Long before Altaf Ganihar started a software company, he was looking at stone. As an engineering student in Hubli, India, he joined the India Digital Heritage project, an effort to reconstruct Hampi in three dimensions. The medieval city is all pillars, courtyards, towers and weathered geometry. Capturing it meant translating a place with deep physical texture into points, surfaces and navigable models. It also meant spending time with architectural software that Ganihar found painfully slow.
The mismatch was hard to miss. His research involved computer vision, graphics and geometry processing - fields built around teaching machines to interpret and manipulate the visible world. Yet many of the tools used to describe a building still demanded repetitive drafting, modeling and handoffs. The technology could recognize complex form, but the workflow kept asking people to redraw, re-export and reconcile.
Ganihar began building an automation project in his spare time. It was crude, by his own account, but it did something useful: it turned flat plans into three-dimensional models. When he showed it to architects and builders, they did not treat it as a student exercise. They saw a working answer to a job they disliked doing. That reaction changed the direction of his life.
01 · The forkA safe path, left open
Ganihar had reasons to stay in research. By graduation he had worked on computer graphics, point-cloud processing and video summarization, co-authored international papers, and received invitations to pursue a PhD. The appetite for technical depth started earlier. In eighth grade, a pamphlet about Einstein's theory of relativity pulled him into self-directed mathematics and physics. By twelfth grade, he had published a paper on non-linear transformations.
But the architects who leaned toward his prototype offered a different kind of peer review. Their interest said the problem was not just intellectually interesting. It was present, expensive and shared. Ganihar chose to build a company after finishing his engineering degree in 2015. Professors supplied an early round of backing. That money let him hire interns and spend roughly a year and a half turning the experiment into a product.
“Premature product optimisation at any phase of product development leads to the doom of great products.”Altaf Ganihar, on his early product philosophy
The remark sounds like a warning to founders who polish before they understand. Ganihar's own sequence was closer to research: observe the failure, build a test, put it in front of practitioners, then follow the evidence. Early Snaptrude appeared as a SketchUp extension that could create 3D real-estate models from computer-generated or hand-drawn floor plans. It was a specific wedge into a much broader problem.
The wedge also gave him a language practitioners could test. A floor plan was familiar. A generated model was visible. An architect could decide quickly whether the output saved time or created cleanup. Snaptrude later joined the Brigade Real Estate Accelerator Program, placing the young product near developers and construction workflows rather than leaving it inside a research lab. The setting widened the questions. A visualization might help someone understand a space, but a working design system had to preserve measurements, materials, quantities and revisions. Each new requirement pulled the product away from a clever conversion and toward a dependable model of the building itself.
02 · The productThe hidden price of a revision
A building changes many times before it exists. Move a wall and the room area changes. Change the area and the program, cost assumptions, environmental performance and drawings may need to follow. The design decision itself can take seconds. The consequences can occupy a team for days.
That is the tax Ganihar keeps returning to: not simply the time required to draw, but the cost of changing your mind. When project knowledge is split across models, spreadsheets, presentations and messages, each revision becomes a small coordination project. A team explores fewer options because every option creates more reconciliation. The software quietly shapes the design by making some questions too expensive to ask.
Snaptrude's pitch expanded accordingly. The point was no longer only to automate a sketch into 3D. Ganihar began describing a collaborative building-design environment that joined the ease of freeform modeling with the structured information of BIM. In practical terms, the model would carry geometry and data together. Multiple people could work in the browser. A change could move through the shared system instead of becoming another file to circulate.
03 · The foundationSix years beneath the interface
By 2024, Ganihar was describing three foundations that had occupied much of the previous six years: a modern geometry kernel that runs in the browser, a data schema designed to address interoperability, and a database stack able to handle real-time updates. These are not naturally cinematic product features. They are infrastructure. Together, they determine whether a building can behave like shared, living information rather than a stack of documents.
This technical patience clarifies Ganihar's view of AI. He has argued that an intelligent design assistant needs access to the actual, current building model - its spaces, rules, relationships and revision history. An AI peering at an exported file sees only a slice. One operating on structured project data can propose a layout, check adjacency or revise a model while preserving context.
The distinction matters because architecture is constrained creativity. A promising arrangement can still fail a code check. A beautiful mass may violate a site rule. A plausible room list may create a bad circulation pattern. Ganihar's stated goal is not to remove the architect from those judgments. It is to move repetitive research, checking and coordination closer to the moment of design.
“Tools and platforms should empower, not hinder, the process of designing and constructing our built environment.”Altaf Ganihar · 2024
04 · The companyCapital for an unglamorous rebuild
Professional design software is a difficult market for a young company. The incumbents are embedded in contracts, training, libraries and client requirements. A better interface is not enough. New software must read and write the formats teams already use, preserve the information they care about, and survive projects where mistakes become steel and concrete.
Ganihar has spoken about software adoption as an empathy problem as much as an engineering one. A firm does not switch because a founder declares the old workflow obsolete. Its people have deadlines, habits and years of knowledge stored in existing systems. Migration must respect that investment. Collaboration must work both live, when a client is in the room, and asynchronously, when a production architect needs three quiet hours. The observation is ordinary and consequential: professional software succeeds inside the day people already have.
Snaptrude raised a $6.6 million seed round co-led by Accel and Foundamental in January 2023. That November it announced a $14 million Series A led by Foundamental, with Accel, Fortius Ventures and angel investors participating. The money gave Ganihar's team room to continue the less visible work: interoperability, performance, model structure and workflows broad enough for architecture firms rather than isolated demos.
His language became more expansive without losing the engineering sequence. First connect people, tools and data. Then reduce rework. Then make design decisions more informed, transparent and responsive. Sustainability enters the story through timing: performance and quantity information are most useful when a team can still change the building, not after the important choices have hardened.
05 · The through-lineGive attention back to the designer
There is a consistent personal logic from the student reading relativity to the researcher processing Hampi to the CEO demonstrating a building generated from a brief. Ganihar is drawn to complex systems, but his product instinct is subtractive. Remove the manual conversion. Remove the stale handoff. Remove the search for which file is current. Let the structure underneath absorb more of the complexity.
That is also why the Hampi story remains more than founder mythology. A heritage site is a reminder that buildings outlast the software used to draw them. The decisions made at a desk become rooms people occupy, routes they follow and materials that remain for decades. Speed alone is a thin goal. The more consequential possibility is that teams can consider more options, see more consequences and keep more people inside the decision before construction closes the window.
Ganihar's aspiration is to make good design more accessible while helping teams create efficient, sustainable buildings faster. Snaptrude still has to earn that ambition inside a profession rightly cautious about changing its tools. But the path so far has a useful discipline: start with work people resent, understand the data beneath it, and rebuild until a revision feels less like starting over.