AI Manager Space

Overview

AI Manager Space brings scattered management activities into one modern workspace. That covers training videos, vehicle requests, purchasing and approvals, legal records, a combined master calendar and safe migration of historical data from legacy systems. Light AI assistance helps with vehicle assignment, purchasing decisions and calendar suggestions. It serves managers who handle these responsibilities today across disconnected tools, message threads, shared folders and old processes.

Challenges

Critical management workflows were trapped in disconnected legacy systems with no unified manager view. That meant training videos, vehicle requests, purchasing, legal records and calendars. Those legacy features were tied to old code and interfaces. Copying them directly was risky and they couldn't be maintained on a modern tenant-aware platform. Vehicle requests, purchasing and approvals had no structured workflow. Managers had no clear submit-and-approve path and no visibility into what vehicles or stock were actually available. Migrating years of historical data risked ambiguity and corruption without agreed export formats, controlled cutover timing and reliable update rules. It also carried real cross-tenant import risk if migration ran without strict scoping. Purchasing and receipt handling were informal. There was no vendor management, price comparison or pre-approval control for the items the business buys again and again. Stakeholders wanted AI to help with vehicle assignment, purchasing and calendar decisions. But unchecked AI risked overreaching into autonomous action rather than staying as recommendations. And without disciplined scope control a transfer project like this could balloon into rebuilding the whole platform.

Solution highlights

We transferred and modernised each area onto the new platform rather than copying legacy code across. Training videos gained proper hosting, upload, listing, management and playback. Vehicle requests were rebuilt with data models, APIs and an interface wired into task, operations and inventory workflows. Vehicle inventory is exposed to AI so it can suggest a suitable vehicle and driver for the manager to review. Receipt handling grew into a full purchasing module. It covers invoice and receipt upload and management. A manager pre-approval workflow lets a requester submit and their direct supervisor approve. There is vendor management and price shopping for commonly purchased inventory-linked items. An AI purchasing assistant recommends rather than decides. Legal records moved across with list, detail, upload, models, APIs and front-end support. That gives a structured place to store and review important documents. A master calendar brings events and volunteer days into a single view with an AI-assisted display and suggestion layer. Under all of it sits JSON-based data import. It runs through tenant-scoped management commands doing authoritative idempotent create-or-update migrations. An import can be re-run safely and it won't create duplicates. A dedicated final phase handled cross-area wiring, QA, cutover and documentation.

Key Features

• Training video hosting with upload, management, listing and playback
• Vehicle request models, APIs and interface wired into task, operations and inventory workflows
• AI-assisted vehicle and driver assignment suggestions for manager review
• Purchasing module with invoice and receipt upload and management
• Manager pre-approval workflow where a requester submits and their direct supervisor approves
• Vendor management and price shopping for commonly purchased items
• AI purchasing assistant that only recommends
• Legal records with list, detail, upload, API and front-end support
• Master calendar combining events and volunteer days in one view
• AI-assisted calendar display and suggestion layer
• JSON-based data import from legacy systems
• Tenant-scoped idempotent migration commands safe to re-run

Our Role & Approach

Our role was to modernise rather than replicate. We rebuilt each transferred feature as backend APIs plus a React front end instead of copying legacy code and interfaces. The result is maintainable rather than simply relocated. Discovery stayed deliberately lean with a one-page specification per transfer area. That kept the project moving without documentation overhead nobody needed. AI features were built MVP-first and held strictly to suggestions and display. Never autonomous action. Managers keep final control over vehicles, purchasing and scheduling. We used the JSON exports from the client as the authoritative migration source and separated development fixtures from production data. Each migration command is tenant-scoped and idempotent. Compatible workstreams ran in parallel to compress delivery. We reserved the final weeks for integration, migration, QA and cutover rather than shipping it all on the last day.

Technology Stack

Backend: Django with a plugin registry, multi-tenant model and tenant management commands. Frontend: React. AI: Butterfly AI layer. Integrated modules: training, vehicle repair, calendar and meetings, tasks, flower shop, farm automation, inventory. Data migration: JSON exports, tenant management commands, idempotency rules.

Outcomes

Managers now work from one modern tenant-aware workspace spanning training, vehicles, purchasing, legal records and calendars. No more juggling several disconnected legacy systems. The transferred features run on clean backend APIs and a React front end so they're far easier to maintain and extend without inheriting old code. Vehicle requests flow through structured workflows connected to tasks, operations and inventory with AI suggesting assignments to speed up decisions. Purchasing now supports invoices, receipts, vendor management, price comparison and a clear submit-then-approve workflow. Spending that used to be informal has controls around it. AI assistance stays safely scoped to recommendations across vehicles, purchasing and calendar so managers keep the final call. Legacy data migrates reliably from the exports the client supplies with idempotent updates that can be re-run without creating duplicates. Tenant-scoped commands protect against cross-tenant leakage during import. A reserved integration and release phase with QA and controlled cutover cut the risk of a disruptive all-at-once launch.