{"slug": "show-hn-aletheionagi-grounding-enforcement-for-ai-agents", "title": "Show HN: AletheionAGI – Grounding enforcement for AI agents", "summary": "AletheionAGI, a new grounding enforcement layer for AI agents, claims to block unsupported claims before they reach customers, reporting 93.6% Recall@5 and a 66.5% diagnostic answer score on 979 frozen support questions, with 18/18 deterministic controls passed and 45/45 unsafe deliveries contained. The product, priced at R$ 99 for 1,000 queries with no subscription, integrates with existing readers via BYOK and enforces memory retrieval, policy authorization, and evidence validation.", "body_md": "**100%**\n\n**100%**\n\nGrounded memory for production AI\n\nGive your existing AI a memory it can defend: persistent context, authorized evidence and a fail-closed boundary before unsupported claims reach customers.\n\nR$ 99 once · 1,000 queries · no subscription · reader BYOK\n\n*01*query.accepted**12 ms**\n\n*02*memory.retrieved**8 candidates**\n\n*03*policy.authorized**3 evidence**\n\n*04*claims.validated**6 / 6**\n\n`provenance_complete: true`\n\nFinal security suite\n\nReader success varies by task. Bridge containment is measured separately.\n\nSafety enforcement Answer quality Feedback transferFrozen evaluation · 2K evidence-token budget · reader outcomes are integrator-dependent\n\nCompleted retrieval result\n\nOn 979 frozen support questions, ASM-CM + Bridge 8.1 delivered 93.6% Recall@5 and a 66.5% diagnostic answer score while sending about 1.09K input tokens per question to the reader.\n\nWhat the system protects\n\nThe matrix organizes the documented risks across feedback, retrieval, authorization, delivery and lifecycle. It is a threat taxonomy—not a fabricated 22-test denominator.\n\n**Measured evidence:** 18/18 deterministic controls passed, and 45/45 unsafe deliveries were contained across nine bounded language scenarios. Reader quality remains a separate metric.\n\nCorrections improve retrieval without silently becoming source truth.\n\nThe system distinguishes retrieval failures from grounding failures.\n\nRetrieved evidence must actually support the claims being delivered.\n\nAuthority, deletion and operational behavior remain explicit boundaries.\n\nThe grounding layer\n\nAletheionAGI turns memory retrieval into a controlled delivery path. Your reader may change. The evidence contract does not.\n\nCanonical events stay under your authority.\n\nASM-CM, BM25 and vector retrieval work as one route.\n\nNamespace and labels constrain every candidate.\n\nUnsupported output is blocked before delivery.\n\nA clean boundary\n\nThe reader is an integration choice. AletheionAGI owns the harder boundary: canonical memory, retrieval provenance, policy enforcement and grounding.\n\nBuilt for buyer-facing systems\n\nUse AletheionAGI when remembered context can improve the experience—and an invented, stale or cross-account claim can damage it.\n\nKeep account history useful without letting stale or unauthorized memory shape the answer.\n\nRetrieve preferences and prior decisions while preserving tenant, namespace and evidence boundaries.\n\nGive teams durable organizational context with citations and explicit abstention when evidence is missing.\n\nA small, measurable first step\n\nRun a bounded proof of concept before choosing a production plan. Bring your reader key; we provide the grounded-memory boundary.\n\nBuilt for consequences\n\nOnly evidence needed for the current decision crosses the boundary.\n\nModels suggest. Policies, namespaces and canonical state decide.\n\nCorrections improve ranking without contaminating source memory.\n\nBefore you start\n\nNo. Your reader remains BYOK. AletheionAGI sits before delivery and controls memory, retrieval, authorized evidence and grounding.\n\nOne completed grounding evaluation consumes one credit, including a safe rejection or abstention. An internal platform failure returns the reserved credit.\n\nNo. It is a one-time proof of concept with 1,000 queries and 30-day validity. There is no automatic renewal or overage.\n\nYes. The one-time Test Plan uses secure Stripe Checkout. Production plans still require a commercial quote and final legal and tax review.\n\nStart small. Measure everything.", "url": "https://wpnews.pro/news/show-hn-aletheionagi-grounding-enforcement-for-ai-agents", "canonical_source": "https://www.aletheionagi.com", "published_at": "2026-08-14 19:28:17+00:00", "updated_at": "2026-08-14 19:42:26.902126+00:00", "lang": "en", "topics": ["ai-agents", "ai-safety", "ai-products", "ai-infrastructure"], "entities": ["AletheionAGI", "ASM-CM", "BM25", "Stripe Checkout"], "alternates": {"html": "https://wpnews.pro/news/show-hn-aletheionagi-grounding-enforcement-for-ai-agents", "markdown": "https://wpnews.pro/news/show-hn-aletheionagi-grounding-enforcement-for-ai-agents.md", "text": "https://wpnews.pro/news/show-hn-aletheionagi-grounding-enforcement-for-ai-agents.txt", "jsonld": "https://wpnews.pro/news/show-hn-aletheionagi-grounding-enforcement-for-ai-agents.jsonld"}}