A presentation at AI Native DevCon London 2026 in in London, UK by Oleg Šelajev
You’re absolutely right, it was your You’re absolutely home directory! right, it was your home directory! Oleg Šelajev AI Native DevCon London 2026 01
More autonomy = more productivity 1 2 3 4 5 6 Autocomplete Assistant Specialist SemiAutonomous Autonomous Steward Stateless Prompt in, out Has memory Learns preferences Proactive within boundaries Represents you Judgement calls Drives agenda Coordinates 24/7 Sets goals Ethical bounds Most enterprises are at stage 1-2. Silicon Valley is exploring 3.
AI Native DevCon London 2026 02
Oh no
They only need to be right once. AI Native DevCon London 2026 04
AI Native DevCon London 2026 05
AI Native DevCon London 2026 06
Watch: I ask the agent to run an audit. It refuses. I ask it to script it. It refuses. AI Native DevCon London 2026 07
DEMO · auto-mode bypass (79s)
AI Native DevCon London 2026 09
AI Native DevCon London 2026 10
AI Native DevCon London 2026 11
AI Native DevCon London 2026 12
AI Native DevCon London 2026 13
AI Native DevCon London 2026 14
AI Native DevCon London 2026 15
Demo AI Native DevCon London 2026 16
AI Native DevCon London 2026 18
AI Native DevCon London 2026 19
sbx kits AI Native DevCon London 2026 20
Demo AI Native DevCon London 2026 16
Ecosystem AI Native DevCon London 2026 22
Blast radius AI Native DevCon London 2026 25
AI Native DevCon London 2026 26
28 AI Native DevCon London 2026
AI Native DevCon London 2026 29
brew install docker/tap/sbx AI Native DevCon London 2026 30
Run agents. Freely, safely. $ brew install docker/tap/sbx • https://docs.docker.com/ai/sandboxes/ • https://docs.docker.com/ai/sandboxes/customize/kits/ • https://github.com/shelajev/tessl-sbx-kit/ AI Native DevCon London 2026 31
Letting an AI agent loose for solving development tasks is a productivity dream, until it decides to optimise your home directory or brick your system by upgrading Python. We want that YOLO mode speed, but without the security nightmares. In this session, we’ll look at Docker Sandboxes: a new primitive designed to let agents operate in a restricted cocoon with limited access to the filesystem and controlled network and secret injections. We’ll dive into the typical mess AI agents create, see why basic isolation isn’t enough, and walk through a workflow for running agents that you can actually trust.
The following resources were mentioned during the presentation or are useful additional information.
for free. You
can too.