01 / Build. Grow.
AboutKaya.
Building software has never been easier. Getting customers still is. Kaya is the AI agent that does your marketing, so founders can keep building.
“Founders should build the future, not feed marketing dashboards.”
Faust & Uche · Co-founders of Kaya
Why Kaya exists
Building got easy. Distribution didn't.
So we built the loop that finds customers.
AI made shipping software fast and cheap. The bottleneck moved to distribution: positioning, channels, creative testing and growth. The next great tool doesn't help you build. It turns what you built into customers.
How it works today
Advice is everywhere. Execution, measurement and memory are nowhere.
- ✕A dozen tools to learn: ads managers, analytics, SEO, email, social schedulers, each with its own dashboard and its own numbers.
- ✕Advice, not action: courses, templates and chatbots tell you what to try. Nobody runs it, measures it and reports back.
- ✕Vanity metrics: impressions and clicks look good in a report and say nothing about signups, customers or MRR.
- ✕No memory: what was tested, why, and what it proved gets lost, so the same mistakes get paid for twice.
The Kaya loop
One system that observes the business, decides, executes, measures money and learns, then acts again.
- ✓One command center: your goal, your numbers and the next best actions, ranked, in one screen.
- ✓Execution within your rules: Kaya drafts, launches and measures, and waits for your approval whenever your guardrails say so.
- ✓Judged on revenue: every experiment is scored on signups, customers and MRR from your connected tools.
- ✓A memory that compounds: every hypothesis, action and result becomes a learning that shapes the next recommendation.
How it works
Understand, decide, experiment, measure, learn
Understand
Paste your URL
Kaya reads your site, pricing and positioning, finds your audience and competitors, and keeps every fact with its source and confidence.
Decide
Channel fit
It scores channels against your product and your economics, and says it plainly when a channel isn't worth your money yet.
Experiment
Measurable bets
Each idea becomes an experiment: hypothesis, audience, channel, budget, metric and a success threshold set before launch.
Measure
Down to revenue
Results are tied to signups, customers and MRR, and evaluated with statistics before anything is called a winner.
Learn
Compounding memory
Winners shape the strategy, losers are remembered, and the next recommendation changes because of the evidence.
Our principles
Four rules behind every decision
They come from the product itself: what Kaya measures, how it decides, and when it stays still.
Revenue over vanity
Kaya judges experiments on signups, customers and MRR, not on impressions. If it didn't grow the business, it didn't work.
“Did it actually grow the business?”
Evidence over guesswork
Every fact has a source and a confidence level. What you haven't confirmed never drives a decision, and every action can be explained.
“Why did the agent do this?” always has an answer.
Execution over advice
A plan only matters once it runs. Kaya drafts the assets, launches the experiment and brings the result back to you.
“Don't give me a marketing course. Give me customers.”
Guardrails over blind autonomy
Budget caps are enforced outside the AI. Risky actions wait for your approval, and sensitive ones always do.
“The agent should know when not to act.”
The people building Kaya
Kaya is built by two founders who ship software and wanted the marketing half of the job to work like the building half.
Faust Oswald
Co-founder
Full-stack developer who builds web and mobile products for clients, from Next.js to native iOS. He builds Kaya's product and agent, and shares the journey in public.
Uche Egboluche
Co-founder
Co-founder of Kaya, building the growth operator every software company should have.
“Every software company deserves an autonomous growth team. We are building it.”
Faust & Uche
Built to be trusted
Rules you can check
No inflated metrics here: these are the rules the product enforces today.
4
Autonomy modes
Observe, Suggest, Copilot or Autopilot. You choose how far the agent goes, and change it anytime.
5
Risk classes
From reading data (R0) to sensitive changes (R4). Each action is classified before it runs.
R4
Always needs you
Sensitive actions such as pricing changes are never run without your explicit approval.
0
Unlogged actions
Every tool call, approval and external change is recorded, so you can always see what happened and why.
1
Source per fact
Business facts carry their source and a confidence level, and stay proposals until you confirm them.
2
Languages
Kaya works in English and French, including the answers the agent writes for you.
How we build
We use Kaya before we sell it
A marketing product has to prove it can market itself. So we build it the way we ask founders to grow.
Our own products first
Kaya runs on the products we build ourselves, starting with StableVPS: paid search, comparison pages, creators and lifecycle email.
In public
We share what we test, what worked and what didn't, with the same honesty we expect from the agent.
Evidence before features
A feature ships when it helps a founder make a better growth decision, not because a competitor has one.
Guardrails first
Before the agent can do something new, we decide what it must never do without asking.
Grow with us
We are looking for our first design partners: founders with a live product who want to grow it with Kaya and shape what we build next.
Want to build Kaya with us, or become a design partner? Write to us directly.
Your next chapter
You build. What comes next grows with Kaya.
Start free. No credit card. Nothing spends without you.
From the first signal to the next milestone.