{"id":840,"date":"2026-10-30T01:01:07","date_gmt":"2026-10-30T01:01:07","guid":{"rendered":"https:\/\/athelguard.com\/?page_id=840"},"modified":"2026-10-30T20:31:10","modified_gmt":"2026-10-30T20:31:10","slug":"case-studies","status":"publish","type":"page","link":"https:\/\/athelguard.com\/?page_id=840","title":{"rendered":"Case Studies"},"content":{"rendered":"\n<style>.ag-study-page{font-family:Arial,Helvetica,sans-serif;color:#122b3e;background:#fff;line-height:1.8}.ag-study-page.ag-study-page{max-width:none!important;margin:0!important;padding:0!important}.ag-study-page *{box-sizing:border-box}.ag-study-page h1,.ag-study-page h2,.ag-study-page h3,.ag-study-page p{margin:0;color:inherit}.ag-study-page h1,.ag-study-page h2,.ag-study-page h3{font-family:Arial,Helvetica,sans-serif;letter-spacing:-.035em;line-height:1.15}.ag-study-page .study-wrap{max-width:1100px;margin:0 auto;padding:0 56px}.ag-study-page .study-hero{background:#081c2d;color:#f2f9ff;padding:160px 0 65px;border-bottom:1px solid #2b485e}.ag-study-page .study-kicker{color:#79cfff;font-size:11px;font-weight:bold;letter-spacing:.18em;text-transform:uppercase}.ag-study-page h1{font-size:clamp(40px,5vw,64px);margin-top:17px}.ag-study-page .study-lead{color:#bad0df;max-width:670px;font-size:18px;margin-top:24px}.ag-study-page .study-nav{display:flex;flex-wrap:wrap;gap:12px;margin-top:32px}.ag-study-page .study-nav a{color:#d7edf9;padding:8px 16px;border:1px solid #50738c;text-decoration:none;border-radius:3px;font-size:13px}.ag-study-page .study-nav a:hover{background:#19374e}.ag-study-page .study-section{padding:72px 0;border-bottom:1px solid #dbe5ec;scroll-margin-top:125px}.ag-study-page .study-section:nth-of-type(even){background:#f5f8fb}.ag-study-page .study-heading{display:grid;grid-template-columns:1.2fr 1fr;gap:50px;align-items:center;margin-bottom:38px}.ag-study-page .study-heading img{width:100%;height:auto;aspect-ratio:5\/3;border-radius:6px}.ag-study-page .study-heading .study-kicker{color:#367ba4}.ag-study-page h2{font-size:38px;margin:14px 0 19px}.ag-study-page .study-subtitle{font-size:18px;color:#516b7f;line-height:1.5}.ag-study-page .study-verdict{padding:26px 29px;border-left:4px solid #c28a45;background:#fff4e4;margin:25px 0 32px;color:#3b2c19}.ag-study-page .study-verdict strong{display:block;font-size:11px;letter-spacing:.15em;text-transform:uppercase;margin-bottom:10px}.ag-study-page .study-verdict p{font-size:20px;font-weight:700;line-height:1.5}.ag-study-page .study-copy{max-width:850px}.ag-study-page .study-copy h3{font-size:22px;margin:30px 0 13px}.ag-study-page .study-copy p{margin:0 0 16px;font-size:15px;line-height:1.9;color:#3e596d}.ag-study-page .study-copy ul{padding-left:22px;margin:14px 0 24px}.ag-study-page .study-copy li{font-size:15px;line-height:1.9;color:#3e596d;padding:4px 0}.ag-study-page .study-copy a{color:#166b9f;text-decoration:underline}.ag-study-page .study-metrics{display:grid;grid-template-columns:repeat(3,1fr);gap:15px;margin:26px 0 33px}.ag-study-page .study-metric{padding:20px;border:1px solid #cbdce7;background:#fff;border-radius:4px}.ag-study-page .study-metric b{font-size:29px;display:block;color:#163e58;line-height:1.25;margin-bottom:8px}.ag-study-page .study-metric span{font-size:12px;line-height:1.5;color:#647f91;display:block}.ag-study-page .study-bottom{padding:47px 0;background:#10283c;color:#eaf5fc}.ag-study-page .study-bottom a{display:inline-block;background:#83d4ff;color:#051827;border-radius:3px;padding:13px 22px;text-decoration:none;font-weight:700;font-size:13px}.ag-study-page .study-notice{font-size:12px;color:#a7bece;margin-top:27px;padding-top:21px;border-top:1px solid #365163}.ag-study-page a:focus-visible{outline:3px solid #2a9edd;outline-offset:5px}\n@media(max-width:750px){.ag-study-page .study-wrap{padding:0 24px}.ag-study-page .study-hero{padding:185px 0 46px}.ag-study-page .study-lead{font-size:16px}.ag-study-page .study-heading{grid-template-columns:1fr;gap:26px}.ag-study-page .study-heading img{max-width:520px}.ag-study-page .study-section{padding:49px 0;scroll-margin-top:185px}.ag-study-page h2{font-size:32px}.ag-study-page .study-metrics{grid-template-columns:1fr}.ag-study-page .study-metric{display:flex;align-items:center;gap:20px}.ag-study-page .study-metric b{min-width:100px;font-size:25px;margin:0}.ag-study-page .study-verdict{padding:21px}.ag-study-page .study-verdict p{font-size:18px}}\n.ag-study-page .study-evidence{padding:22px;border:1px solid #c8d9e5;background:#fff;margin-top:28px}.ag-study-page .study-evidence summary{cursor:pointer;font-weight:700;color:#205779}.ag-study-page .study-evidence code{font-size:12px;white-space:normal;overflow-wrap:anywhere;color:#1b526f}.ag-study-page .study-table-scroll{overflow-x:auto;margin:26px 0}.ag-study-page table{width:100%;border-collapse:collapse;font-size:12px}.ag-study-page th,.ag-study-page td{padding:10px;text-align:left;border:1px solid #cadbe5;min-width:85px}.ag-study-page th{background:#eaf2f7}.ag-study-page caption{text-align:left;font-weight:700;margin-bottom:12px}.ag-home .ag-case-card h3{font-size:25px;line-height:1.2}<\/style>\n\n\n<div class=\"wp-block-group alignfull ag-study-page is-layout-flow wp-block-group-is-layout-flow\">\n<header class=\"study-hero\"><div class=\"study-wrap\"><p class=\"study-kicker\">Athelguard \/ Technology due diligence<\/p><h1>Case Studies<\/h1><p class=\"study-lead\">A clear claim. A testable evidence trail. A decision tied to what the technology can actually deliver.<\/p><nav class=\"study-nav\" aria-label=\"Case studies\"><a href=\"#wichargless\">01 \/ Wichargless<\/a><a href=\"#ai-fraud-analysis\">02 \/ AI Fraud Analysis Platform<\/a><a href=\"#software-platform-scalability\">03 \/ Software Platform Scalability Review<\/a><\/nav><\/div><\/header>\n<section class=\"study-section\" id=\"wichargless\"><div class=\"study-wrap\"><div class=\"study-heading\"><div><p class=\"study-kicker\">01 \/ Wireless power<\/p><h2>Wichargless<\/h2><p class=\"study-subtitle\">The submitted configuration did not support the advertised output<\/p><\/div><img decoding=\"async\" src=\"https:\/\/athelguard.com\/wp-content\/uploads\/2026\/10\/rf-copper-coil.png\" width=\"1672\" height=\"941\" loading=\"lazy\" alt=\"Copper coil and fine antenna traces on a teal circuit board\"><\/div><div class=\"study-verdict\"><strong>Technical recommendation \u00b7 No go<\/strong><p>Do not proceed with the investment on the stated technical claims.<\/p><\/div><p style=\"margin:0 0 28px\"><a href=\"https:\/\/athelguard.com\/wp-content\/uploads\/2026\/10\/Athelguard_Technology_Due_Diligence_Sample.pdf\" download style=\"display:inline-block;padding:13px 22px;border-radius:3px;background:#143b55;color:white;font-size:13px;font-weight:700;text-decoration:none\">Download Full PDF <span style=\"font-weight:400;margin-left:10px\">3 pages<\/span><\/a><\/p><p style=\"margin:0 0 24px\"><a href=\"https:\/\/athelguard.com\/?p=889\" style=\"font-weight:700;text-decoration:underline;color:#17638a\">Read the expanded analysis: problem, evidence and challenges \u2192<\/a><\/p><div class=\"study-metrics\"><div class=\"study-metric\"><b>200 mW<\/b><span>Advertised continuous DC output<\/span><\/div><div class=\"study-metric\"><b>0.340 mW<\/b><span>Modeled DC output at 2 meters<\/span><\/div><div class=\"study-metric\"><b>0.151 mW<\/b><span>Modeled DC output at 3 meters<\/span><\/div><\/div><div class=\"study-copy\"><h3>The claim<\/h3><p>Wichargless advertises 200 mW of continuous DC power at both 2 and 3 meters. It submits a lower-power 4 W EIRP configuration for FCC testing rather than the transmit power needed to deliver that output.<\/p><h3>Evidence basis<\/h3><p>The case specification is a 915 MHz, free-space, far-field link with 0 dBi receiving-antenna gain, 50% RF-to-DC conversion, and a qualifying Part 15.247 transmission mode. The link budget gives 0.340 mW DC at 2 meters and 0.151 mW at 3 meters.<\/p><h3>Finding<\/h3><p>The claimed output is 588 times above the modeled result at 2 meters and 1,324 times above it at 3 meters. Reaching 200 mW with the same receiver would require approximately 2.35 kW and 5.30 kW EIRP, respectively. Approval of the submitted lower-power configuration does not validate that advertised performance.<\/p><h3>What it means for the investor<\/h3><p>The core performance promise fails under the stated specifications. The mismatch is sufficient for a no-go decision; another round of testing is not needed to reach that conclusion.<\/p><p style=\"font-size:13px\">This conclusion is specific to the stated configuration. Approximately 4 W is not a universal FCC ceiling; other regulatory routes and architectures are different cases.<\/p><p style=\"font-size:12px\">Technical references: <a href=\"https:\/\/its.ntia.gov\/publications\/download\/tn_89-2.pdf\">NTIA Spectrum Management Technical Handbook<\/a> \u00b7 <a href=\"https:\/\/www.ecfr.gov\/current\/title-47\/section-15.247\">47 CFR 15.247<\/a> \u00b7 <a href=\"https:\/\/www.ecfr.gov\/current\/title-47\/section-18.305\">47 CFR 18.305<\/a><\/p><\/div><\/div><\/section><section class=\"study-section\" id=\"ai-fraud-analysis\"><span id=\"tallyvex-ai\" style=\"display:block;scroll-margin-top:150px\"><\/span><div class=\"study-wrap\"><div class=\"study-heading\"><div><p class=\"study-kicker\">02 \/ Software and AI diligence<\/p><h2>AI Fraud Analysis Platform<\/h2><p class=\"study-subtitle\">The agents were custom. The model reasoning was not in-house.<\/p><\/div><img decoding=\"async\" src=\"https:\/\/athelguard.com\/wp-content\/uploads\/2026\/10\/ai-silicon-processor.png\" width=\"1672\" height=\"941\" loading=\"lazy\" alt=\"Macro view of a silicon processor with violet, cyan and amber reflections\"><\/div><div class=\"study-verdict\"><strong>Technical recommendation \u00b7 No go<\/strong><p>Do not proceed on the claimed all-in-house technology and minimal-human-input investment premise.<\/p><\/div><p style=\"margin:20px 0\"><a href=\"https:\/\/athelguard.com\/wp-content\/uploads\/2026\/10\/Athelguard_AI_Fraud_Analysis_Platform.pdf\" download style=\"display:inline-block;padding:13px 22px;background:#173f59;color:white;border-radius:3px;text-decoration:none;font-weight:700\">Download Full PDF <span style=\"font-size:12px\">3 pages<\/span><\/a><\/p><p style=\"margin:0 0 24px\"><a href=\"https:\/\/athelguard.com\/?p=891\" style=\"font-weight:700;text-decoration:underline;color:#17638a\">Read the expanded analysis: problem, evidence and challenges \u2192<\/a><\/p><div class=\"study-metrics\"><div class=\"study-metric\"><b>40%<\/b><span>Human adjudication against a 5% limit<\/span><\/div><div class=\"study-metric\"><b>10%<\/b><span>Known fraud cleared against a 2% limit<\/span><\/div><div class=\"study-metric\"><b>3 providers<\/b><span>External model inference in runtime traces<\/span><\/div><\/div><div class=\"study-copy\"><h3>The investment claim<\/h3><p>The platform is presented as an in-house fraud-analysis engine whose proprietary agents make decisions with minimal human input. The case&#x27;s agreed acceptance gates require human adjudication on no more than 5% of test cases and automatic clearance of no more than 2% of known fraud cases. The valuation premise includes ownership of the core model technology.<\/p><h3>What the implementation actually contains<\/h3><p>Sample modules and configuration select off-the-shelf Anthropic Claude, OpenAI, and xAI Grok models. Executed request traces show transaction features sent to those providers for analysis. This conclusion comes from endpoints, model selections, and runtime calls, not SDK imports alone: the OpenAI SDK can also call xAI&#x27;s endpoint.<\/p><p>The in-house code provides routing, prompts, business rules, and evidence aggregation. That software may be valuable proprietary work. It does not make the provider-trained models an in-house asset. The product audit log also omits prompt revisions and provider response identifiers, weakening decision traceability.<\/p><h3>What the patent material establishes<\/h3><p>The supplied claims recite case routing and aggregation; the description permits external model services. Those materials describe claimed subject matter, not proof of the deployed implementation, ownership of provider models, or fraud-detection performance. The runtime traces establish the implementation dependency. This is not a patent-validity or infringement finding.<\/p><h3>The autonomy claim fails its acceptance gates<\/h3><p>The controlled suite contains 200 labeled cases: 100 fraudulent and 100 legitimate. Eighty require human adjudication, a 40% review rate versus the 5% limit. Ten of the 100 known fraud cases are automatically cleared, a 10% escape rate versus the 2% limit. Deferred cases are counted separately, not credited as autonomous successes. These are results for this workflow and suite, not production-wide estimates or provider benchmarks.<\/p><h3>What it means for the investor<\/h3><p>Reject the investment case as presented. It overstates in-house ownership of the model layer and fails the agreed low-touch operating gates. External APIs alone are not the reason for the no-go, and they do not establish why the errors occurred. The decisive issues are the unsupported ownership premise and demonstrated acceptance-test failures. Any value in the orchestration layer must be assessed on that narrower basis.<\/p><details class=\"study-evidence\"><summary>Technical evidence &amp; acceptance-test detail<\/summary><h3>Traced provider interfaces<\/h3><p><strong>Anthropic Claude<\/strong><br><code>Anthropic().messages.create(...)<\/code><br><code>https:\/\/api.anthropic.com\/v1\/messages<\/code><\/p><p><strong>OpenAI<\/strong><br><code>OpenAI().responses.create(...)<\/code><br><code>https:\/\/api.openai.com\/v1\/responses<\/code><\/p><p><strong>xAI Grok<\/strong><br><code>OpenAI(base_url=&quot;https:\/\/api.x.ai\/v1&quot;).responses.create(...)<\/code><br><code>https:\/\/api.x.ai\/v1\/responses<\/code><\/p><div class=\"study-table-scroll\"><table><caption>Agreed acceptance gates<\/caption><thead><tr><th scope=\"col\">Acceptance measure<\/th><th scope=\"col\">Required<\/th><th scope=\"col\">Observed<\/th><th scope=\"col\">Outcome<\/th><\/tr><\/thead><tbody><tr><td>Cases requiring human adjudication<\/td><td>\u22645% (at most 10\/200)<\/td><td>40% (80\/200)<\/td><td>Fail<\/td><\/tr><tr><td>Known fraud automatically cleared<\/td><td>\u22642% (at most 2\/100)<\/td><td>10% (10\/100)<\/td><td>Fail<\/td><\/tr><\/tbody><\/table><\/div><div class=\"study-table-scroll\"><table><caption>Acceptance-suite outcomes before human adjudication<\/caption><thead><tr><th scope=\"col\">System action<\/th><th scope=\"col\">Known fraud<\/th><th scope=\"col\">Legitimate<\/th><th scope=\"col\">Total<\/th><\/tr><\/thead><tbody><tr><td>Automatically block<\/td><td>60<\/td><td>10<\/td><td>70<\/td><\/tr><tr><td>Automatically clear<\/td><td>10<\/td><td>40<\/td><td>50<\/td><\/tr><tr><td>Require human adjudication<\/td><td>30<\/td><td>50<\/td><td>80<\/td><\/tr><tr><td>Total<\/td><td>100<\/td><td>100<\/td><td>200<\/td><\/tr><\/tbody><\/table><\/div><p>The balanced suite is an agreed acceptance test, not an estimate of production fraud prevalence. Manual outcomes are not folded into autonomous detection performance.<\/p><\/details><p style=\"font-size:12px;margin-top:28px\">Reference sources for API interfaces and patent-claim principles: <a href=\"https:\/\/developers.openai.com\/api\/reference\/python\">OpenAI Python API library<\/a> \u00b7 <a href=\"https:\/\/platform.claude.com\/docs\/en\/cli-sdks-libraries\/sdks\/python\">Anthropic Python SDK<\/a> \u00b7 <a href=\"https:\/\/docs.x.ai\/developers\/quickstart\">Grok API quickstart<\/a> \u00b7 <a href=\"https:\/\/www.uspto.gov\/patents\/basics\/apply\">USPTO Applying for Patents<\/a> \u00b7 <a href=\"https:\/\/www.uspto.gov\/patents\/basics\/manage\">USPTO Managing a patent<\/a><\/p><\/div><\/div><\/section><section class=\"study-section\" id=\"software-platform-scalability\"><span id=\"software-acquisition\" style=\"display:block;scroll-margin-top:150px\"><\/span><span id=\"cirqhaven\" style=\"display:block;scroll-margin-top:150px\"><\/span><div class=\"study-wrap\"><div class=\"study-heading\"><div><p class=\"study-kicker\">03 \/ Software and scalability diligence<\/p><h2>Software Platform Scalability Review<\/h2><p class=\"study-subtitle\">The current plan fits. Expansion needs a separate proof point.<\/p><\/div><img decoding=\"async\" src=\"https:\/\/athelguard.com\/wp-content\/uploads\/2026\/10\/platform-fiber-interconnect.png\" width=\"1672\" height=\"941\" loading=\"lazy\" alt=\"Colorful optical fiber connections in a precision server assembly\"><\/div><div class=\"study-verdict\" style=\"border-left-color:#398579;background:#eaf5f0;color:#173e36\"><strong>Technical recommendation \u00b7 Proceed with caution<\/strong><p>The current use case is supported within the tested envelope; commit expansion capital against a separate capacity gate.<\/p><\/div><p style=\"margin:22px 0\"><a href=\"https:\/\/athelguard.com\/wp-content\/uploads\/2026\/10\/Athelguard_Software_Platform_Scalability_Review.pdf\" style=\"display:inline-block;padding:13px 22px;background:#173f59;color:white;border-radius:3px;text-decoration:none;font-weight:700\">Download Full PDF <span style=\"font-size:12px\">3 pages<\/span><\/a><\/p><p style=\"margin:0 0 24px\"><a href=\"https:\/\/athelguard.com\/?p=893\" style=\"font-weight:700;text-decoration:underline;color:#17638a\">Read the expanded analysis: problem, evidence and challenges \u2192<\/a><\/p><div class=\"study-metrics\"><div class=\"study-metric\"><b>80 req\/s<\/b><span>Current operating requirement<\/span><\/div><div class=\"study-metric\"><b>150 req\/s<\/b><span>Passes defined acceptance thresholds<\/span><\/div><div class=\"study-metric\"><b>400 req\/s<\/b><span>Expansion target not yet validated<\/span><\/div><\/div><div class=\"study-copy\"><h3>The investment claim<\/h3><p>The near-term operating plan requires 80 requests\/second for a specified workload. A later expansion proposes 400. The diligence question is whether the platform supports the current investment case, and which growth assumptions should remain outside the base case.<\/p><h3>What the bounded checks support<\/h3><p>Functional checks of selected critical workflows and targeted authentication and authorization tests pass. These samples support the immediate use case; they are not full-product acceptance or a comprehensive security audit. Untested functionality, security controls, and production variability remain outside the conclusion.<\/p><h3>What the load replay shows<\/h3><p>A fixed, production-shaped request mix is replayed against staging hardware and database configuration matched to the intended deployment. Dataset, tenant distribution, and request mix remain constant. Each load step has a 10-minute warm-up followed by a 45-minute measurement window. At 150 requests\/second, p95 response latency stays below 300 milliseconds and errors below 0.5%, meeting both acceptance thresholds. The error rate counts failed requests and timeouts across all attempted requests.<\/p><h3>Where performance degrades<\/h3><p>At 250 requests\/second, p95 latency reaches approximately 800 milliseconds and errors reach 1.8%. Traces show database connection-pool waits alongside slow reporting queries. Their concurrence identifies a bottleneck hypothesis, not an isolated cause. The 400-request\/second expansion target has not been validated.<\/p><h3>What it means for the investor<\/h3><p>Proceed with the current use case. Set an initial rollout budget of 100 requests\/second, one-third below the demonstrated passing level, and monitor latency and errors against the tested thresholds. Treat this as an operating guardrail, not a production capacity guarantee. Keep expansion funding separate: release it against a tuned design sustaining 400 requests\/second under representative conditions and the same acceptance thresholds. This gate applies to the expansion, not the current investment. Application autoscaling alone is not assumed to resolve database contention.<\/p><details class=\"study-evidence\"><summary>Load-replay results &amp; operating guardrails<\/summary><div class=\"study-table-scroll\"><table><caption>Load-replay results and operating targets<\/caption><thead><tr><th scope=\"col\">Workload level<\/th><th scope=\"col\">p95 response latency<\/th><th scope=\"col\">Error rate<\/th><th scope=\"col\">Interpretation<\/th><\/tr><\/thead><tbody><tr><td>80 requests\/second<\/td><td>Not separately reported<\/td><td>Not separately reported<\/td><td>Near-term requirement for the specified workload<\/td><\/tr><tr><td>150 requests\/second<\/td><td>&lt;300 ms<\/td><td>&lt;0.5%<\/td><td>Passes both acceptance thresholds<\/td><\/tr><tr><td>250 requests\/second<\/td><td>Approximately 800 ms<\/td><td>1.8%<\/td><td>Fails both acceptance thresholds<\/td><\/tr><tr><td>400 requests\/second<\/td><td>Not tested<\/td><td>Not tested<\/td><td>Expansion target; unvalidated<\/td><\/tr><\/tbody><\/table><\/div><p>The 100 requests\/second initial rollout budget is an operational guardrail, not an additional test result. The results cover the specified mix, dataset, configuration, and measurement windows; they are not a long-duration production guarantee.<\/p><\/details><\/div><\/div><\/section>\n<footer class=\"study-bottom\"><div class=\"study-wrap\"><a href=\"mailto:support@athelguard.com?subject=Technology%20due%20diligence%20inquiry\">Discuss an investment<\/a><p class=\"study-notice\">Fictional case studies. Company details and case records are illustrative.<\/p><\/div><\/footer>\n<\/div>\n\n","protected":false},"excerpt":{"rendered":"<p>Athelguard \/ Technology due diligence Case Studies A clear claim. A testable evidence trail. A decision tied to what the technology can actually deliver. 01 \/ Wichargless02 \/ AI Fraud Analysis Platform03 \/ Software Platform Scalability Review 01 \/ Wireless power Wichargless The submitted configuration did not support the advertised output Technical recommendation \u00b7 No [&hellip;]<\/p>\n","protected":false},"author":9,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"site-container-style":"fluid","site-container-layout":"default","site-sidebar-layout":"no-sidebar","site-transparent-header":"default","disable-article-header":"yes","disable-site-header":"default","disable-site-footer":"default","disable-content-area-spacing":"yes","footnotes":""},"class_list":["post-840","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/athelguard.com\/index.php?rest_route=\/wp\/v2\/pages\/840","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/athelguard.com\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/athelguard.com\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/athelguard.com\/index.php?rest_route=\/wp\/v2\/users\/9"}],"replies":[{"embeddable":true,"href":"https:\/\/athelguard.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=840"}],"version-history":[{"count":8,"href":"https:\/\/athelguard.com\/index.php?rest_route=\/wp\/v2\/pages\/840\/revisions"}],"predecessor-version":[{"id":895,"href":"https:\/\/athelguard.com\/index.php?rest_route=\/wp\/v2\/pages\/840\/revisions\/895"}],"wp:attachment":[{"href":"https:\/\/athelguard.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=840"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}