For founders and core teams installing an AI-native delivery system
Install a software factory in your company — not another AI tool.
A six-week program for a founder, CTO, and lead engineer. Build the context, execution, verification, and delivery system that lets a small team ship like a much larger one.
What you leave with
You leave with a working pipeline, a repo context pack, verification gates, a cost model, and a 90-day rollout plan. The goal is not 100% automation. The goal is a factory your team can trust.
6 weeks
Team program
3 seats
Per company
5 max
Companies
Own repo
Implementation labs
Admissions
Request the program sheet & proposal
Tell us where you are and what you are trying to move forward. Every note is read personally. No lists, no sequence, no pressure.
Prefer the full working syllabus? Use the private curriculum request in the curriculum section. It is not published or indexed.
Engineering standards grounded in 15+ years at
Who this is for
Built for people who own the outcome.
This is not a survey course. The work assumes you have real decisions to make and can bring a current system, team, or delivery problem to the labs.
A fit if you are…
- Founder/CEO who owns priorities, economics, and the operating model.
- CTO or head of engineering who owns architecture, repository shape, and technical risk.
- Lead engineer who owns implementation and will run the factory day to day.
- Companies with a real product, a real repo, and at least one delivery workflow to improve.
Not a fit if you want…
- Solo learners looking for prompt tips or a tool walkthrough.
- Teams without a working repository, CI, or a product to ship.
- Companies that want the instructor to implement the factory for them.
What changes
What you will be able to do afterwards.
Map what to automate, what to keep human, and what the factory should cost.
Write a context layer that lets agents work in your repo without constant re-explaining.
Run parallel agent work with worktrees, subagents, model tiering, and MCP.
Install verification gates: tests, security, E2E, and real-data QA.
Ship through self-hosted CI with blue/green, health checks, and rollback.
Leave with a 90-day rollout plan and a cost/throughput dashboard.
Curriculum
The work, week by week.
Each session produces a decision, a design, or a working slice. The capstone connects them into one architecture you can defend.
The private version includes weekly exercises, lab briefs, assessment rubrics, and the capstone checklist. It is not published or indexed.
Week 1Factory blueprint and economicsArtifact: Factory blueprint + economics model
Design the delivery system before installing tools. You will map the current workflow, choose automation boundaries, assign roles, and build the cost model.
Week 2The context layerArtifact: Repo context pack + AGENTS.md / CLAUDE.md
An agent can only work at the speed of the context it can find. You will build the repo layer that explains the system, the rules, and the definition of done.
Week 3The execution layerArtifact: Working agent execution workflow
Move from one chat at a time to controlled, parallel execution. You will install the agent workflow and the permissions that keep it safe.
Week 4The verification layerArtifact: Verification gate design + test plan
Automation is only safe when the pipeline can prove the work. You will design the gates that catch errors before a human has to.
Week 5The delivery and operations layerArtifact: CI/CD pipeline + operations dashboard
Ship the work and run it like a system. You will wire deployment, rollback, observability, and the metrics that tell you whether the factory is improving.
Week 6Growth layer and capstoneArtifact: Factory readiness review + 90-day plan
Connect the factory to the business, then defend the whole system. Each company presents its blueprint, pipeline, and rollout plan.
Production proof
The factory is already running
This program is based on the delivery system built for Neonbee: a 20+ module platform with roughly 60 deployable services, reusable GitHub Actions workflows, a self-hosted CI runner, blue/green deployments, a universal rollback workflow, and AI agents with approvals and audit. You are learning the operating model, not a vendor demo.
The case used in the labs
The delivery layer is modelled on neonbee-devops: build on a self-hosted runner, gate on tests and security, deploy by slot, back up the previous release, and roll back through one workflow.
~60
Services deployed
3
Seats per company
6 weeks
To a working pipeline
90 days
Rollout plan
Instructor: Praveen VR
Ex-Dell Principal Engineer · Nokia & Alcatel-Lucent R&D · Founder, Moonhive & Neonbee AI
What you leave with
Artifacts, not notes.
Every deliverable is built during the program and reviewed. You can take it into a real review, a planning cycle, or a hiring conversation.
1
Factory blueprint
The six layers, automation boundaries, roles, and the economic model.
2
Repo context pack
AGENTS.md / CLAUDE.md, codemaps, API registry, schemas, and task templates.
3
Working agent pipeline
One real ticket from spec to agent branch to verification to preview.
4
Verification gates
Tests, security checks, review rules, and the both-pass-or-stop CI gate.
5
Delivery and rollback
Self-hosted CI, blue/green, health checks, and a rollback path.
6
90-day rollout plan
A staged plan with owners, metrics, and the first three things to change.
Format & investment
How it runs.
Training-only, by design
No code access, no PR reviews, no repository audit, no NDA, and no done-for-you implementation. You learn the pattern and build the artifact; your team keeps ownership.
Per-company cohort pricing
Build Your Software Factory
One fee covers the three-seat team. Request the current cohort sheet for the program fee, dates, and the factory readiness review.
Live instruction and review directly with Praveen VR.
All templates, artifacts, and session materials.
Employer sponsorship brief and itemised invoice.
Questions
Clear answers before you apply.
Still unsure? Send a note through the form and Praveen will give you a straight answer.
Why three seats per company?
The factory only works if priorities, architecture, and implementation move together. The founder owns the economics, the CTO owns the architecture and risk, and the lead engineer owns the day-to-day execution. Sending one person usually produces notes instead of a system.
Do you need access to our repository?
No. Your team implements in its own environment. We review artefacts, diagrams, and pipeline designs — not production code. This is training-only; no repository access, credentials, customer data, or NDA required.
What if we are not using Claude?
The operating model is tool-agnostic. The labs are Claude-first because that is the most common stack, but the context, verification, and delivery layers apply to Cursor, Copilot, or self-hosted models.
How much time should the team budget?
Two hours of live session per week, plus four to six hours of implementation per company. The work happens in your repo, so the time is spent on changes you would need to make anyway.
What size company is this for?
It works best for product companies and agencies with a real repo, a small engineering team, and a delivery workflow that already exists. Pre-revenue teams can join if they have a working product and CI.
How is pricing structured?
Pricing is per company for the three-seat team, not per seat. Request the current cohort sheet for the program fee, dates, and what is included.
Next step
Build the factory your company keeps.
Applications are reviewed personally. If the fit is right, you will get a short alignment call and a clear next step. If not, we will tell you.