Communication & Coordination

It is worth stating plainly, because it reframes the whole topic: coordinating information is not a support function of construction management - it is much of the job. A project manager's day is largely spent moving information between people who need it: relaying a design decision to the site, chasing an answer to an RFI, making sure the trade that pours the slab knows the trade that laid the conduit has finished, telling the client what changed and why, ensuring a safety instruction reached every gang. The physical work is done by others; the manager's distinctive contribution is keeping the vast web of people aligned on a shared, current picture of the project. When that web is coherent, work flows in the right sequence; when it frays, the consequences are immediate and expensive.

And it frays constantly, because construction communication is fragmented and lossy by nature. Information lives in silos - the designers' models, the site's daily reports, the subcontractors' own systems, a hundred email threads and WhatsApp groups - that do not join up. It passes through many hands, and detail is lost or distorted at each handover, like a game of telephone with money at stake. It crosses boundaries of language, discipline and organisation, each with its own vocabulary. And it arrives late, or not at all: the report no one had time to read, the change that reached the office but not the gang, the decision made in a meeting that half the affected people never heard about. The results are the familiar failures - a crew building a superseded design, two trades clashing because their sequences were never reconciled, a delay that rippled because it was not flagged, a dispute rooted in "you never told us." This is the chronic, unglamorous pain that AI communication tools aim at: not a lack of information, but a failure to move the right piece of it to the right person in time. Understanding that the problem is flow, not volume, is the key to seeing what AI can and cannot fix.

AI's contribution to coordination is to make the flow of information faster, lighter and less lossy - and there are four genuinely useful moves. Summarise: a project generates far more reporting than anyone can read - long daily and progress reports, hour-long meeting recordings, sprawling message threads. AI can condense these into the decisions taken, the issues raised and the actions agreed, so a manager gets the substance of a two-hour meeting in two minutes and a week of site reports in a page. Extract and route: beyond summarising, AI can pull out the discrete action items - who agreed to do what, by when - and help route each to the right owner, turning a meeting into a tracked to-do list rather than a memory that fades. It can help direct an RFI or an issue to the person best placed to handle it, so things stop falling into the gap between roles. Translate: on multilingual sites - a daily reality in India and on international projects - AI translation can bridge languages between office and site, between a specification and a worker, between trades who do not share a tongue, lowering a real and dangerous communication barrier. Roll up and surface: AI can aggregate status from many sources into a coherent picture, and surface the item that needs attention from the noise.

What unites these is that AI acts as an intelligence layer over the project's communication - reading the flood, distilling it, and helping the right piece reach the right person. For an industry where coordination failure is a leading cause of trouble, that is a real gain. But the same discipline applies as everywhere in this course. Every AI output is an input a human relies on, and it can be wrong: a summary can drop the one decision that mattered or misattribute an action, a translation can garble a safety-critical instruction, a routing can send an issue to the wrong person. So summaries and action lists must be verified before they are trusted and acted on - especially where safety or contract is involved - and the manager remains responsible for what was communicated. AI can carry the information faster; it does not take over the duty to make sure people actually understood.

Here is the boundary that keeps this honest, and it is different from the data-and-accountability limits elsewhere in the course: some of the most important parts of coordination are not information problems at all, and AI cannot touch them. Moving data between people is one thing; the human dimension of construction is another. Trust is what makes coordination actually work: a foreman acts fast on a manager's word because of a relationship built over months, not because a system routed a task to them. Negotiation resolves the daily conflicts of a build - a clash between trades, a dispute over a variation, a supplier who needs to be persuaded to prioritise you - and it turns on reading people, giving and taking, and understanding interests, none of which an AI does. Motivation gets a tired crew to pour before the rain or fix a defect properly rather than hide it, and it comes from leadership and respect, not a notification. Reading the room - sensing that a subcontractor is quietly in trouble, that a client is losing confidence, that a safety culture is slipping - is a human judgement that precedes any data, and often the most important thing a good manager does.

The risk of AI communication tools is that they tempt a project to mistake the flow of information for the running of the project - to believe that if the summaries are circulated and the actions routed, coordination is handled. It is not. A perfectly summarised meeting whose decisions no one is committed to changes nothing; a flawlessly routed action ignored because the relationship is broken is not an action. Worse, over-relying on AI-mediated communication can erode the very relationships that make a project work: the manager who sends AI-drafted updates instead of walking the site and talking to people loses the trust, the informal knowledge and the early warning that only human contact provides. So the competent stance is to use AI to remove the drudgery of moving information - the summarising, extracting, routing and translating that clogs a manager's day - precisely so there is more time for the human work that AI cannot do: building trust, negotiating, motivating, and being present on the site. AI handles the pipes; people run the project.

Using AI coordination tools well comes down to a few disciplines. Treat summaries and extracted actions as drafts to verify, not truth to circulate: before a meeting summary or an action list goes out, someone who was there checks it caught the real decisions and attributed them correctly - a wrong summary that everyone trusts is worse than no summary. Guard the safety-critical channel especially hard: a translated or summarised safety instruction must be confirmed understood, not just sent, because a garbled hazard warning can kill. Use the time AI frees, deliberately, for the human work - more presence on site, more direct conversation, more relationship-building - rather than letting AI-mediated communication replace human contact and quietly hollow out the trust the project runs on. And keep the accountability clear: the manager owns what was communicated and understood, whoever or whatever drafted it.

The Indian context makes both the promise and the boundary vivid. Indian construction sites are often highly multilingual - workers, supervisors, engineers and clients may speak several different languages - so AI translation to bridge those gaps addresses a genuine, sometimes safety-critical barrier, and is one of the more immediately valuable applications on organised sites. At the same time, much of Indian construction runs on relationships, informal networks and verbal coordination, especially in the vast small-scale and informal segment, where the human, trust-based dimension of coordination is not a nicety but the actual operating system of the project - and where digitised communication tools may have little purchase at all. The honest picture, then, is the same as the rest of the module: AI can genuinely ease the flow of information on organised, digitised projects - summarising, routing, translating - and free managers for the human work; but it augments coordination, it does not run it, and the trust, negotiation and presence that actually hold a build together stay human. Binding decisions communicated through these tools - a safety instruction, a contractual notice, an instruction to change the works - remain the responsibility of the accountable people and are governed by the contract and the law, never by the tool that carried them.