Operator-led product studio

Build the systems the future depends on.

Fairy Hill Ventures builds products for security, infrastructure, and applied AI. We focus on systems where reliability, visibility, and control matter.

Configuration → observability → control

01 / Studio thesis

Critical layers, built deliberately.

Software is becoming more autonomous. The infrastructure around it must become easier to inspect, govern, and operate. We build at that boundary.

01

Secure by design

Put identity, policy, data protection, and auditability into the system rather than around it.

02

Infrastructure that holds

Make complex systems easier to configure, observe, recover, and run across environments.

03

Applied AI with a job to do

Use models where they improve a defined workflow, with clear controls and inspectable behavior.

02 / Operating approach

Work from the constraint outward.

We start with the part of the system that cannot fail, then work toward a product that can be tested in real operating conditions.

Map

Define the system

Identify the user, the critical workflow, the failure modes, and the boundary the product must enforce.

Build

Make the smallest complete loop

Connect configuration, execution, and feedback before expanding the surface area.

Harden

Add operational evidence

Instrument behavior, test recovery paths, and make policy visible where decisions are made.

Release

Put the work where it can be inspected

Publish clear documentation, version the changes, and keep the path from issue to release legible.

03 / Open source

Proof in the repository.

These projects by Sean Fraser explore the same operating problem from three directions: configuration, observability, and control.

Configuration

Forge

Agent infrastructure as code

Define agent configuration in version-controlled files, validate changes, inspect diffs, deploy across environments, and roll back when needed.

Public repository MIT OSS release v0.2.7 last verified 28 Sep 2026

Observability

Lantern

Agent observability and tracing

Capture instrumented model calls, tool invocations, and custom spans; attribute token costs; and score behavior so teams can inspect system activity over time.

Public repository MIT OSS release v0.3.6 last verified 28 Sep 2026

Control

Bastion

Security gateway for AI agent traffic

Enforce policies, redact PII, apply rate limits and injection checks, route across providers, and record audit logs through a single proxy.

Public repository MIT OSS release v0.2.7 last verified 28 Sep 2026

04 / Operator-led

The thesis stays close to the work.

Fairy Hill Ventures operates at the product layer. Public repositories by Sean Fraser provide a practical place to inspect how configuration, observability, and security fit together in working systems. The result is a studio practice grounded in implementation, not a list of abstract themes.

The projects are documented in their repositories. Commercial availability and support are not implied. Repository ownership by Fairy Hill Ventures is not asserted.

Start with the hard part

What has to be true for the product to work?

If you are building around security, infrastructure, or applied AI, send a concise note about the system, the constraint, and what you are testing.