BUILT, NOT BRIEFED

AI agents built to
make delivery faster.

A project manager who ships product builds product. Helm — my live AI orchestration app — plus a set of AI agents and n8n workflows that take the repetitive parts of running projects off the PM's plate.

Executive analytics dashboard
Live Product · AI Orchestration

Helm

An AI command center that runs the project — not just tracks it.

A web app I designed and shipped end to end (live on Vercel). Helm brings AI into the heart of delivery: one workspace that orchestrates tasks, teams and status across a project, turning the coordination a PM normally does by hand into an assisted, repeatable flow.

How it works
  • Centralises scope, tasks and status in one AI-aware workspace.
  • Uses AI to coordinate the moving parts and surface what needs attention next.
  • Built and deployed full-stack — from concept to a running product.
Live product — designed, built and deployed end to end.
Visit the live app
AI Agent

Scope & Requirements Agent

Turns messy client inputs into structured user stories and acceptance criteria.

Takes raw client inputs — emails, call notes, briefs — and drafts the first-pass scoping document: user stories, acceptance criteria and effort buckets. What I do by hand on a discovery, done as a first draft in minutes.

How it works
  • Ingests raw inputs — emails, call notes and briefs.
  • Structures them into user stories, acceptance criteria and effort buckets.
  • Drafts the first-pass scoping doc for the PM to refine.
Hours saved per engagement on discovery and scoping.
Live GPT · ⌘/Ctrl + Enter
AI Agent

Meeting Intelligence Agent

Transcribes meetings, extracts action items, and pushes tasks to the tracker.

Pairs transcription (Otter.ai / Deepgram) with Claude for extraction and n8n to push tasks straight into the project tracker — owners assigned, nothing lost in the notes.

How it works
  • Transcribes the meeting automatically.
  • Extracts action items and assigns owners.
  • Pushes tasks directly into Jira, Asana or Linear.
Action items land in the tracker, not a forgotten doc.
Live GPT · ⌘/Ctrl + Enter
AI Agent

Client Communication Drafter

Drafts client-ready emails, reports and change-request replies in the right tone.

Given project context and recent updates, it translates internal dev-speak into clear, client-friendly updates — the first draft of every weekly report, status email and change-request response.

How it works
  • Reads project context and recent updates.
  • Drafts emails, weekly reports and CR responses in the right tone.
  • Leaves the PM to review and send — not write from scratch.
The tone-translation tax on client comms, removed.
Live GPT · ⌘/Ctrl + Enter
AI Agent

RFP / Proposal Assistant

Reads an RFP — even in another language — and drafts mapped proposal sections.

Ingests an RFP (including multilingual ones, like an Arabic Oman tender), extracts every requirement, maps them to a capability matrix, and drafts the proposal sections — a pain point I have lived through, automated.

How it works
  • Ingests the RFP and extracts each requirement.
  • Maps requirements to the capability matrix.
  • Drafts proposal sections ready for refinement.
Faster, more complete proposals — in any language.
Live GPT · ⌘/Ctrl + Enter
AI Agent

Retrospective Analyst

Collects retro feedback, clusters themes, and drafts real improvement items.

Gathers retro input asynchronously via form or Slack, clusters recurring themes across sprints, and drafts actionable improvements — so the insights live and get acted on instead of dying in a doc.

How it works
  • Collects feedback async via form or Slack.
  • Clusters themes and spots patterns across sprints.
  • Drafts actionable improvement items.
Retro insights that survive past the meeting.
Live GPT · ⌘/Ctrl + Enter

AUTOMATION LIBRARY

n8n workflows for delivery

The always-on side of the toolkit — scheduled workflows and background daemons that watch the project and act, so the PM doesn't have to. Each one replaces a task a person used to do by hand.

Standup & Status Agent

Scheduled
Jira / Linear + Slack + Git Claude summarise Detect blockers Standup / status report

Pulls updates from the tracker, Slack threads and Git commits, then generates the daily standup or weekly status report — with blocker detection and auto-escalation nudges built in.

Replaces Chasing "what happened yesterday" by hand.

Download n8n workflow (.json)

Risk & Dependency Tracker

Scheduled daemon
Tracker API Velocity analysis Predict bottlenecks Alert the PM

Less chatbot, more background daemon: it monitors task dependencies, flags slipping timelines, and predicts bottlenecks from velocity trends — alerting before things go red.

Replaces Finding out a timeline slipped only once it is red.

Download n8n workflow (.json)

Resource & Capacity Planner

Scheduled
Allocation data Detect overcommit Suggest rebalance Notify the PM

Tracks team allocation across concurrent engagements, flags overcommitment, and suggests rebalancing — built for running multiple APAC / EMEA / US projects at once.

Replaces Spreadsheet capacity juggling across projects.

Download n8n workflow (.json)

Real, importable workflows — download a .json and drop it into any n8n instance (Workflows → Import from File). Credentials and endpoint URLs are left as placeholders for you to wire to your own stack.

WHY I BUILD

If it's repetitive,
it shouldn't be manual.

Every one of these started as a bottleneck in a real project — a painful discovery, a status chase, an RFP in a language I don't speak, a report that ate a day. Instead of accepting the friction, I build the fix. That's what "AI-driven delivery" actually means: not a buzzword, but agents and workflows that give teams their time back.

See them in the case studies Talk about your bottleneck

LET'S TALK

Have a program that needs
to actually ship?

Whether it's a stalled project to turn around, a transformation to lead, or a delivery role to fill — I'm open to the conversation.

Start a conversation Email me