Founder's Note — August 24, 2026
There is a whiteboard outside my office with five columns drawn on it. Website. Helios. Polaris. Meridian. Orion. Underneath each one, in my handwriting, is everything that has to be true before it goes live. Across the top: 8/24/2026.
Six months to the day since Quantum Leap.
Two founders' notes ago, I claimed the gap between impossible and shipped is measured in days. Fifty days later, I tried to falsify that claim by building the most regulated system in higher education, but I couldn't. Both of those notes were about building.
Building turned out to be the easy part.
What six months actually taught us
The first hundred days answered a question about capacity: can a fifteen-year-old company move like a startup? Three million lines of code said yes.
The next fifty answered a question about rigor: does that velocity survive contact with FERPA, GLBA, a registrar, and a financial aid office? Meridian said yes.
The last thirty answered a question I didn't know I was asking, and it's the one on that whiteboard. Can we land it? Can we put five products in front of the five hundred institutions we already serve — some going live net-new this week, most on a migration path we've published — with support that answers, uptime we can promise in writing, and no customer who has run on Illuminate since 2016 left stranded?
That is a different company than the one that shipped the demos. Engineering capacity stopped being our constraint somewhere around month four. Everything downstream of engineering became the constraint instead.
Nobody writes a founder's note about a cutover plan. That's precisely why I'm writing one.
The pattern I didn't see until month four
Look at the five things going live this week, and there's a temptation to read them as five products. They aren't. They're the same lesson, learned five times.
Orion had a brilliant fraud engine wrapped in an interface built with a DevOps mindset rather than a product mindset. It took my team weeks to ship the smallest improvement. Six months ago, that was a UX complaint. Now I read it as an architecture complaint — the UI was slow to change because every customer was a deployment, not a tenant. Moving Orion to a SaaS chassis cut per-customer deployment time from weeks to minutes, and the interface finally became something we could fix in an afternoon. This week the last of our bespoke Orion instances move onto that chassis. The direction of travel is one-way — we are not leaving customers on the old model.
Polaris started as a bespoke build for one college. It became a routing platform — any incoming request was sent to the right agent. Then something happened that told me the shape was right: we stopped launching Polaris for customers and started running it under our other products. Starting this week, Polaris is the front door for Orion, Meridian, Helios, and the website itself. The bespoke build for one college is now the support layer for the whole company.
Helios exists because Illuminate's 2016 foundations couldn't support agents. That's the uncomfortable version. The libraries we chose in 2019 were never designed for what we wanted to build, and no amount of refactoring would change that. Legacy code is a toxic asset the moment agents can't safely act on it. This week isn't a Helios announcement — we announced it in June. This week is the harder thing: moving legacy customers, in-flight implementations, net-new institutions, and all three partner tiers onto it.
Meridian was a test I ran to prove myself wrong. Build the system of record fast enough, and you stop hand-building screens; you start describing objects — what a thing is, who may touch it, what happens when it changes, and what gets written to the audit log. Build that engine well enough, and you haven't built a student information system. You've built the thing that builds them.
Four products. One pattern. Each of them stopped being an application and became a chassis, and each of them became more valuable the moment we became its first serious customer.
That's the strategic finding from these six months, and I'd trade all the line counts for it: in an agentic company, the durable asset is the chassis, not the feature. Features are now cheap enough to be disposable. What is expensive — still, permanently — is tenancy, permissions, audit, governance, and the discipline to make those the floor rather than the retrofit.
The bottleneck moved, and we moved with it
If engineering is no longer the constraint, then the org chart built around the assumption that engineering is the constraint is wrong.
That is what the last sixty days of internal change have been about, and it's why launch week has PMO, DevOps, and a service desk on the same whiteboard as the product launches. Projects, milestones, and go-live activities have owners. Infrastructure, monitoring, and SLA have owners. Stress testing and the NOC run all week and don't stand down when the announcements go out.
We're also replacing our project management with the status tracker we built on our own platform. Onboarding, recruiting, and compliance run on Meridian. Our internal service requests route through Polaris. Not as a pilot. This is how the company operates starting today.
If we won't run our business on this stack, why would we ask a chancellor to run their business on it?
Why five in one week
The obvious play is to stage them. Launch one per month, with six weeks of air cover each, a clean narrative per product.
We're not doing that for two reasons.
The first point is that these five aren't independent. Polaris supports the other four. Helios carries the integrations that Orion and Meridian sit on. The website is where all of it becomes real to someone who isn't already a customer. Staging them means shipping four of them into a support model that doesn't exist yet, and the seam between products is exactly where enterprise software fails.
The second reason is the honest one. Staged launches give us somewhere to hide. One date, with everything visible at once, means the gaps surface in the same week and get fixed then — with the whole company watching, instead of one team quietly absorbing it.
Is this riskier? Yes. Deliberately. The last six months were spent proving we can build under pressure. This week is the first real test of whether we can operate under pressure.
The part I'd rather say myself
Some of this is not where I want it to be.
Orion's interface is better than it was, but it's still not what I'd call good — that work continues beyond this week, and I'm not going to pretend a launch date fixed it. Meridian is early and not generally available; we are working with our first institutions, and I'd rather you hear that from me. Some of our migrations will be messier than the plan indicates. Something will break on the 24th, and it will most likely break in a place none of the five columns on that whiteboard anticipated.
I'd also reiterate the caveat I made at 150 days, because it remains true. Since January, we've shipped 5,172,393 AI-assisted line edits, and just under a million of those came through agents I directed myself. Lines of code measure throughput, not craftsmanship.
But there is one number on that same dashboard that measures something else, and it's the one I'd show any CEO who asks what this job actually looks like now. Over those six months, I accepted 8,367 agent diffs. I used 24 tab completions.
Twenty-four.
I'm not sitting there typing code with autocomplete. I'm writing specifications, reading what comes back, and deciding what ships. That ratio — roughly three hundred and fifty reviewed decisions for every line I finished by hand — is the real work. It's also the clearest evidence I have that the skill that changed in six months wasn't typing speed. It was judgment, applied at scale. A large share of this was written by agents under my direction. The number I care about isn't how much we wrote. It's how much of it survives contact with a registrar in October.
What the next six months have to prove
Quantum Leap plus twelve months lands in February 2027, and the question then will not be velocity. That one's answered.
It will be adoption. How many institutions actually built their own agents on Helios rather than asking us to? How many campus systems were built on Meridian by a business analyst who has never written code? Whether the fraud numbers — 1.8 million applications stopped across our installed base, more than a billion dollars in student aid protected — grow because the product got easier to run, not because the problem got worse.
Volume proved capability. Only usage proves value. Everything we ship this week is instrumented to tell us the difference, and I intend to publish that honestly, whichever way it goes.
Today
Six months ago I stood in front of a sixty-foot screen with a prototype and a promise.
A demo is a promise. Today we keep it.
— 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.