AIToday
AI Safety & AlignmentAI Business & IndustryTomasz Tunguz (Theory Ventures)Published: Oct 14, 2026, 04:00 JST

Theory 2026 AI Survey: 455 builders rank security top barrier

Theory 2026 AI Survey: 455 builders rank security top barrier

3 Key Points

  1. What happened

    Theory's 2026 AI Survey polled 455 senior technical builders and found security & governance jumped from sixth to first among AI pain points in twelve months, cited by 43%, with evals second at 22%.

  2. Why it matters

    Security has moved from a peripheral concern to the leading blocker these builders report when adopting AI.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Theory's 2026 AI Survey, which drew replies from 455 senior technical builders, marks a shift from the previous edition. In 2025, the survey found that context limited AI performance; in 2026, that appears solved. The pain-point ranking moved sharply in the same period, with security & governance climbing from sixth to first in twelve months, ahead of evals, which came second at less than half that share.

On the tooling side, the survey points to concentration in one layer: of the 259 teams writing code with agents, 62% host their sandboxes on Cloudflare, three times the closest competitor. Cost optimisation and model selection now look standard, with 45% of teams routing queries across models.

Analysis of agent behaviour also appears to be converging. 56% of respondents use business intelligence tools, 48% rely on coding agents, and 41% use dedicated observability tools for the same job, while spreadsheets persist at 30%. Theory notes the full survey covers open versus closed source adoption, the rise of MCP, and still-early adoption of evals.

FAQ
How many teams route queries across multiple AI models?
45% of teams route queries across models, with 23.7% using a router they trained and 21.5% using a proprietary one. 54.7% still send every query to a single model.
Where do teams writing code with agents host their sandboxes?
Of the 259 teams writing code with agents, 62% host sandboxes on Cloudflare, three times the closest competitor. Other and Vercel follow at 17% each.
How do teams analyze agent performance?
56% use business intelligence tools, 48% rely on coding agents, and 41% use dedicated observability tools. Spreadsheets are used by 30%.
Tomasz Tunguz (Theory Ventures)Read Original Article

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleScan Computers, OLO Robotics partner to ease robot deployment