MeltPlan Wants to Fix Construction Before Construction Starts

Kanav Hasija did not set out to build in construction. He got there the way many people do: through a renovation that went sideways. After buying and remodeling a 1938 home in Berkeley in 2021, he watched the project blow past both budget and timeline by roughly 50 percent. What surprised him even more was the reaction from people around him. By industry standards, they told him, that was not unusual. That experience pushed him to look deeper into why building projects so often run late, cost more than expected, and feel chaotic long before anyone picks up a hammer.

MeltPlan is focused on pre-construction, the planning phase that happens before physical work begins. In Kanav’s view, that phase is where construction projects are either set up for success or quietly put at risk. A building can spend months, and in some cases years, in planning before the first hole is dug, because architects, structural engineers, mechanical engineers, electrical engineers, and other specialists all need their work to align. If one part changes, the rest often has to change with it.

MeltPlan is building tools to help teams make sense of that complexity. Today, the company helps with code research and is moving into bid management and bid leveling. Kanav’s broader vision is a system where architects and engineers can upload plans and quickly get answers to the questions owners care about most. Is the design code-compliant? What will it cost? How long will it take to build? Are there changes that could save money or improve returns?

Kanav had already built a billion-dollar company in healthcare, so construction was not the obvious next chapter. But what he saw in construction was a problem hiding in plain sight. The industry, he argues, is not failing because people are not working hard. It is failing because the work is fragmented.

He describes a world where a large project can involve around 40 different skills spread across 30 companies and 100 people. That makes coordination hard, accountability blurry, and mistakes expensive. In his telling, the issue is less about individual performance and more about the number of handoffs. Too many moving pieces. Too many chances for something to break.

The other thing that caught his attention was the timing. Construction runs on documents, drawings, images, and plans, all formats that traditional software has struggled to interpret deeply. Kanav saw AI as the first real technology shift that might change that. Not because it solves everything today, but because it can finally start to read and reason through the kinds of materials construction depends on.

What makes MeltPlan interesting is not just that it uses AI. It is where the company is aiming it.

Kanav’s argument is simple. Construction problems do not begin in the field. They begin earlier, when plans are incomplete, misaligned, or slow to adapt. He puts it bluntly. The productivity game in construction is won or lost in pre-construction. Spend more time getting the plan right and the chaos during construction drops. Rush through planning and the problems show up later in steel, plumbing, scheduling, and cost overruns.

He also talks about AI with more restraint than hype. At MeltPlan, the test is not whether a model sounds smart. The test is whether it performs on the work. To measure that, the team assembled around 3,000 building inspector practice questions and used them as a benchmark for their code-research product. Kanav says the system improved from roughly 75 percent accuracy at the start to about 95 percent. That benchmark helped the company judge whether the system was actually improving.

Just as important, he does not frame AI as a replacement for human judgment. He compares AI agents to employees. They are useful, productive, and worth supervising. In his view, the best near-term use cases are the ones that wear humans down, like reading thousand-page specification documents or tracing code references across multiple sections. Creativity, strategy, and novel problem solving still belong to people.

Kanav shares the perspective of an owner as much as a founder. He describes the frustration of asking for a design change and waiting six to eight weeks to hear whether it is feasible. That lag, repeated across multiple iterations, can turn pre-construction into a year-long slog. His goal is to compress that cycle dramatically, so owners can ask for dozens of changes and get answers back in a day or two instead of over several weeks.

He does not provide a detailed customer case study with named metrics in the material provided. The strongest example here is his own experience with renovation overruns, slow feedback loops, and a process that feels much more manual than it should.

MeltPlan appears to sit at the intersection of several groups that rarely have a clean, shared view of the same project. Kanav specifically mentions architects, engineers, inspectors, design-build firms, and owners as users who benefit from faster code research, clearer planning, and quicker answers about project feasibility.

Owners may be the most obvious beneficiaries. They are the ones asking basic but consequential questions throughout planning. Can this change be made? How much will it cost? What happens to the schedule? Today, those answers can take weeks. MeltPlan is trying to turn that into something closer to real-time decision support.

There is also a broader audience in the background. Teams that spend hours doing repetitive, mentally draining work. Kanav believes AI is especially well suited to tasks that create fatigue through sheer volume rather than difficulty. In that sense, MeltPlan is not only trying to speed up projects. It is also trying to reduce the grind in how projects get planned.

Kanav’s long-term picture of construction is vivid. By 2045, he expects much more prefab construction, with more building components produced offsite and assembled later. He sees robotics fitting naturally into factory environments where conditions are controlled, rather than across every unpredictable job site. But that future depends on better planning. Prefab works best when plans are locked in early rather than constantly shifting.

That is where MeltPlan fits. Kanav’s five-year vision is a workflow where design teams check plans into the system continuously and receive rapid feedback on compliance, cost, build time, and profit implications. The goal is not a slightly faster version of today’s process. It is a different one, where pre-construction stops being a long bottleneck and becomes an active, iterative feedback loop.

He also sounds restless in the way many founders do when the market opportunity is larger than the first product. Customers are already asking MeltPlan to solve adjacent problems. He mentions code research today, bid management and bid leveling next, and a longer list still waiting. His insomnia, he jokes, comes from wanting to get there faster.

MeltPlan is chasing a deceptively simple idea. Fix more before construction begins. That means treating planning not as paperwork before the real work starts, but as the place where the real leverage lives.

Kanav Hasija did not enter construction romanticizing the industry or dismissing it. He came in frustrated, curious, and convinced that the people in it were already working hard. What he believes they need is better coordination, faster feedback, and tools that can finally understand the mass of documents and decisions that shape every building. If he is right, the future of construction may look less like chaos on a job site and more like clarity upstream, when it still matters most.