Construction projects often go over budget by 30% and take 20% longer than planned because fragmented pre-construction planning makes teams commit to scope, cost, and code without complete data, leading to change orders and rework.
Design tools cover aesthetics and execution, and software manages on-site control, but the important early-stage trade-off optimisation is still done manually and in isolation.
MeltPlan fills this gap with its “Planning Engine,” an AI system built specifically for construction that checks constraints such as code, cost, schedule, and value before plans are finalised.
The San Francisco startup, founded in 2025, recently raised $14 million in Seed funding led by Bessemer Venture Partners, with noa and WND Ventures joining in. This funding will speed up the development of their connected decision systems.
Making construction ‘boring’ by putting the real work into planning
Kanav Hasija, CEO and former co-founder of the $3 billion health tech unicorn Innovaccer, brings big-picture experience to tackle construction’s productivity challenges. His co-founder, Tanmaya Kala, COO, has years of hands-on experience as a project executive at DPR Construction, managing multimillion-dollar commercial, healthcare, and life sciences projects.
Their goal is to make construction “boring” by moving the heavy lifting to planning and highlighting trade-offs early to cut down on rework. They’re applying Hasija’s healthcare experience to the $14 trillion construction industry, using AI to catch mistakes before fixes become costly.
MeltPlan rejects generic AI wrappers and builds a construction-native system that already scores 95%+ on building inspector exams. Its Planning Engine comprises four integrated modules: code, cost, schedule, and value.
Jess Clemans, Principal, noa, shares with TFN, “MeltPlan is building an end-to-end preconstruction platform with cutting-edge AI tools that power every part of the workflow. The launch of the first three product verticals is going to be announced very soon, so readers should watch this space.”
Clemans adds, “MeltPlan’s unique moat is their deep product experience and technical acumen, building proprietary AI models from scratch — not fine-tuned off-the-shelf models. These achieve 93–98% accuracy, compared to roughly 85% for commercial RAG platforms.”
What stands out?
Expert reasoning that actually understands construction methods and procurement limits, full transparency (no black boxes), and shifting decisions earlier.
Competitors like Autodesk Construction Cloud (focused on execution), Procore (field management), and Reconstruct (BIM-focused) exist, but none bring together pre-construction tradeoffs using AI at scale.
What’s next?
MeltPlan is already piloting with DPR Construction in California and Innovo Group in the UAE, aiming to slash planning errors.
The big vision is to make pre-construction the most intense and most profitable phase in the industry, finally closing the productivity gap that’s stumped ConTech for decades.
