Why the most expensive construction mistakes happen before ground is broken, and how we designed Jengafy to move decisions earlier
There's a familiar story on building sites across East Africa. Someone buys a plot. They find a design they like, often from a picture. A contractor starts. Halfway through, the money runs out, or the design turns out not to fit the plot, or the real cost of materials arrives as a surprise. The walls stand unfinished for months or years.
The mistake in that story didn't happen on site. It happened before construction started, when the key questions (what can I realistically build here, with this budget, and what will it actually cost?) were never answered with real numbers.
That's the problem we set out to solve with Jengafy, an AI-powered construction platform I founded within Tamalaki Business Network.
The decision problem, not the drawing problem
When people think of "AI in architecture" they picture image generators producing beautiful renders. Renders are fun, but a render doesn't tell you whether you can afford the building.
So we framed Jengafy around a different question: how do you make a confident building decision before you commit money? The inputs are things almost every person has before they talk to a professional:
- The plot: location, size, orientation.
- The budget: what you can actually spend.
- The vision: what you want to build and why.
The outputs are what you need to decide: design concepts validated against the plot and the brief, an itemised cost estimate in local currency, a Bill of Quantities (BOQ), a 3D view you can walk through, and a project plan you can hand to a team.
Seven steps from idea to a buildable plan
We structured the product as a guided journey we call Dream-to-Build:
- Decide what to build for your plot, budget and goals.
- Generate architectural concepts.
- Validate them against the site, regulations and your brief.
- Refine for cost, space and buildability.
- Estimate construction costs and generate the BOQ.
- Visualise the decision in 3D.
- Hand over a build-ready plan.
Each step exists to make one decision easier. The ordering matters: cost comes before beauty, or at least alongside it. A design that fails step 4 should change before anyone gets attached to the render.
Why the BOQ is the unsung hero
If you've never seen a Bill of Quantities, it's essentially the shopping list of a building: every material and work item, quantified and priced. Contractors bid against it. Banks and investors read it. Homeowners rarely see one before construction, and that's exactly the problem.
A BOQ turns a vague number ("about this much per square metre") into a list you can question: why this much steel? Why this roofing? What if we change the finish? Generating a first-pass BOQ early doesn't replace a quantity surveyor. It makes the conversation with one far more productive, because you arrive with a structured starting point.
Designing for where people actually build
Generic tools trained on someone else's context can produce designs that don't fit. Jengafy is built to understand tropical climates, local materials such as rammed earth and makuti, Swahili-coast aesthetics, and the economic realities of building in emerging markets. Costs are expressed in the user's local currency, because a cost estimate in the wrong currency or with the wrong material prices isn't an estimate.
We also built different paths for different users: homeowners, property developers, architects and contractors, plus churches, schools, hospitality and commercial projects, because a church committee planning a community build within a tight budget has different questions from a developer modelling returns.
What AI should and shouldn't do here
I'm careful about overclaiming, so here's where I draw the line:
- AI should compress the early, repetitive work (concept generation, first-pass quantities, cost roll-ups) so people can explore more options before committing.
- AI should make trade-offs visible: change the floor area or the finish, and see the cost move.
- AI shouldn't be the final authority on structural safety, regulatory approval or a binding price. Those still need qualified professionals and local approvals. The platform's job is to get you to that conversation better prepared.
We summarise readiness as a project readiness score: a single indicator of whether you're ready to build, or still have open questions.
Pricing that matches how people build
One small product decision I'm glad we made: no subscriptions, just credits. Most homeowners build once or twice in a lifetime. Asking them to subscribe monthly to plan a single house never made sense.
The bigger point
For many families and businesses, a building is the biggest investment they'll ever make, and too often that money is committed before anyone has answered basic questions with real numbers. If AI can do one useful thing in construction today, it's this: move the decision earlier, when changing your mind is free.
You can try the estimating and design tools at jengafy.com.
About the author
Isaac Mwendwa Muthui is the Founder & CTO of Tamalaki Business Network, a Nairobi-based innovation holding company building AI-native platforms across agriculture, manufacturing, construction and exports. He founded Jengafy, AgriTrust Exchange, FactoraERP and CEA Consulting, and is co-founder and CEO of Purevado. isaacmuthui.com · tamalakibn.com
Top comments (0)