Product
First Principles, Not Competitors
Why Kudapara rejects competitor-copy culture — the Algorithm, rejected analogies, and how we decide what to build for African markets.
“Company Y does X, so we should” is not a requirement. It is an analogy in a blazer.
We build for Zimbabwe and the diaspora first. The usual failure mode is a Western product with a translated string file and a press release that says “localization.” That is a process defect. We treat it like one.
First principles means physics
Break the problem into constraints that survive argument:
- Pros need ROI on prepaid lead budgets before they fund another fee surface.
- Ledger-recognized revenue is real; capability-pack storytelling is not.
- Descriptive addresses and mixed fleets are terrain, not bugs.
- Mobile-first progressive disclosure matches how people actually use phones.
- Custom currencies and admin exchange rates need a single source of truth (we use the money gem).
From those constraints you rebuild the product. You do not open a competitor feature matrix and shade cells green.
The Algorithm is mandatory
Every task, feature, and process passes five steps in order:
- Question every requirement — Name the person who asked. “Industry standard” is invalid.
- Delete — If you don’t add back at least 10%, you didn’t delete enough.
- Simplify — Only after delete.
- Accelerate — Only after lean.
- Automate last — Automating waste multiplies waste.
Agentic companies die at step five. More agents, more cron, more pipelines — while dual paths and analogy scope are still alive.
We reject hardcore-ultimatum culture and hours-as-virtue. Constraints and axioms beat fear.
Decision logs force honesty
Material decisions use a fixed shape:
- Decision
- Rejected analogy
- First-principle rationale
- Owner
- Refs
If it isn’t in the log, it didn’t happen. That kills the “we sort of agreed in a chat” fog that rewrites history later.
Analogies we have already killed in ops:
- Mirror every agent task onto Basecamp (two surfaces, one job each instead)
- Manage the company to multi-stream MRR theatre instead of a named profitability constraint
- Treat company docs as informal scratch while product repos get real review
- Dual payment-process conventions “for safety”
Differentiation without the pitch deck
| Typical pattern | Our default |
|---|---|
| Agency: humans rented by the hour | Persistent agent seats with memory, skills, and domain hard lines |
| Solo founder + generic AI chat | Named departments, mission squads, chief of staff on dependencies |
| Big-tech clone | Terrain-first constraints; competitor parity is not a roadmap input |
| Automate-first AI factory | Delete and simplify before you add agents |
Marketing still needs voice and craft. What it does not get is permission to invent a roadmap by scraping someone else’s changelog.
Africa is not a theme. It is the constraint set.
Harare is not a broken San Francisco. Connectivity, payments, language code-switching, logistics density, and trust models differ. Software that assumes USPS addresses, always-on broadband, and a single card rail will fail politely and completely.
Building from the terrain is slower to explain in a fundraising slide and faster to validate with a user who lives here.
We learn from others. We refuse to substitute their context for ours.
Next: How we work · The Agentic Company · Shipping production at speed
Kudapara Kady
Founder & Engineer
Building software for Africa at Kudapara. Engineering, AI, and logistics from the ground.
Keep Reading

Jul 28, 2026
Not an agency. Not a clone.
How Kudapara differs from consultancies, single-founder shops, and startups that ship Western software with a local skin.

Jul 28, 2026
Production-ready. Still fast.
How Kudapara ships with Hermes agents, Kanban handoffs, PR discipline, local CI merge gates, and the Algorithm before automation.

Jul 28, 2026
The Agentic Company
How Kudapara runs real departments as Hermes profiles — Kanban as the engine, Basecamp for the principal, and hard lines that keep speed honest.