You bring one boring recurring task from your job. Six weeks later it runs itself, your IT department has approved it, and you have the numbers to prove it.
3 hrs→20 min
One Monday report, automated by week 4. Measured, not promised.
Learning AI has never been easier. Becoming the person who implements ithas never been harder.
Learning alone
Uses AI for emails and summaries only
Watches every posting add AI requirements
Has nothing to show a manager
Can't put company data into random tools
vs
Syntax Fellowship
One real process from your own job
Approved by IT
Measured result
A portfolio that proves it
Most people consume AI content. AI Operators ship approved automations.
You already understand something AI does not: how your company actually works. The opportunity is to become the person who redesigns that work with AI.
"Hands-on experience with AI-assisted development tools such as Cursor and Claude, including creating custom skills, agents, and rules … AI no-code tools and business process automation (e.g., n8n)."
Custom Claude skills, agents, rulesWeek 2 · Skills, plugins, company context
Business process automationWeeks 3–4 · always-on workflows on real systems
One posting from one function. The same requirement is appearing in marketing, finance, operations, HR, and sales roles.
10,000+
Playwright QA postings on Indeed US
76%
of QA leaders run AI test generation, up from 31%
12,397
AI PM postings since January
Baseline
GenAI in analyst job postings
+23%
BLS growth for analyst roles to 2032
Proof, not promises
What one approved automation changes
Priya · QA engineer, 6 yrs · Illustrative
Her Thursday smoke check runs itself
A workflow triggered on each release runs the checks, retries the flaky ones, and posts a summary to Slack with logs attached. She reviews the exceptions and signs off before lunch.
Before2.5 hrs every Thursday
After the program15 min reviewed & signed
−90%Less time on release checks2.5 hrs → 15 min, every Thursday
Marcus · Data analyst, 4 yrs · Illustrative
His Monday metrics are drafted before he logs in
A scheduled flow pulls the data overnight, refreshes the metrics, and drafts the stakeholder email with charts. A quality check catches a wrong figure before anyone reads it.
Before3 hrs every Monday
After the program20 min checked & sent
−89%Less time on Monday reporting3 hrs → 20 min, every Monday
Dana · Marketing manager, 5 yrs · not an alumna · Illustrative
Her monthly campaign report builds itself from four platforms
A scheduled flow pulls the numbers from four ad platforms, a Skill writes the commentary in the house voice, and a review step catches a wrong figure. She presented it to leadership in five slides.
Before2 days every month
After the program2 hrs reviewed & presented
−88%Less time on campaign reporting2 days → 2 hrs, every month
From the community
What graduates tell us
”
Amazed on just week 1... can't imagine what 5 more weeks will look like.
Gabriel Mendez
Marketing Operations Lead
”
It was great! Especially when you can see the result from the prompts provided.
Afa Faraji
Built her AI workforce
”
Love it so far! Practice made it very interesting!!!
Olga Lytvynenko
Program Manager
The program journey
Six modules, one process, one approved automation
Your own recurring process, carried through six stages: Find, Frame, Design, Build, Trust, Scale.
WEEK 1
Bring a processtime-and-pain audit
↓
WEEK 2
Write it downSOP + reusable Skill
↓
WEEK 3 - 4
Build the workflowalways-on, real tools
↓
WEEK 5
Earn the trustquality checks + approval pack
↓
WEEK 6
Approved & measuredhandover + before/after
01
Module 01
Find, Identify the process
A time-and-pain audit of your own working week. You map what repeats, what it costs, and what breaks, then pick the one recurring task worth automating and measure its baseline
What you'll learn
The time-and-pain audit: finding what repeats and what it costs
Impact × effort scoring, which tasks are worth automating
What makes a process automatable (and what makes IT say no)
Measuring a baseline you can compare against later
Learning outcomes
A shortlist of automatable tasks from your own job
One selected process with a measured baseline
A one-paragraph case your manager would recognise
Used in industries
OperationsHRRetailFinancial Services
Projects you'll build
01
A time-and-pain audit
Your recurring work mapped and costed.
02
The selected process
One task chosen, with a measured baseline.
Tools you'll use
ClaudeExcelNotion
Career skills gained
Opportunity spottingProcess mappingBaseline measurementSkill Building
02
Module 02
Frame, Write the job down
Turn the task into a spec a machine can follow: the SOP, the requirements, the ROI, and your first Claude Skill, with plugins and company context, doing a real slice of the work.
What you'll learn
Writing an SOP precise enough to hand to software
Requirements and ROI, the numbers that justify the build
Claude Skills: creation, instruction writing and testing
Plugins and company context, applied safely
Learning outcomes
A written SOP and requirements for your process
Your first Claude Skill, built and tested
An ROI estimate your manager can check
Used in industries
StartupsConsultingEducationOperations
Projects you'll build
01
The written SOP
Your process specified end to end.
02
Your first Claude Skill
A reusable, tested slice of the work.
Tools you'll use
ClaudeNotionGitHub
Career skills gained
SOP writingClaude SkillsPluginsROI framing
Fellowship experience
Assignments & Activities
Each module is reinforced through practical assignments, guided activities and structured feedback, so you apply what you learn as the programme progresses.
Week by week
Week 1
Assignment· 45 min
Run the time-and-pain audit on your own week
Assignment· 30 min
Score five tasks on impact × effort and pick one
Activity· 60 min
Automate the task in Claude Cowork, log before and after
Week 2
Assignment· 60 min
Write the SOP and build your first Claude Skill
Assignment· 30 min
Peer review, can a stranger follow your SOP?
Week 3
Assignment· 60 min
Design the always-on flow in n8n with checkpoints
Assignment· 30 min
Break it on purpose, walk every failure path
Activity· 30 min
Run it unattended five times and review the log
Week 4
Assignment· 75 min
Connect two workplace systems, zero copy-paste
Assignment· 45 min
Optional: a small internal tool built with Claude Code
Week 5
Assignment· 75 min
Run evals on 15+ real examples and set guardrails
Assignment· 45 min
Write the one-page IT approval pack
Week 6
Assignment· 60 min
Build a knowledge-base agent the low-code way
Assignment· 60 min
Handover doc + measured before/after
Capstone project
Bring everything together in your final capstone
Present your automated process end to end, the workflow, the evals, the IT approval pack, and the measured before/after, to your cohort and mentors.
A live working solution, running unattended
Handover docs someone else can operate from
Eval results and a five-slide impact presentation
Present your approach and outcomes
Leave with a portfolio-ready asset
Download the complete module-by-module fellowship breakdown.
What you'll have shipped
Six weeks, six shipped milestones.
Every week something real exists that did not exist the week before. Not demos, not tutorial projects. By week six it runs without you.
One recurring task, automated in Cowork
Your process packaged as a reusable Skill
Your flow runs on a trigger, unattended
Two systems connected
Approval pack, IT signed
Runs without you
How it works
Four steps from one process to proof
STEP 01
Bring one process
One boring, recurring task from your own job. That's the entry ticket.
STEP 02
Get selected
We review every process for fit, recurring, measurable, worth automating.
STEP 03
Learn and build
Six weeks of live sessions and mentor-reviewed builds on your own process.
STEP 04
Graduate with proof
An approved automation, a measured result, and the pack that proves it.