Name: Projects management that the AI can drive.
Brand: Odexy

The wedge

You can't run a real engineering or services team on a kanban board. You need cycles you can plan against, custom fields the AI can fill, time tracking the finance team trusts, and capacity planning that updates when someone takes leave. Odexy Projects ships all of that — and the same actions that drive the UI drive the AI agent.

What ships today

Cycles — sprint-style time-boxed plans with status rollups and burn-down. Tasks belong to a cycle; the cycle inherits the project's settings.
Saved views — filter / sort / group combinations persisted per user. Switch from "my open this week" to "team backlog by priority" in one click.
Custom fields — typed (number / select / date / url / link) and added per project. Forms auto-pick them up; the AI fills them with provenance.
Time tracking — start / stop / manual entry, weekly digests, auto-approval rules, edit history. Coverage-symmetric rollups so leave overlays correctly.
Capacity planning — per-employee availability after leave + holidays. The schedule view shows red where a sprint is over-committed.
Task detail sheet — comments threaded by entity, sub-tasks, dependencies, attachments, secrets storage (AES-256-GCM at rest).
Realtime presence — Slack-grade unread, who's viewing the same task, typing indicators.

How it integrates

Projects is part of the suite, not a silo:

Chat — every task has a thread; mentions notify and write to the audit log.
HRM — task assignees draw from the live employee directory; leave overlays the schedule view.
Recruitment — onboarding tasks fire automatically when a candidate signs the joining pack.
Sales — quotation acceptance can spawn a project from a template.
Accounting — billable time exports straight into invoice line items.
Email — every notification routes through the unified Email module, with one suppression list.

How the AI helps

The Projects AI is the same agent that runs the rest of Odexy — the same permissions, the same activity trail. Concretely:

Plan a sprint from a brief: "Cut a 2-week cycle for the auth migration, include security review and rollback rehearsal."
Triage backlog: "Group anything tagged blocker by owner and propose deadlines based on capacity."
Daily standup: "Summarize what shipped yesterday and what's at risk in the active cycle."
Capacity-aware reassignment: "Move Maria's open tasks since she's on leave Thursday + Friday."

Every action goes through projects.task.update, projects.cycle.create, etc. The same handlers your team's clicks invoke.

Built for scale

Activity trail — every change is recorded: who made it, whether they were a person or the AI, and what it looked like before and after.
Permissions — scope-suffixed: projects:task:update:own, :team, :any. Custom roles can grant a precise slice.
Performance — server-side pagination on every list view; cursor-based, not offset. 100k task tables stay responsive.
Realtime — Postgres LISTEN/NOTIFY fan-out via WebSocket. No long-polling.
Data export — spreadsheets, structured files and the full activity history, any time. Nothing held hostage.

Odexy Projects vs. the category

|---|---|---|---|

| One schema with CRM / HR / Accounting | ✓ | — | — |

| Custom fields the AI can fill | ✓ | ✓ | ✓ |

| Capacity planning with HR leave | ✓ | — | partial |

| AI uses the real UI (not a chatbot) | ✓ | — | — |

Linear is excellent at issue tracking but you'll still buy a CRM, an HR system, a payroll system. Jira covers more but charges per seat per module. Odexy bundles all 14.

Design-partner take

> "We replaced ClickUp, Toggl, and a spreadsheet capacity tracker with Projects + HRM. The cycle view is what we'd have built if we had the time." — Design partner, 28-person services team

Projects management that the AI can drive.