Mobolaji Ogunbiyi

KS App

A multi-tenant platform for CRM, projects, finance, communications, campaigns, knowledge, reporting, automation, and client operations—with an AI business operator inside the same workspace.

Designed and built by Mobolaji Ogunbiyi · May 2026—Present

Status: active private developmentRole: founder, product design, full-stack architecture and implementation
DemoCase StudyEvidencePrivate repository

Synthetic-company workspace

This read-only fixture uses a fictional wellness company to demonstrate implemented product boundaries without exposing client data, credentials, or a private backend.

Northstar WellnessSynthetic tenant · fixture data · read only

Operations overview

The workspace combines records that normally live across separate CRM, project, finance, and communication tools.

active_clients: 12
projects_on_track: 7
invoices_due: 2
campaigns_running: 3

CRM and client operations

Tenant-scoped contacts, organizations, opportunities, communication history, and next actions are connected to the work they generate.

account: Harbor Pediatrics
stage: proposal_sent
owner: Jordan Lee
next_action: follow up Friday

Projects and delivery

Project state, milestones, tasks, files, and client-facing work remain linked to the same tenant and account context.

project: New patient intake launch
health: on_track
milestone: staff review
blocked_items: 0

AI business operator

A natural-language request selects tenant-aware skills and tools, prepares a bounded set of record updates and artifacts, and records the action history.

request: Prepare Friday's client update
read: project status + recent messages
prepare: progress summary + next actions
result: draft saved; no external message sent
Read the complete fixture
  1. Review the combined operations overview.
  2. Inspect a tenant-scoped CRM account.
  3. Connect the account to project delivery state.
  4. Prepare a client update through the tenant-aware assistant without sending it.

Contextualized evidence

The current benchmark uses synthetic business data and verifies persisted outcomes. It is regression evidence—not proof of every live integration or production workload.

Business fixtures
3 companies
Operational coverage
15 checkpoints
Simulation horizon
1 month

Checks span saved records, generated artifacts, action histories, safety behavior, and regression assertions across simulated operations.

What I engineered

Full-stack operations product

Next.js and TypeScript product surfaces for CRM, work, finance, communications, campaigns, knowledge, reporting, and automation.

Tenant-aware backend

PostgreSQL and Supabase boundaries for users, workspaces, records, permissions, and operational state across multiple businesses.

AI inside the product

A business operator that selects tools and skills to update records, prepare documents and communications, analyze operations, and produce reports.

Architecture

AI actions use the same tenant and domain boundaries as ordinary product actions; the agent is not a separate ungoverned system.

Engineering decisions and tradeoffs

Tenant context is mandatory

Every record and assistant action resolves through a workspace boundary. That adds implementation friction but prevents convenience code from becoming cross-tenant risk.

AI augments domain workflows

The operator uses existing record operations instead of inventing a parallel data model. This keeps actions inspectable, but tool design must stay aligned with product changes.

Fixtures verify outcomes

The benchmark inspects records, artifacts, and histories rather than grading prose alone. Synthetic coverage is repeatable, though it does not replace connected production evidence.

Current status

KS App is a substantial active private platform spanning browser product surfaces, application services, tenant-aware persistence, integrations, operations, and an AI business operator. The portfolio demo is intentionally fixture-driven and exposes no client workspace or production backend.