Fifty days ago I published 100 Days That Changed Everything, and I ended it with a claim: the gap between "impossible" and "shipped" is now measured in days — for those willing to start with the prompt.
Claims like that age badly. So I spent the last fifty days trying to break mine.
I picked the hardest thing on campus
Every higher-ed technologist knows the system of record is the hardest thing on campus to build and the most dangerous to touch. It holds a student's entire life at an institution — enrollment, grades, financial aid, the bursar's ledger, the transcript. It's bound by FERPA and GLBA. It's multi-tenant, multi-role, and unforgiving: a wrong number in a GPA isn't a bug, it's a compliance incident. The last generation of enterprise systems took years and armies of consultants to stand up.
That's exactly why I picked it. I wasn't trying to enter that market. I was trying to falsify my own claim. If "start with the prompt" only holds on marketing sites and demos, it's a slogan. Put it against the hardest, most regulated class of application in higher education and you find out in weeks.
It held.
The numbers, and one honest caveat
Since January 1, across the fifteen repositories I primarily author, my own commits plus the autonomous agents I direct produced 518 commits and roughly 875,000 net lines of code. The test build alone accounts for 265 commits and about 383,000 of those lines, nearly all of it in June and July.
For context, my team used to average about 10,000 lines per developer per year. This is a different physics.
But I'll say the uncomfortable part before anyone says it to me. Lines of code measure volume, not craftsmanship. A large share of this was authored by AI agents working under my direction, so the raw number runs high. Read it as throughput.
The number I'm actually proud of isn't 383,000. It's that a governed, auditable, multi-tenant system with real financial aid and registrar logic exists at all — built the way I said it would be fifty days ago.
What actually changed was me
The first article was about velocity. This one is about skill. The most important thing that changed in fifty days wasn't how much code got written. It was how much better I got at directing the machine that writes it.
I think about it like a kid with a bin of Legos.
In February I was dumping the bin on the floor. My prompts were one line — "make the chat work," "fix this bug." My commit messages said "latest changes."
By April I was reading the box. I started naming what I wanted and who it was for.
By July I was designing my own sets. A real prompt from July 25th: fix one named subsystem as a single security workstream, work locally, do not push or deploy — these changes alter who can do what, so I review before anything ships. Investigate, fix, prove, in that order. Here are the exact files and line numbers, and the lines may have shifted slightly since I looked.
Five weeks earlier, June 18th, the same person wrote: go research standard business workflows for higher education and build the next generation platform.
One is a kid handing over the bin and saying make something cool. The other is an architect handing a bounded spec to a trusted crew.
Then I stopped prompting in the moment altogether and started writing the rulebook. In the flagship repository alone: 23 standing rules, 14 specialist critics that poke holes in my work before I ship it, 52 automated test suites that keep the agent from silently regressing, 38 recorded architecture decisions, and a 3,600-line master brief the agent reads on every task.
Early on I told the machine what to build. Now I tell it how to think, what to check, and how to prove it worked.
The part I didn't plan for
Here's what I found at the bottom of all that.
To build a system of record that fast, you have to stop hand-building screens. You start describing objects instead — what a thing is, who's allowed to touch it, what happens when it changes, and what gets written to the audit log when it does.
Build that engine well enough to survive a registrar and a financial aid office, and you haven't built a student information system.
You've built the thing that builds them.
Introducing Meridian
So that's what we're announcing today. LightLeap Meridian is a platform for building governed enterprise applications — the ones your institution needs and no vendor will ever build for you.
A business analyst describes a business object: what it is, the roles that touch it, the workflow it moves through, the rules that govern it. Meridian generates the application. Database-per-tenant isolation, least-privilege by default, an audit trail on every write — not features you bolt on later. The floor you start from.
Parking permits and appeals. Housing assignments and room changes. Citation management. Student conduct cases. Club and organization registration. Grant tracking. Facilities requests. Space reservations.
Not glamorous. That's the point. These are the systems that run today on an Access database somebody's assistant maintains, a shared inbox, and a spreadsheet nobody wants to own. Every campus has two hundred of them. None of them will ever get funded.
What this is not
Let me be direct. Meridian is not a replacement for Banner, Colleague, PeopleSoft, or Workday, and we are not asking anyone to rip one out. We've spent fifteen years building the integrations that keep those systems connected to everything else on campus, and we intend to spend the next fifteen doing exactly that.
Every institution I know runs a student information system plus a couple hundred things that are not one. That second category is the work Meridian is for.
Why this matters beyond us
N2N has spent fifteen years as the connective tissue of higher education — the integrations that keep more than 500 institutions running, the fraud detection that has stopped over 1.8 million fraudulent applications and protected more than $1 billion in student aid. That heritage is exactly why this leap counts. We didn't get here as a feature company. We got here as an institution-grade platform company.
Meridian is the proof that institution-grade rigor and startup-grade velocity can live in the same building. Governance isn't the brake on velocity. In a regulated domain, it's the only thing that makes velocity shippable.
Where it stands
Meridian is early. It is not generally available. We're working with our first institutions now, and I'd rather tell you that myself than have you read it somewhere else.
If you want to see it, let's talk.
Fifty days ago I said the gap between impossible and shipped is measured in days. I built the hardest system in higher education to try to prove myself wrong.
I couldn't.
— Kiran Kodithala, Founder & CEO, N2N Services
Fifteen years of building. The leap is now.
Talk to N2N about partnerships, press, or careers — or visit LightLeapAI to see what we built.