Skip to main content
Caspian Almerud
TECHNICALOngoing

Obsidian KB / EMAX Operating System

Caspian Almerud portrait
Caspian Almerud
Systems Architect & Automation Developer · Stockholm, Sweden
Published

A client-operations knowledge base that turns scattered SEO work, meeting notes, CRM records, tasks, and search data into a repeatable agent-assisted workflow.

Project outcomes

Problem

Agency client work was spread across meetings, CRM notes, tasks, search exports, transcripts, and ad hoc decisions. The main risk was losing context between sessions and repeating work that had already been solved.

Outcome

Built a traceable client-ops layer where AI-assisted workflows can load client profiles, recent activity, strategy, performance, patterns, and open tasks before making recommendations.

Metrics

Client ops scope
Active EMAX portfolio
Workflow types
Reviews, GSC pulls, CRM notes
Knowledge assets
Profiles, patterns, solutions
Role

Systems Architect & Automation Developer

Category

Agentic Operations

Skills
Systems DesignWorkflow AutomationKnowledge ManagementSEO OperationsAgentic Workflows
Tools
ObsidianTodoistGoogle Search ConsoleFirefliesCRM APIPythonShell Scripts

The challenge

Client work at an agency moves through many surfaces: meetings, CRM notes, Google Search Console exports, Todoist tasks, Fireflies transcripts, performance reports, and specialist decisions. The risk is not just losing information. The bigger risk is losing context: why a recommendation was made, what was already tried, which client constraints matter, and what should happen next.

I needed a system that could keep client context alive across weeks of work, support recurring review workflows, and make AI-assisted execution more reliable without turning the process into a brittle black box.

What I built

I built an Obsidian-based operating system for EMAX client work. Each client has a structured folder with profile, strategy, activity log, performance notes, meeting notes, deliverables, and active ideas. A central mapping layer resolves client names to the right folders, project IDs, CRM IDs, and aliases so workflows do not create duplicate or misplaced context.

On top of the knowledge base, I created reusable skills and scripts for recurring workflows:

  • Weekly and biweekly client reviews
  • Google Search Console pulls and keyword/page analysis
  • Reach Top 3 ranking diagnosis
  • CRM note drafting and posting workflows
  • Todoist task creation and follow-up reminders
  • Fireflies meeting processing
  • Monthly performance summaries
  • Pattern and solution capture for repeatable SEO tactics

The system is intentionally traceable. Activity logs show what was done. Strategy files preserve current priorities. Performance files keep monthly results visible. Pattern files capture approaches that worked across clients. Solution files document technical fixes so they can be reused instead of rediscovered.

Result

The result is a practical client-ops layer where AI assistance has memory, guardrails, and source context. Instead of starting from scratch for each client request, the workflow can load the relevant profile, recent activity, strategy, prior patterns, and current tasks.

The system now supports work across the active EMAX client portfolio and gives recurring workflows a clear paper trail: what happened, why it happened, and what should happen next.

Proof points

  • Client profile, strategy, activity-log, and performance file structure
  • Reach Top 3 workflows that turn SERP and GSC data into implementation briefs
  • Weekly reviews that turn completed work into client-facing CRM notes
  • Pattern library examples such as PageSpeed optimization, service-area expansion, and disavow audits