Agent Design · Forthcoming
What a production role needs beyond a system prompt
A practical breakdown of triggers, tools, memory, approvals, evaluation, and ownership.
Article-ready topicJournal
Notes from designing and operating AI agents.
Agent architecture, guardrails, integrations, professional-services workflows, and the details that determine whether a system survives production.
Production note
From Slack request to production change: what Brajko actually proves.
The interesting part is not code generation. It is maintaining project context, respecting a design system, editing a real codebase, shipping through a deployment path, and leaving enough evidence for the team to trust the result.
Open Brajko’s profileB
Guardrails · Forthcoming
Autonomous, approval-required, forbidden
How to build an autonomy matrix that operations, security, and business owners can all read.
Article-ready topicProfessional Services · Forthcoming
Why intake is a strong first agent deployment
Intake combines measurable state, high repetition, human review, and clear downstream value.
Article-ready topicIntegrations · Forthcoming
API coverage is part of product design
Why an attractive agent demo can fail when the system of record cannot support safe writes.
Article-ready topicProduction Notes · Production note
What Brajko taught us about handoff
The website-building agent is useful because logs, context, deployment access, and rollback are part of the workflow.
Article-ready topicAI Readiness Audit
Have a workflow worth turning into a production note?
In one working session, we map the workflow, systems, approval boundaries, risks, and the most useful first deployment.
Workflow mapTop three opportunitiesRisk and approval mapIntegration inventory
Book the 60-minute audit