Every major Maratha military operation documented in the historical record followed the same sequence: gather intelligence, confirm capability and disposition of the enemy, identify the specific vulnerability to be exploited, then move. The intelligence step was not optional and was never shortened. Shivaji did not launch opportunistic raids; he launched operations whose outcomes had been shaped before the first soldier moved by a prior period of information gathering that could last weeks.
The plain version: the Maratha army was already fighting before anyone drew a sword, because the intelligence network had been fighting for weeks in advance — identifying weaknesses, mapping routes, counting weapons, conditioning targets. By the time the physical operation began, many of the key decisions had already been made.
Shivaji's intelligence chief is named in the historical record: Bahirji Naik. That a 17th-century Indian ruler had a named, identifiable head of intelligence — as opposed to a general practice of spying — suggests an institutionalized intelligence function rather than ad hoc information gathering.1
Bahirji Naik's documented activities include:
The Bahirji Naik documentation is thin by modern standards — Purandare's account is the primary narrative source, and primary documents are not cited — but the pattern of intelligence-informed operations is consistent enough across the documented campaigns that it likely reflects genuine institutional practice rather than retroactive attribution.
The 1663 night raid on Shaista Khan in the Lal Mahal (Pune) is the clearest example of intelligence as the primary enabler of a military operation. Shivaji's force used a wedding procession as cover to approach the mansion at night; they entered through the kitchen rather than any guarded entrance; they found Shaista Khan in his private quarters. This level of operational precision required:
This is not the intelligence gathering of scouts reporting enemy troop positions. This is penetration intelligence — the kind that requires sources inside the target environment, either planted agents or corrupted insiders. The operational precision of the Lal Mahal raid is only achievable with that level of prior access.
The basket escape from Agra (1666) is typically read as an escape story. It is more precisely an intelligence and behavioral conditioning operation. Shivaji did not escape on impulse; the preparation took weeks:1
This is intelligence tradecraft — conditioning the environment in advance so that the operation, when it happens, appears to the target to be a continuation of normal patterns. The escape required weeks of patience in conditions of extreme danger, with Mughal guards watching the house and Aurangzeb's court potentially ordering execution at any moment. The capacity to sustain that patience while running a deliberate conditioning operation is a specific psychological and organizational capability.
The consistent pattern across the Surat raid, the Afzal Khan ambush, the Lal Mahal operation, and the Agra escape is what might be called intelligence-first sequencing: no operation begins until the intelligence phase has confirmed the specific vulnerability that the operation will exploit, and the intelligence has been gathered with enough specificity to shape the operational design.
This is different from "gathering information before you fight," which is standard military practice. It is a doctrine in which the intelligence determines what operation is possible, rather than the operation determining what intelligence is needed. The sequence is: intelligence → possible operation (from that intelligence) → operation design → execution. Not: intended operation → intelligence needed for that operation → gather it → execute.1
The distinction matters because it changes what the intelligence function is for. In most military organizations, intelligence supports a pre-determined operation. In the Maratha system as Purandare describes it, intelligence determines which operation is possible at all.
[POPULAR SOURCE] — Bahirji Naik's activities are described in Purandare's narrative without primary-source citation. The interior layout of the Lal Mahal and the specific details of the basket conditioning are from Marathi chronicle traditions whose reliability is uncertain. The consistent pattern across multiple documented campaigns is the strongest evidence for the claim — individual episodes could be legendary elaboration; the pattern is harder to dismiss.1
Tension with Sun Tzu's five spy types: Sun Tzu's taxonomy (local/inward/converted/doomed/surviving spies) treats the converted spy as the most valuable intelligence asset because they provide access to the enemy's internal information. The Lal Mahal penetration suggests the Maratha intelligence network included either a converted-spy equivalent (someone inside the Mughal establishment reporting back) or a local network of informants. Sun Tzu's framework would classify this as an institutional intelligence capability, not just tactical scouting.
History — Sun Tzu Intelligence and the Five Spies: Sun Tzu — Intelligence and the Five Spies — Sun Tzu's argument that "foreknowledge cannot be obtained by induction from experience, nor by calculation, nor by any other means except from men" is validated by the Maratha case. The Surat raid's success depended on Bahirji Naik's human intelligence — not on calculation of Surat's defenses from public information but on embedded sources reporting from inside. The two sources together produce a principle neither contains alone: foreknowledge from human intelligence is not just tactically superior to other intelligence types — it enables categories of operation (the Lal Mahal penetration, the Agra conditioning) that are simply impossible with signals or observational intelligence alone.
History — Sun Tzu Field Intelligence and Signal Reading: Sun Tzu — Field Intelligence and Signal Reading — Sun Tzu describes passive signal reading (dust clouds, bird flight, soldier posture as indicators of enemy state). The Maratha intelligence system as described goes further: it actively shapes the signal environment (the sweetmeat conditioning at Agra) so that the enemy's own perceptual system is manipulated. Passive reading of signals versus active shaping of the signal environment — both involve intelligence, but they operate in opposite directions. Together they define the full spectrum of intelligence-as-operational-tool.
The Sharpest Implication The intelligence-first sequencing doctrine inverts the conventional relationship between intention and information. Most actors decide what they want to do and then gather intelligence to support that decision. The Maratha doctrine decides what the intelligence makes possible and builds the operation from there. The inversion is uncomfortable because it requires suspending the goal until the information arrives — which requires tolerating uncertainty about whether a viable operation exists, potentially for weeks or months. Most decision-makers can't sustain that tolerance; they default to the intended operation and use intelligence to confirm it rather than to determine it. The Maratha cases suggest that the operations that seem most brilliantly conceived (the Lal Mahal penetration, the basket escape) were not conceived in advance and then executed; they were made possible by prior intelligence that revealed which specific vulnerability existed. The brilliance was in the patience to wait for the intelligence to determine the operation.
Generative Questions
The doctrine described here now has its named architect on his own page: Bahirji Naik the Spymaster traces the spy-chief from the camp at Afzal Khan to the alleys of Surat. The route reconnaissance that fed the first big raid is documented in The First Sack of Surat 1664 — weeks of advance work before a soldier moved. And the Agra basket escape, read above as conditioning-as-operation, gets its own granular treatment in The Sweetmeat Basket Escape, where the routine deliveries that disarmed the guards are followed deed by deed.