Manual, Automated & Intelligent

The control ladder has three clear rungs, and naming them precisely prevents most of the muddled thinking in this area. The first rung is manual: a person operates the adaptation directly - opens the window, cranks the shutter, slides the partition, tilts the blind. The intelligence, the sensing and the decision all live in the human; the building element is a simple, passive mechanism that a hand moves. The second rung is automated: a machine operates the adaptation without a person doing the physical work - an electric motor, actuator or mechanism driven by a timer, a simple switch, or a basic trigger such as a rain sensor closing a skylight. The decision is pre-set and mechanical, not thoughtful; the system does the same thing at the same time or in response to one simple signal, every time. The third rung is intelligent: the adaptation is driven by sensing and computation - sensors perceive conditions (sun, temperature, occupancy, air quality), a controller computes what to do, and actuators carry it out, often adjusting continuously and combining many inputs.

The essential thing to see is that these rungs differ not in what moves but in where the sensing and decision-making live. In manual control, sensing and decision live in the person. In automated control, the decision is frozen in advance into a timer or a simple trigger, and there is no real sensing of the wider situation - a timer-driven blind lowers at three o'clock whether the day is bright or overcast. In intelligent control, sensing and decision are embedded in the building itself, which perceives and computes in real time. This is exactly the boundary between the course's three strands turned into a practical control choice: manual and automated can both be purely kinetic (movement without intelligence), while the intelligent rung is where the smart strand - sensing and computation - genuinely enters.

Each rung, crucially, is a superset of the burdens of the one below plus its own. A manual element needs a mechanism and a person. An automated element needs all that plus a motor or actuator, a power supply, a control device and the wiring between them. An intelligent element needs all of that plus sensors, a controller, software, calibration, and the integration of the whole into a working system. So climbing the ladder is not swapping one thing for a better thing - it is adding layers, each of which costs money, consumes power, and introduces its own ways to fail. This is why the ladder is a trade-off and not simply a hierarchy of better options, and why the choice of rung deserves as much thought as the decision to adapt at all.

The manual rung is the oldest, the cheapest, and - this lesson insists - very often the best, yet it is the one designers reflexively skip past in the rush toward automation. In manual control the human being who is already present does the sensing, the deciding and the moving: they feel the room is stuffy and open the window, see the low sun and adjust the blind, decide the space needs dividing and slide the partition. This is an extraordinary bargain. It requires no motor, no power, no sensor, no controller, no software, and no wiring - just a well-designed mechanism and a handle within reach. It cannot suffer a power cut, a sensor drift, a software bug, or a controls failure, because it has none of those things. And it draws on the most sophisticated, adaptable and freely available controller in existence: a person's own judgement, which effortlessly weighs comfort, mood, activity and preference in ways no automated system matches.

Manual control has real limitations, and honesty requires naming them. It needs a person present and willing to act - it cannot adapt an empty building, or respond to a fast change nobody is there to catch, or perform a continuous fine adjustment a person would find tedious. It depends on the occupant knowing how and remembering to operate it, which good, intuitive design must support. And it cannot do the genuinely superhuman - track the sun's position every few minutes across a whole facade of hundreds of louvres, or manage a system too large or too fast for hands. Where these limitations bite, the higher rungs earn their place. But the limitations are narrower than the automation instinct assumes: a great many adaptations are occasional, happen while people are present, and are exactly the kind of thing a person does well.

The manual rung is also the heart of the Indian and vernacular case this course keeps making. The operable shutter, the sliding screen, the adjustable verandah, the opened-and-shaded courtyard are all manual adaptive architecture, and they have delivered comfort reliably for centuries precisely because they have nothing to break and depend on nothing but a hand. In a context of uneven maintenance, dust and monsoon that punish mechanisms, and variable power that disables automation, the manual rung's robustness is not a quaint limitation but a decisive advantage. When a brief calls for adaptation and people will be present, the honest first proposal is almost always a well-designed manual element - and the burden of proof lies on anyone who wants to climb higher.

The automated rung adds a machine to do the physical work and a simple, pre-set rule to decide when. An electric motor, a linear actuator or a powered mechanism moves the element, driven by a timer (lower the blinds at three), a schedule, a simple switch, or a single-signal trigger (a rain sensor that closes a skylight, a wind sensor that retracts an awning). Automation buys two genuine things the manual rung cannot offer: it can operate when no person is present or willing, and it can perform repetitive or tedious adjustments consistently without relying on someone remembering. For an unoccupied building, a large array that would exhaust a person to adjust by hand, or a change that must happen reliably at a set time, automation earns real value.

But the automated rung also introduces the whole apparatus of powered machinery and everything that can go wrong with it, and this is where the honest costs begin in earnest. Now there is a motor that can burn out, an actuator that can jam, a mechanism that wears with every cycle toward fatigue failure, a power supply that can be interrupted, wiring and connections that can corrode or fail, and a control device that can break - none of which a manual element has. Every one of these is a maintenance item and a potential point of failure, and when any of them fails, the automated element usually fails closed or fails stuck, often worse than a manual element that a person could simply move by hand. Automation also consumes energy to run, and its pre-set logic is dumb: a timer-driven blind lowers on a dark, cool day when it should not, because it senses nothing and merely obeys the clock.

The automated rung therefore earns its place only where its two genuine advantages - operating unattended and performing tedious repetition reliably - clearly outweigh the added cost, energy, wear and failure modes it brings, and where the maintenance to keep it working is genuinely assured. The trap is automating for its own sake: motorising an element that a present occupant could happily operate by hand, thereby paying for a motor, power and controls, accepting new failure modes, and gaining nothing but the appearance of sophistication. Before specifying automation, the honest questions are whether anyone is actually there to operate a manual version, whether the change is truly too frequent or tedious for a hand, and whether the building will really receive the ongoing maintenance the motor demands. In much of India, that last question alone sends many automated proposals back down to the manual rung.

The intelligent rung is the top of the ladder and the one the field romanticises most. Here the adaptation is driven by genuine sensing and computation: sensors read the sun, temperature, glare, occupancy and air quality; a controller computes the best response, often balancing many inputs and even learning patterns over time; and actuators carry it out, adjusting continuously and precisely across a whole system. Intelligent control can do things the lower rungs genuinely cannot - track the sun across a facade in real time to optimise daylight and glare for the tasks inside, respond to conditions no person could monitor continuously, coordinate many elements together, and hold performance near an optimum that shifts constantly. Where the optimisation is genuinely complex, variable and continuous, and the benefit is real, the intelligent rung is the only rung that can capture it.

But the intelligent rung also carries the full, stacked weight of every layer below plus its own formidable additions, and this is the honest reckoning the romance omits. On top of the mechanism, motor, power and wiring of the automated rung, it adds sensors that drift, foul and fail; a controller and software that can be misconfigured, buggy, or badly tuned; calibration that must be maintained; integration between many components that can break at any interface; and a dependence on specialist skills to commission, understand and maintain the whole. It is the most expensive rung to build, the most expensive to run, the most complex to keep working, and the one with by far the most failure modes - and when it fails, it often fails obscurely, leaving occupants unable even to understand why the building is behaving as it is, let alone fix it. The intelligent facade that is magnificent when commissioned is the one most likely, a decade later, to sit seized and switched off in the maintenance graveyard.

So the lesson closes where it began: higher is not better, and the disciplined choice is the lowest rung that genuinely meets the need. Match the rung to the job. Occasional change with people present: manual, almost always. Frequent, predictable change with nobody to operate it: automated, if the maintenance is assured. Complex, continuously variable optimisation with a real, large benefit and guaranteed upkeep: intelligent, and only then. The honest, uncomfortable truth for a field in love with intelligence is that a great many adaptive needs are best met by a person and a well-made handle, that automation is over-specified far more often than it is justified, and that the seized intelligent facade would frequently have been beaten by an operable shutter its occupants could have worked forever. Climb the ladder only as far as the need truly reaches - and defer the binding controls, actuator, sensor and integration engineering of any automated or intelligent system, and its maintenance regime, to qualified specialists and tested manufacturer systems.