Nehad Al Ghadri on stage, holding a microphone

Academies, agents, and tools I’ve built

Programs, AI agents, and tools I’ve built across L&D, leadership development, and AI-assisted learning. Most were built for clients and ran in production, where the clients tracked the results: time saved, completion rates, attrition, and productivity. A few are tools I built for my own work.

Academies 4

A three-stage blended sales academy I designed and ran across five markets for a Fortune 500 insurance client. New agents go from onboarding to certification, with a pass gate at each stage and a full compliance record. They reach productivity in 37% less time than under the old program.

37%Time-to-productivity reduced
95%Cert first-pass rate
25%Top-performer attrition cut

A yearlong leadership program I built for field managers and aspiring leaders at a Fortune 500 insurance client. It runs in two stages, recruitment and then leadership, and people move from the first to the second by passing a formal assessment center that tests what they can do on the job.

95%Participant satisfaction
30%Internal promotion rate up
70%Assessment-center pass rate

A library of soft-skills programs I built for a Fortune 500 insurance client across five markets, covering communication, negotiation, emotional intelligence, time management, and influence. Each program is a module, so the client could combine them into a different learning path for each audience.

9Micro modules built
60%Per-module dev time saved
3Academies powered

An internship program I designed and ran across several companies in the UAE (automotive, insurance, and professional services), recruiting students from universities in the UK, Canada, India, Lebanon, and across the Gulf. I built it from scratch so interns did real work, followed a development track, and left with experience worth putting on a CV.

6+Organizations across automotive, insurance & professional services
5+Countries represented per cohort
30+Interns placed across all program cycles

AI agents 5

A retrieval-augmented onboarding assistant I built for new joiners’ first 90 days. They ask in plain language, and it looks up the answer in company policy, information about the organization, and their earlier conversations, so they don’t need to know which document to open.

94%Policy lookup time cut
40%Knowledge retention up
<30sAverage time to answer

A Copilot Studio agent I built to help supervisors run the same well-structured safety observation conversation every time, without it sounding scripted. It replaces improvised, yes-or-no questioning with a guided conversation that always ends with an action the worker commits to.

87%Prep time reduced
5xObservation completeness
90%Supervisor adoption rate

A three-prompt pipeline I built that turns a job description into a role-based competency framework and exports it to Excel through Power Automate. L&D teams get a defensible competency model in minutes, where it used to take six weeks of expert workshops.

99.9%Build time cut (6wk → 6min)
85%Match vs expert frameworks
200+Roles mapped in production

A Copilot Studio agent I built to score uploaded training decks against a 21-criterion curriculum rubric and write the results straight to Excel. It cuts a manual review that took 1 to 2 days down to 10 to 15 minutes, and the team can see and edit the criteria for each engagement.

98%Review time cut (2d → 15min)
21Criteria assessed per deck
100%Audit-trail coverage

An agent I built that reads a finished eLearning course against the client’s own rulebook and source documents, and returns a terminology table and a list of gaps.

256Source documents read
55Term pairs checked
19Rulebook rules checked

Tools, software & programs 14

A four-stage Claude API pipeline I built to turn interview notes from subject-matter experts into a High-Level Design Document (HLDD), the plan an eLearning course is built from. It checks every output against 13 quality rules I took from a gap analysis of three strong HLDDs.

99%Draft time cut (7d → 30min)
13/13Quality rules pass rate
12+HLDDs in production

A small command-line tool I built that turns a SCORM package (the standard way eLearning courses are packaged for a learning platform) into a clean, paginated PDF of every screen. Reviewers and regulators get a static record of an interactive course without opening a course player. It handles pages built with JavaScript, click-to-reveal interactions, and screens that advance on a timer.

<60s200-page deck processed
100%Page coverage incl. JS
30+Courses extracted

A web app I built to replace the 30-minute write-up an instructional designer does at the start of every project. You give it a topic, a target role, an industry, and how many learning objectives you need, and it returns them in Mager format, tagged with Miller’s level and 70-20-10 placement, ready to export to Word or Excel.

95%Generation time cut
100%Mager format compliance
500+LO sets generated

A content system I built that turns campaign goals, audience notes, and live trend data into a LinkedIn content brief: 10+ posts per run, each matched to a format and a goal, exported as Word documents in under two minutes.

10+ postsper content brief
<2 minfull brief generated
Goal-drivenevery post mapped to the objective

A personal app I built that turns one topic a week into three narrated episodes of about 10 minutes, released Monday, Wednesday, and Friday. Each one comes with chapter timestamps, show notes, a podcast feed, and an automatic archive to my notes. A script on my Mac writes the episodes, and a Next.js web app plays them.

3Episodes per weekly topic
~10mPer episode, ±30s tolerance
6Pushes per week

A command-line tool I built that turns any real decision into a researched debate between several AI advisors. It writes a decision file, records the discussion as a short podcast (Kokoro text-to-speech), and checks back with me at 30, 90, and 180 days so I can see how the decision held up.

12Persona-modeled advisors
~7mDeep-mode decision cycle
3Post-mortems at 30/90/180d

A phone-first Next.js web app I built that turns my daily Claude Code tips and terms into audio I can scrub through and notes I can read, with two push notifications a day, chapter timestamps, lock-screen controls, and an automatic archive to my notes. It reuses about 80% of The Rabbit Hole’s code.

2Daily pushes, morning + backup
4Measured chapters per day
~3hPort time from sibling app

An engine I built that turns a subject-matter expert’s course outline into a finished learner guide deck on the client’s own template, word for word.

~2hOutline to finished guide
190/190Client lines kept word for word
10Guides live on the client site

An engine I built that studies two approved instructor-led decks, then builds the next course’s full deck and facilitator notes in the same style.

265Slides in one delivered deck
4Documents built for each course
2 daysCold start to a 20-slide deck

A web app I built where a client signs in and builds a High-Level Design Document section by section.

24Block types to build with
9Guided prompts built in

A proposal builder I made that drafts a consulting proposal from a brief, with a swappable pack for each industry sector.

12Industry sectors recognised
5Sector packs with their own cases
88Slide layouts in the library

A game engine I built for one-hour learning games, where each game is a pack of content dropped onto the same board.

5Modules in the first game
40Interactive beats to play
9 of 10To pass the final check

An analysis engine I built that audits a training program for instructional-design quality.

78Quality checks in the catalogue
130Findings on one course series
11Audits run on real programs

A tool I built that makes other AI skills to one house standard, so each new one starts tested and documented.

6Skills packaged to the standard
12Checks each skill must pass