 in 15,787 Chance Blog Post

    I wrote a post titled Congratulations, Robot. You’ve Been Promoted! in which OpenAI declared that their AI coders were…

    31 Comments
  • OpenAI's $1 Trillion Infrastructure Spend

    OpenAI has committed to spending $1.15 trillion on hardware & cloud infrastructure between 2025 & 2035.

    28 Comments

Others also viewed

Explore content categories