For most of 2025 we had four SaaS in production at the same time, run by one engineer (with help). Carriva for retirement advisors. PrepareMesCours for French teachers. DraftMyLesson for English-speaking teachers. PrzygotujLekcje for Polish teachers. Plus the studio site at draftedby.com and a couple of in-development bets. People who hear that ask "how?" The honest answer is "imperfectly, and with a lot of things broken at any given moment." Here is the unvarnished account of solo founder shipping multiple SaaS, what made it possible, what we paid for, and what we would do differently.
Why four products instead of one
This is the question we get most often, and it deserves an answer before the tactical stuff.
If you are running one SaaS, your bet is concentrated. If it works, you win big. If the niche is wrong, you start over. We picked four because the four niches share enough that we can leverage shared infrastructure, but they live in independent markets so a bad week in one does not zero our income.
Concretely: the three lesson-planning tools are forks of the same engine, in different languages and curricula. Carriva is a different domain entirely (retirement advisory) but uses the same tech stack and ops playbook. The studio's hard skill is not "education software". It is "shipping vertical AI SaaS", and the four products are four bets on the same skill.
The risk we take is that four mediocre products beat zero great products only if you can keep all four above the bar. The minute one of them starts dragging you down, you have to either fix it or kill it. We have done both.
The architectural choice that made parallel possible
The single biggest lever was the lesson-planning monorepo. PrepareMesCours, DraftMyLesson, and PrzygotujLekcje share a Next.js codebase with locale-specific configuration and content. We finished Plan A of the unification on April 24, 2026, and the difference was night and day.
Before unification: a bug fix in PMC needed to be re-applied to DML and PL by hand, sometimes weeks apart, sometimes never. Around 30% of the active issues at any moment were "PMC has feature X but DML still has the old behavior". The forks were rotting.
After unification: one PR. Three deploys. Same fix everywhere. The cognitive load of maintaining "the same thing in three places" went away. We talked about the technical pattern in detail in our piece on multi-tenant SaaS architecture because tenant-style isolation maps cleanly onto locale-style isolation.
We are now doing the same with Creaclases (the Spanish multi-country fork covering 10 LATAM countries). Same engine, different country configs. If you are running multiple products and you are not constantly asking "could these share a backbone?", you are paying for parallel maintenance with no upside.
The tooling that turned one engineer into a small team
We are picky about tools, and most of the productivity story is one tool: Claude Code, the Anthropic CLI we use for the bulk of engineering work. We tried Cursor first and moved off because of context window limits and the pricing. Claude Code's persistent skills/memory and 1M-context Opus 4.7 are the unlocks for someone juggling four codebases. We wrote about why we use Claude Code for everything and what we tried first because it is the single biggest reason solo founder shipping multiple SaaS is doable in 2026.
The next layer is operational tooling. We run a Proxmox host with Coolify managing most app deploys and a Cloudflare Tunnel exposing services. AdGuard handles DNS for everything internal. The whole homelab is documented and standardized so that whichever product we are debugging at 9pm on a Tuesday, the operational shape is the same.
The third layer is content production. We use AI to draft blog posts (this one was drafted with heavy editing), email replies, French-to-English translations, and a thousand small textual things. The bottleneck for a one-person studio is rarely code. It is usually English (or French, or Spanish) prose. AI moved that bottleneck.
The leverage is not "AI does my work for me". The leverage is "AI eliminates the tasks I would have skipped, which would have killed me later".
What broke in parallel mode
Now the honest part.
Customer support is unevenly amazing and unevenly terrible
When two customers email at the same time about two different products, one of them waits. Our average response time across the studio is around 4 hours during the day, but the variance is brutal. Some emails sit for 18 hours. We have lost customers to slow response, especially on the English market where competitors reply in 30 minutes.
Our patch is templated responses for the top 8 inquiry types per product, plus a deliberate slot in the morning where we triage everything in one batch. It is not best-in-class. It is sustainable.
Marketing is the weakest leg
We do good engineering. We do mediocre marketing. With four products and one person, marketing is the discipline that gets cut first when something is on fire. Our SEO outputs are inconsistent, our social presence is sparse, and our cold outreach (Carriva's 534-prospect list got 20 contacts in mid-April, not 200) ran slower than it should have.
We are not pretending this is fine. It is the reason we are deliberately slower than venture-backed competitors and the reason we keep our cost base low. We are buying time with frugality.
Profnova is not in the four
You will notice the marketplace product, Profnova, is not on the parallel list. It is roughly 65% done, has 177 audit findings open, and the payment flow is broken. It is the cautionary tale of "we should have shipped a smaller version first". We started it before we had the discipline of going from idea to production in 6 weeks, which is now our default scope.
The right move on Profnova is either a hard cut (kill it) or a hard refocus (ship a smaller v1). Indecision is the worst option. Telling you this publicly is part of forcing ourselves to decide.
The operational rhythm that keeps it sane
Three habits keep four products from collapsing into chaos:
- Weekly product day. One day per product per month, blocked on the calendar. That product gets undivided attention for 8 hours. Customer development calls happen, deep features ship, the backlog gets pruned. The other days the product runs on autopilot.
- A unified deploy pipeline. Coolify builds, automated rollbacks, alerts via Sentry. We do not hand-roll deploys for any one product. The day a product needs a special deploy process, it stops being a fit for the studio.
- A single inbox. All customer email funnels into one address with labels per product. No per-product email accounts. The cost of context switching between four inboxes is real, and we eliminated it by funneling into one.
We track maybe six numbers across all four products in a spreadsheet that we look at every Sunday. MRR, active users this week, support emails this week, bugs reported this week, qualified outbound this week, and one product-specific KPI per product. Six numbers. Sunday. That is the dashboard.
What we would absolutely do differently
If we were starting fresh in 2026, we would change three things:
- Stop running four products and run two. Carriva and the lesson-planning monorepo (which counts as one product because they share a codebase). The fourth product (Profnova) was a stretch and is paying us back in audit findings. Three is the new four.
- Pick the niches with shorter sales cycles first. Carriva sales cycle is several weeks because CGPs are deliberate buyers. PrepareMesCours sales cycle is minutes (a teacher signs up, tries it, pays). The cash-flow shape is much friendlier on PMC and we should have leaned in there earlier.
- Hire help one quarter sooner. We have done freelance contracting for design and parts of engineering for short bursts. The general lesson is that if you are running four products and you keep saying "I'll do it next week", "next week" arrives 6 months late.
The case for staying small
Despite all of the above, we have no intention of ballooning the studio. Four products run by one engineer with a couple of contractors is a deliberate architecture. The economics are friendly: no funding, no investor pressure, no growth-at-all-costs. The cost: marketing is slower, support is patchier, and we will never look glamorous.
The trade we made is "consistency and sanity over hype and speed". Two years in, the trade still looks correct. We are profitable on three of the four products, the fourth is climbing, and we sleep most nights.
If you are weighing whether to be a solo founder shipping multiple SaaS, the question to answer is not "can I do it?" The question is "am I willing to be patient enough that doing four mediocrely beats doing one excellently?" For us yes. For others maybe no.
The honest financial picture
A note for founders considering the same path: we are profitable on a unit-economics basis on three of the four products. Total revenue across the studio is in the low five figures per month, growing month over month at a healthy but not viral pace. We do not raise. We do not pay ourselves a market salary. We reinvest in the products and the infrastructure.
Two-year revenue trajectory looks like a slow line going up and to the right, with one product (PrepareMesCours) carrying most of the weight, Carriva accelerating in 2026 as the outreach pipeline matures, and DraftMyLesson and PrzygotujLekcje as smaller but profitable additions.
Could one of these products by itself, with full attention and venture funding, be 5 times bigger? Possibly. We are not optimizing for that. We are optimizing for "the studio still exists and pays its bills in 2030". The math is different.
If you are weighing solo founder shipping multiple SaaS against the venture-funded single-product alternative, the question is not "which has a bigger upside?" The question is "which fits your tolerance for risk and your taste for ownership?" There is no universally correct answer. There is only the answer that fits the founder.
The skill that turned out to mattermost
If we had to name one skill that holds it all together, it would not be coding. It would not be customer development. It would be sequencing.
Sequencing is the discipline of knowing what to do this week, what to do next month, and what to defer to next quarter. With four products, the sequencing question is constant: which product gets attention right now, and why? We answer it with the weekly product day discipline, the Sunday dashboard review, and the explicit refusal to context-switch within a day.
Founders who fail at running multiple products usually fail at sequencing. They try to push all four forward simultaneously and end up moving none of them. The right move is one product, focused, this week. Next week the next one. Repeat.
Running four products is not parallel. It is fast serial. The sequencing is the skill.
What is next
In 2026 we expect to consolidate, not expand. Creaclases is going from pre-prod to GA inside the lesson-planning monorepo. Carriva is investing heavily in the free B2C detector and the CGP outreach pipeline. Profnova gets a decision (cut or refocus) by end of Q2. No new product launches before the existing four are uniformly above the bar.
Solo founder shipping multiple SaaS is a discipline, not a brag. The day we stop being honest about what is broken, the discipline rots. Posts like this one are the public version of that discipline. We will keep writing them.
