AI is reshaping how construction and real estate companies connect data, systems and decisions. Speaking to KAVITA PARAB, Siddharth (Sid) Mitra, COO and CTAIO at sirrus.ai, explains how an agentic AI layer can connect siloed systems, optimise project schedules, identify risks in real time and unlock latent enterprise value.How does sirrus.ai connect market intelligence, project feasibility, construction planning and management to help developers make better decisions?sirrus.ai is more than a construction platform. We have created what we call an agentic plane, starting with an ontology — a set of entities connected through defined relationships. This can bring together project execution, markets, pricing, construction planning and other parameters that influence a developer's decisions.We then apply reasoning engines over context and knowledge graphs. Real-time construction updates, market intelligence, sales and revenue data can be brought together to create an intelligent conversational platform. Developers can understand project expenditure, sales velocity, risks and project status, and assess different scenarios before making decisions.Unlike conventional platforms that solve one problem at a time, we are bringing multiple entities together through a conversational, agentic platform.How does the platform help developers manage the cost and time impact of delays?A project schedule is a fully loaded schedule containing the BOQ, costs, materials, timelines and milestones. Our proprietary algorithm optimises it against cost, time or a combination of both. For instance, a developer can choose to optimise 60 per cent for cost and 40 per cent for time.The algorithm recommends workflows that then go through an approval process involving project managers or task owners. It also factors in events beyond the project's control, such as geopolitical situations, natural events and other external factors.The system continuously considers site progress and current signals, incorporating them into the micro-schedule and overall schedule. This brings together optimisation, approvals, site progress and external risks, which are generally handled in silos.The solution is not completely autonomous. Recommendations still go through a maker-and-checker process. We have developed this as a proprietary algorithm and filed patents for the technology.What site-level data does the platform capture?Data can come through an app, WhatsApp for Business, email or voice messages. When a task is updated, the system can capture its start and end dates, progress percentage and supporting evidence such as photographs.This information flows into the micro-schedule, then the look-ahead schedule and ultimately the overall project plan. So a site update can cascade into the daily progress report, look-ahead and full project schedule.How does the platform improve coordination and identify potential risks?Coordination is critical to large construction and infrastructure projects. You need a central predictive engine that can bridge communication gaps between developers, contractors and project teams.Our AI engine connects schedules, BIM models, daily logs and RFIs. If a site update indicates that a task has not been completed as planned, that information flows through the micro-schedule and look-ahead to the relevant milestone, where a risk can be flagged. This gives the developer time to take an alternative course of action.The platform can also track real-time data from cameras, laser scans and IoT devices against BIM models and the master schedule. AI can identify design discrepancies through BIM clash detection and monitor resource and material bottlenecks using supply chain and other external data.Because these entities are connected through an agentic plane, we can identify execution risks before they translate into project delays.How does sirrus.ai integrate with existing ERP, CRM and project systems?We are not trying to replace systems such as SAP or Salesforce. The objective is to make these systems talk to each other.We do not introduce ourselves as a new system of record. We sit on top of existing systems and extend them with an AI layer. The agentic plane consumes project, customer, sales and marketing data and brings it together into one intelligent engine.We call this AI interoperability or the synchronisation of agents. We can also plug in third-party agents and orchestrate them alongside our own. The architecture is designed to be extensible, with the ontology, reasoning layer and control plane working together to orchestrate agents.What should developers consider when adopting construction technology?The short answer is interoperability.Whether there is one system or multiple systems, the important question is whether they can talk to each other. A design system should be able to communicate with the scheduler; the scheduler should provide milestone information to finance; and planning and budgeting should receive this information in real time.I don't have a problem with multiple systems performing specialised functions, provided they communicate. What is required is a common platform that brings together project delays, risks, milestones and external factors.The next step is to make this AI-enabled, with construction, scheduling, DPR and finance agents communicating with each other so decisions can be taken faster.What is holding developers back from adopting this architecture?Trust is a key factor, particularly around data privacy and how organisational data is used. There is concern about whether data will leave the company's environment or be used by AI language models for training.I believe this will be addressed through structured AI platforms. We see a future where we will not position ourselves as a new system of record. Instead, we will be a plane into which companies can bring their existing systems and get intelligence from the combined data.The BIM system could talk to design management, which talks to project scheduling, which talks to finance. This would enable faster decisions around budgeting, planning, sales and revenue.Can you share a use case?One key use case has been a real-time MIS for a client. Traditionally, MIS is generated periodically to understand the financial position of a project or portfolio.We created a real-time MIS that provides visibility into impending risks and helps identify how they can be mitigated. It can also pick up external signals that could delay a project or milestone and recommend ways to address them.This is more than an AI problem; it is also a data governance challenge. Departments often do not want to share data with one another. Once that is addressed, AI can take care of much of the rest.The client moved from receiving visibility once a month to having it in real time. AI can unearth a significant part of the enterprise value that remains unrealised because organisations may lack the right processes, people or connected systems.What are your plans for the next three to five years?We want to solve business problems and, once we solve them, redefine how that industry operates, starting with real estate.We don't believe in the conventional SaaS model of developing an off-the-shelf product and then customising it. We want to work directly with clients to solve specific requirements. If an agent already exists in our marketplace, we can deploy it. If not, we can create one for that specific need and add it to the marketplace.Over the next five years, our plan is to create an exhaustive marketplace for agents, starting with real estate and eventually extending across other industries.