**Britain's New AI Targeting Technology Is Legitimately INSANE — Transcript & Summary | SozAI** Source: https://sozai.app/transcript/britain-new-ai-targeting-technology/ Analysis of Britain's Project Asgard, an AI-driven targeting network that reportedly cut a corps-level planning cycle from 72 hours to one, built with German and American firms and trained on Ukrainia ## Key Takeaways - Asgard reportedly compresses corps-level targeting from days to hours, a major claimed operational advance. - Its speed was enabled by a departure from traditional British defence procurement, using rapid contracting and multiple vendors. - The architecture is modular and built on open standards, enabling interoperability with NATO allies. - Human oversight and accountability at machine-like speed remain the central unresolved question. - Performance claims are self-reported and need independent validation before being taken at face value. ## What the video covers - A British corps-level planning cycle reportedly dropped from 72 hours to one hour in a live exercise, as reported by General Sir Rolley Walker in June 2026. - Project Asgard was designed to replace slow, layered 'swivel chair' military decision-making with a data network operating at near machine speed, inspired by lessons from Ukraine. - The 'Recce Strike' concept fuses reconnaissance and strike into a single fast loop to shorten the sensor-to-shooter chain. - Asgard went from announcement (October 2024) to a prototype deployed in Estonia (May 2025) in roughly seven months, unusually fast compared with programs like Ajax. - The program used a rapid, commercially inspired procurement model: contracts went to 26 companies, including Helsing, Anduril UK, Systematic, and CGI, with prototypes accepted before perfection. - The system chains together components: Helsing's Altra edge AI classifies drone video, Anduril's Lattice carries data, SitaWare handles command and control, and PRIISM performs collateral damage estimates and legal review. - AI classifies targets while human legal advisors and commanders approve strikes, though the transcript questions whether oversight remains meaningful at operational speed. - Exercise Arcade Strike in May 2026 tested Asgard at corps scale, using a compact underground headquarters designed to improve survivability. - The transcript notes that the performance claims come from the army and MOD evaluating their own system, so the metrics deserve scrutiny. - Concerns raised include dependence on Ukrainian-derived training data, reliance on communications, and whether a faster system can be matched by sufficient effectors to act on its outputs. ## Chapters 1. 00:00 The 72-Hour Problem: Why Targeting Speed Matters 2. 01:22 Lessons from Ukraine and the Recce Strike Concept 3. 02:53 How Asgard Was Built: Speed Through New Procurement 4. 04:48 Exercise in Estonia: Testing the Prototype 5. 06:57 Inside the Asgard Stack: From Drone to Strike 6. 10:03 Scaling to Corps Level: Exercise Arcade Strike 7. 13:20 Strategic Questions: Data Sources, Oversight and Sovereignty 8. 16:40 Conclusion and Open Questions Answers ## Questions about this video What does Project Asgard reportedly achieve? According to General Sir Rolley Walker, a corps-level planning cycle that used to take 72 hours was reduced to one hour during a live exercise, and the number of targets processed per day rose roughly tenfold. The transcript notes these are self-reported results. How was Asgard developed so quickly? The program moved from announcement to a field prototype in about seven months by using a rapid, Ukraine-inspired contracting approach: identifying the problem, contracting multiple suppliers quickly, and accepting a working prototype rather than writing exhaustive requirements first. What concerns does the video raise about Asgard? The main concerns are whether human oversight and legal review remain meaningful at operational speed, the dependence on Ukrainian-derived training data under agreements that could change, and whether the system's performance claims have been independently verified. ## Full Transcript — Download SRT & Markdown 00:00 Speaker A A British core level planning cycle, the kind that coordinates tens of thousands of troops, decides which targets get struck and sequences everything from rockets to drones, used to take 72 hours. 00:13 Speaker A Now, it takes one. 00:25 Speaker A That figure came from General Sir Rolley Walker, Chief of the General Staff, at the RUSI Land Warfare Conference in June 2026. 00:35 Speaker A Not a prediction, a reported result from a live exercise. 00:44 Speaker A The AI software doing this was built by a German company and an American firm. 00:55 Speaker A The training data came from Ukrainian battlefields. 01:11 Speaker A So, what is actually British about Britain's new AI targeting technology? 01:22 Speaker A To understand why this matters, go back to Ukraine, because the British Army certainly did. 01:37 Speaker A The lesson Ukraine posed from 2022 onwards was brutal in its clarity. 01:49 Speaker A A drone operator in a Ukrainian trench can now find a Russian armored vehicle, generate targeting data, and have a loitering munition on the way before a British divisional headquarters has finished its first coordination meeting of the morning. 02:00 Speaker A That gap between the speed at which a modern sensor-to-strike chain can operate, and the speed at which a 20th century military bureaucracy makes decisions, is where Project Asgard begins. 02:05 Speaker A The British Army's response is called the Recce Strike concept. 02:13 Speaker A Fuse reconnaissance and strike into a single fast loop, so that finding a target and engaging it happens within the same tight sequence, rather than through multiple command layers that each introduce their own delay. 02:31 Speaker A The Army calls the old process swivel chairing, the image of soldiers in a headquarters turning their chairs to brief each other, each pass adding minutes and hours to the clock. 02:53 Speaker A Asgard is the digital architecture designed to replace that with a data network that gets the right information to the right person at something approaching machine speed. 03:06 Speaker A How it was built turns out to be almost as interesting as what it can reportedly do. 03:23 Speaker A July 2024, senior army leadership begins discussions about what this system would need to be. 03:40 Speaker A October 2024, the defense secretary announces it publicly. 03:49 Speaker A January 2025, contracts go to 26 companies from the British and American defense tech ecosystem, including Helsing, Anduril UK, Systematic, and CGI. 03:55 Speaker A May 2025, a working prototype deploys to Estonia for a live NATO exercise. 04:12 Speaker A Seven months, concept to field. 04:22 Speaker A I cannot overstate how unusual that is. 04:27 Speaker A The program that produced Ajax, the army's armored reconnaissance vehicle, took 15 years, cost billions more than planned, and nearly broke the soldiers who had to test it. 04:34 Speaker A Asgard went from announcement to Estonia in roughly the time some programs take to agree on what they want. 04:40 Speaker A The approach that made this speed possible was explicitly borrowed from the commercial contracting methodology the Ministry of Defense developed to rapidly supply Ukraine. 04:53 Speaker A Instead of writing exhaustive requirements, running a lengthy competition, and then delivering something years after the requirement had changed, the Asgard team identified the problem, brought in suppliers capable of solving it quickly, and contracted them within months. 05:00 Speaker A Rapid iteration, competition between a group of vendors rather than reliance on a single prime contractor, and acceptance of a working prototype rather than a theoretically perfect final product from day one. 05:13 Speaker A That philosophy is genuinely unusual in British defense procurement, which has a well-documented tendency to do the opposite. 05:28 Speaker A Whether it survives the harder and more expensive scaling phases is something I want to come back to. 05:40 Speaker A But first, what actually happened in Estonia in May 2025? 05:54 Speaker A The 4th Light Brigade, the Black Rats, deployed from Britain to Estonia in 48 hours and integrated into the Estonian Defense Forces division for the first time under NATO command. 06:01 Speaker A They dug trenches, applying lessons from Ukraine, generated two battalion-size battle groups, and tested Asgard alongside everything else they were doing. 06:12 Speaker A That last part did not appear in the Ministry of Defense's initial press release on the exercise. 06:25 Speaker A The capability was apparently sensitive enough that the army deliberately kept it out of the headline communications. 06:38 Speaker A It was running in the forests of Estonia, though, whether the press releases mentioned it or not. 06:50 Speaker A Asgard is not a single piece of hardware. 06:56 Speaker A It is a software targeting network, a connective layer linking sensors, commanders, and strike platforms so that information moves at something closer to machine speed. 07:04 Speaker A An ISR drone carries Altra, an edge AI package built by Helsing, a German defense AI company. 07:15 Speaker A Altra watches the drone's video feed, classifies what it sees, and generates a cursor-on-target message, a standardized open-source data format that different military systems can read regardless of who made them. 07:24 Speaker A That message travels through Lattice, Anduril UK's mesh network, into the soldier's Android Tactical Awareness Kit devices, and from there into SitaWare, the British Army's existing command and control system used from corps to platoon level. 07:36 Speaker A From SitaWare, the target enters PRIISM for collateral damage estimates, weapon-to-target matching, and legal review. 07:42 Speaker A A human legal advisor is present at this stage. 07:57 Speaker A The AI classifies, humans approve. 08:04 Speaker A Once cleared, a fire mission goes either to the artillery fire control application for a shell or rocket or to AltraStrike for a Helsing HX2 one-way attack drone. 08:18 Speaker A What previously took hours of briefings now takes minutes. 08:25 Speaker A The army called this result doubling lethality. 08:32 Speaker A General Walker told allies and industry partners at a London showcase in July 2025 that processes which once took hours now take seconds or minutes. 08:42 Speaker A The Ministry of Defense's own communications described the same transformation. 08:55 Speaker A Worth noting, those claims came from the army and the Ministry of Defense reporting on their own system in exercises they designed. 09:00 Speaker A That is not a reason to dismiss the results, but it is a reason to ask what they were actually measuring. 09:14 Speaker A The army was clear that Estonia was a brigade-level proof of concept, around 5,000 personnel. 09:24 Speaker A The next stated step was to scale Asgard to corps level in 2026, 20 to 45,000 troops, allied formations, and the full allied rapid reaction corps command structure. 09:31 Speaker A Whether a system that impressed in Estonia could hold together at that scale was the real question. 09:42 Speaker A The answer arrived in May 2026 in an underground space beneath Charing Cross Station in London. 09:49 Speaker A Exercise Arcade Strike was the first full exercise of the Allied Rapid Reaction Corps since it took command of the first and third UK divisions and almost all British Army brigades. 10:05 Speaker A NATO's strategic land reserve. 10:12 Speaker A In other words, the corps-level formation that exists to reinforce the alliance's eastern flank in a crisis. 10:20 Speaker A The corps headquarters for this exercise was not a large tented camp with dozens of terminals, hundreds of offices, and the kind of electromagnetic signature that invites an adversary's first strike. 10:32 Speaker A It was compact enough to run in a disused underground facility. 10:39 Speaker A That choice was not coincidental. 10:49 Speaker A The British Army has watched Russia systematically target Ukrainian headquarters in the first hours of any engagement. 10:55 Speaker A And a smaller, faster, harder-to-locate headquarters is a survivability decision as much as a technological one. 11:04 Speaker A What that headquarters produced during Arcade Strike is where the numbers become genuinely striking. 11:22 Speaker A General Sir Rolley Walker at the RUSI Land Warfare Conference in June 2026 put a specific figure on what Asgard Decide delivered. 11:33 Speaker A A corps-level planning cycle compressed from 72 hours to one. 11:48 Speaker A Not an estimate or a target. 12:00 Speaker A A reported result from a live exercise with the Allied Rapid Reaction Corps. 12:15 Speaker A Alongside that, the number of targets a corps could process in a day increased by roughly tenfold. 12:25 Speaker A The British Army described the data throughput as approximately 10 terabytes a day, which they compared to nearly three months of continuous high-definition streaming. 12:41 Speaker A That last comparison is theirs, not 12:49 Speaker A A British-operated digital architecture that European allies can plug into, particularly one that does not route through Washington, is a form of strategic relevance that does not depend on how many frigates the Royal Navy can field in any given month. 13:04 Speaker A And then there is the communications question, which is where the enthusiasm needs to be calibrated. 13:10 Speaker A Asgard's data flows are only as reliable as the links that carry them. At brigade level in Estonia, Asgard used commercially procured radios supplied under the same rapid acquisition approach that built the rest of the system. 13:24 Speaker A Those commercial solutions are capable, but the British Army's primary tactical communication system remains Bowman, a General Dynamics program delivered between 1904 and 2008. 13:36 Speaker A Analysts who have examined Asgard in detail describe Bowman's bandwidth as a single-lane road. What Asgard at core scale actually requires is multiple five-lane motorways. 13:48 Speaker A The commercial radio solves some of that at brigade level. Whether they solve it across a heterogeneous allied core with equipment from a dozen nations in an environment where Russia's electronic warfare capability is specifically designed to degrade exactly these kinds of 14:02 Speaker A network-dependent systems has not been publicly tested. Russian forces in Ukraine jam communications as a first-order tactic, and they are reportedly able to jam their own radios intentionally to deny adversary exploitation of their frequencies. 14:18 Speaker A An AI targeting network that cannot function when its communications are degraded is, in that specific scenario, a very expensive planning tool. 14:27 Speaker A The MOD has not publicly addressed this directly. The Asgard team almost certainly has thoughts on it. 14:34 Speaker A Whether those thoughts are adequate is a question the next phase of testing will need to answer honestly. 14:40 Speaker A Here is where the legitimately insane headline meets the harder arithmetic. Asgard can process 10 times as many targets per day at core level. The Army has documented that. 14:52 Speaker A What the Army has not documented publicly is where the munitions come from to service that list. 14:59 Speaker A Calibre Defensive's analysis of the Estonia exercise, published in July 2025 by defense journalist Sam Cranny-Evans, made this point with uncomfortable directness. 15:09 Speaker A Project Asgard has reduced the time needed to go from finding a Russian vehicle to targeting it, but it has not addressed how the Army will actually strike those vehicles. 15:19 Speaker A An AI targeting system that generates a queue of 200 prioritized targets per day is enormously valuable precisely to the extent that you have 200 things to fire at them. If you have 40, you have a faster way of deciding which 40. That is 15:35 Speaker A useful. It is not the same thing [music] as lethality doubled. The British Army's available strike inventory for a high-intensity land confrontation is not large. The bulk of its indirect fire lethality sits in a relatively small number of M270 multiple 15:51 Speaker A launch rocket systems and a handful of Swedish Archer self-propelled howitzers. For one-way effects, the Helsing HX2 loitering munition used in Asgard exercises is operationally proven in Ukraine and genuinely capable, but specific quantities for British Army service have not been publicly 16:10 Speaker A confirmed. Anduril's Altius M strike drone has been supplied to Ukraine by the UK government and features in Asgard's [music] capability palette, again, without a stated British inventory number. 16:23 Speaker A There is also the Dart one-way effector from Modine, procured directly under Asgard with a published range of 250 km. 16:31 Speaker A That range places Dart in what the Army calls the deep fight, engaging enemy headquarters, logistics nodes, and enablers well behind the front line. 16:41 Speaker A That is a coherent operational concept. What it does not provide is the volume of tactical strike effects that a brigade in close combat actually needs against infantry, light armor, and the waves of cheap drones that constitute a peer adversary's assault at scale. 16:58 Speaker A Ukraine taught the world that tactical strike requires expendables in very large numbers, not a limited stock of expensive precision weapons held back for high-value targets. 17:09 Speaker A The defense investment plan published on the 30th of June 2026 acknowledges this directly. 6 billion pounds allocated to munitions replenishment across all three services, six new ammunition factories, and 1 and 1/2 billion for what the plan calls an always-on pipeline. These are 17:28 Speaker A announced commitments. The factories are not yet built. The stockpiles are not yet filled. The same DIP included a striking admission in its own language. 47 of 49 major defense projects were delayed or overbudget on inheritances. 17:46 Speaker A The government frames this as a problem it received rather than created, and there is reasonable merit to that across a long procurement history. But, it is the baseline against which Asgard's novel rapid acquisition approach must eventually be measured. The prototype 18:01 Speaker A was fast. The scaling, with its harder industrial requirements and larger budgets, will be the real test. The financial architecture around the digital targeting web is substantial by any measure. The initial £1bn DTW commitment came through the 2025 Strategic Defense Review. The June 2026 18:23 Speaker A Defense Investment Plan added a further 1.8 billion specifically for the targeting web. Across the full DIP period, the digital targeting web and backbone together [music] received £8.6bn. 18:37 Speaker A Between 2030 and 2035, [music] the commitment rises to at least 17 billion. That tells you clearly how central the MOD believes this architecture is to Britain's future military credibility. 18:50 Speaker A It also [music] spans at least two parliaments. Depending on governments not yet elected maintaining investment decisions made [music] in 2026. 18:59 Speaker A £17bn for a digital targeting system by 2035 is a plan. Whether it becomes a delivered capability is a question the MOD's own inherited track record gives us reason to hold open. 19:14 Speaker A The sovereignty question running through all of this is one worth sitting with. [music] The two components most responsible for Asgard's headline performance are Ultra, built by Helsing, a German company headquartered in Munich, and the Lattice, built by Anduril, an American 19:30 Speaker A firm operating in the UK through a subsidiary. Neither is a British sovereign capability. That matters in two specific ways. 19:39 Speaker A The first is industrial. If Britain's most critical targeting software is German and American, the British defense industry does not own the intellectual property central to battlefield performance and continued access depends on the commercial and political decisions [music] of foreign companies. 19:56 Speaker A The second is operational. The United States has applied export controls and technology restrictions to allies in ways that have occasionally surprised those allies. 20:06 Speaker A A British core headquarters whose AI targeting layer runs on American software is, in a narrow but real set of stress scenarios, a dependency. [music] The Avengers training data compounds this. 20:19 Speaker A Ukraine's Avengers AI labs is Ukrainian property, generated by Ukrainian sensors, annotated by Ukrainian analysts, and made available under an agreement that could be modified. 20:30 Speaker A Britain is the first international partner admitted to that platform, and the arrangement is genuinely valuable. 20:37 Speaker A But a British AI targeting advantage partially built on German software, American networking architecture, and Ukrainian battlefield experience is a complicated answer to the question of what sovereign military AI capability actually means in 2026. 20:53 Speaker A Then there is the oversight question, and this is the one attracting the least comfortable public discussion. 21:00 Speaker A The legal reviewer is formally present at the PRISM stage. The human is in the loop by design. 21:07 Speaker A But Asgard is explicitly built to compress targeting decisions to seconds and minutes, and the queue arriving at that reviewer during a core level engagement does not contain one item. 21:18 Speaker A Human Rights Watch in June 2026 identified automation bias as a documented risk in exactly this kind of system. The tendency for human reviewers working under time pressure to accept AI recommendations without the scrutiny those [music] recommendations would receive if the clock were not running. 21:37 Speaker A That is not evidence that Asgard has caused civilian casualties. It is a legitimate structural question about whether a system designed for speed is compatible with the quality of human review that meaningful oversight actually requires. 21:52 Speaker A The MOD has not provided a public answer to that question. Given the sums being committed and the decisions the system is designed to support, it probably should. 22:02 Speaker A Return to the opening question. What is actually British about Britain's AI targeting technology? The answer is more layered than either a proud affirmation or a deflating dismissal. 22:15 Speaker A Ultra is German. Lattice is American. The training data Britain just secured came from Ukrainian battlefields. The rapid procurement methodology was borrowed from the emergency process the Ministry of Defense developed to supply Ukraine. 22:30 Speaker A And yet the army that integrated all of this, exercised it under NATO command in Estonia and London underground facility, and connected a core headquarters to formations in two countries simultaneously, is British. 22:44 Speaker A The doctrine, the operational design, the decision-making chain, the soldiers and officers who made it work, and the command authority that approves every strike, those are British. Whether that constitutes sovereign military AI capability in a meaningful industrial or strategic sense is a question October 23:04 Speaker A 2026 has not answered. And it is worth being honest about that gap [music] rather than closing it with enthusiasm. 23:12 Speaker A The NATO architecture ambition is genuinely the most consequential thing about this program, and it has received far less public attention than the planning cycle figures. 23:22 Speaker A Britain's world made the strategic argument clearly. A country that provides the digital layer through which allied forces combine their capabilities, shapes interoperability standards, integration habits, and the tempo at which those allied forces operate. 23:38 Speaker A Asgard connected a core headquarters in London to formations in Estonia across two simultaneous exercises. 23:45 Speaker A The British Army's own account describes it as interoperable [music] by design, built to bring different nations and languages into a common operational environment. 23:55 Speaker A If that holds at scale, Britain is not merely fielding a faster national headquarters. It is positioning itself as the nation that provides the digital connective tissue of NATO's European land deterrents. 24:07 Speaker A European allies have genuine concerns about dependence on American military technology. A British capability that is interoperable with NATO standards and demonstrably functional when the transatlantic link is under pressure offers something neither Washington nor Brussels currently provides. 24:24 Speaker A Whether Asgard is designed to be that strategic hedge, or whether allied connectivity is incidental to a national accelerator, is a decision H and M government has not stated publicly. The test that would validate that alliance claim has not happened. 24:39 Speaker A The exercises completed so far have been cooperative, pre-planned, and conducted with willing partners in conditions designed to demonstrate capability rather than stress test its limits. 24:50 Speaker A The decisive question is harder. Can an allied unit join the Asgard environment quickly? Can that unit's national command authority survive translation into the shared workflow without being silently overridden? And can the network operate with meaningful national control intact when 25:07 Speaker A communications are degraded or when British and allied policy diverges on a specific engagement? Those are not purely technical questions. They are legal and political ones, and they will determine whether European allies actually choose to rely on British digital infrastructure in a 25:24 Speaker A crisis or observe it admiringly from a careful distance. Here is the reinterpretation I have been building toward. 25:32 Speaker A The 72 to 1 planning cycle figure is extraordinary, and it deserves the attention it has received. But the most strategically important thing Asgard has demonstrated is not that number. It is that the British Ministry of Defense can procure critical operational military 25:48 Speaker A software in 7 months, contract 26 companies through a single open framework, deploy a working prototype in a live NATO exercise, and scale to core level within a year. 26:01 Speaker A If that procurement model is genuinely repeatable and transferable beyond this one program, it addresses something that has arguably been the central failure of British defense investment across three decades. 26:13 Speaker A The MOD inherited 47 of 49 major programs delayed or over budget. Asgard used a deliberately different approach and produced a different result. An institution that can do that consistently is worth more in strategic terms than any single targeting 26:31 Speaker A capability, however impressive. Whether that model survives the scaling is the honest uncertainty. The Briganti prototype worked beautifully, partly because it was small, focused, and run by a team with unusual authority to move quickly. 26:46 Speaker A A program eventually absorbing 8.6 billion pounds in one planning period and 17 billion in the following five years is no longer a nimble experiment. 26:57 Speaker A It is a major program with everything that implies. Governance layers, contractor management at scale, parliamentary scrutiny, and the institutional gravity that tends to pull novel methods back toward the patterns that produced the 47 failures. 27:13 Speaker A The MOD's own investment plan was candid about the inheritance it received. Whether Asgard represents a genuine change in how Britain acquires capability or a brilliant exception the system will eventually normalize back into mediocrity is what the next five 27:28 Speaker A years will establish. So, is it legitimately insane? The planning cycle compression. Yes, documented remarkable directly attributed to a named senior officer at a named public event. 27:42 Speaker A The procurement timeline. Yes, seven months from announcement to Estonia is genuinely extraordinary for any government program. The fully sanity claim. Not yet, because the munitions pipeline is still being built. The communication vulnerability under live Russian electronic warfare remains 28:01 Speaker A publicly unaddressed. The Ukrainian training data has only just arrived, and the human oversight model at operational speed carries a structural tension without a public resolution. 28:12 Speaker A Asgard is a genuine proof of concept for an approach Britain urgently needs. It is not yet a war-winning capability. The distance between those two points is what the program must close, and the funding exists to attempt it if the 28:27 Speaker A delivery follows the prototype's pace rather than the inheritance's. The question worth sitting with is this. 28:34 Speaker A When a legal reviewer in the process, what does meaningful human oversight actually look like? Not the formal presence of a reviewer in the process, but the planning cycle compression. Yes, documented remarkable directly attributed to a named senior officer at 28:44 Speaker A a named public event. The procurement timeline. Yes, not the formal presence a reviewer in the process, but the genuine quality of judgment that accountability for a lethal strike requires. 28:55 Speaker A If Britain has built a system designed to move faster than the human judgment it nominally keeps in the loop can reliably follow, that is the conversation defense policy needs to have openly before the system reaches full operational status rather than 29:09 Speaker A after. If the digital targeting web eventually becomes alliance infrastructure, it is a conversation NATO needs to have as well. 29:18 Speaker A Do you think the current oversight model is adequate? Or is human accountability the one thing in this architecture that still needs serious public scrutiny? 29:27 Speaker A Subscribe if you want to follow where this goes, because how the digital targeting web connects to the air domain and to GCAP is a story that has barely started. 29:37 Speaker A [music] Topics: Project Asgard British Army AI targeting military AI Helsing Anduril NATO defence procurement Ukraine lessons human oversight Study this video - [Flashcards from this video](https://sozai.app/tools/ai-flashcard-generator/?from=transcript&id=86565) - [Quiz on this video](https://sozai.app/tools/quiz-generator/?from=transcript&id=86565) Free, made by AI from this transcript. No signup. --- This is the markdown twin of https://sozai.app/transcript/britain-new-ai-targeting-technology/ — the same content, without the markup. Published by SozAI (https://sozai.app). Reuse and quotation are allowed with attribution and a link back. Machine-readable index: https://sozai.app/llms.txt · data API: https://sozai.app/api/