--- title: "Semantica Review 2026: AI Agent Audit Trails" canonical: https://wavect.io/blog/semantica-ai-agent-decision-provenance/ language: en description: "Semantica review for regulated AI agents: decision provenance, PROV-O exports, policy gates, self-hosting, security limits and a pilot scorecard." image: "https://wavect.io/img/blog/headers/header_semantica-ai-agent-decision-provenance.png" --- [**Back**](/blog/overview/) [![Kevin Riedl](/img/team/kevin.webp)](/team/kevin-riedl/) [Kevin Riedl](/team/kevin-riedl/) https://linkedin.com/in/wsdt 13 min read · 14 Aug 2026 Last reviewed August 14, 2026 [**Next**](/blog/cloudflare-kitesurf-browser-ai-agents/) # Semantica Review 2026: Can It Explain Every AI Agent Decision? TL;DR Semantica is an MIT-licensed, self-hostable Python platform that adds a context graph, first-class decision records, causal links, policy checks and provenance exports to an existing agent stack. Its deterministic graph traversal and rules can reproduce an explanation from stored inputs, but they cannot prove that extracted facts or an agent-written reason were correct. Version 0.6.5, released on 11 August 2026, fixed critical authentication and injection flaws, so buyers should pin that version or newer and perform their own security review. Semantica is a credible pilot for consequential, repeated decisions that need evidence and precedent. It is excessive for low-risk chat, simple workflows or teams without a defined decision schema, retention policy and accountable reviewer. **Semantica is worth a controlled pilot when an AI agent makes repeated, consequential decisions and a reviewer must reconstruct the evidence, policy and causal chain months later.** It records decisions as graph objects instead of leaving them as log lines or transient model context. That is a materially different job from ordinary vector memory. Our commercial verdict is conditional. The software is MIT licensed, but an audit-ready operating model still needs access controls, retention rules, tamper evidence, tested policies and an accountable human reviewer. We reviewed the public [Semantica repository and documentation](https://github.com/semantica-agi/semantica) on 14 August 2026. This is an evidence-based buyer review, not a sponsored post, legal opinion or hands-on penetration test. This article owns the product-evaluation question: should you pilot Semantica for AI agent decision provenance? For the broader architecture decision between workflow graphs, experiment DAGs, SQL, vector RAG and knowledge graphs, use our [graph engineering guide for AI agents](/blog/graph-engineering-ai-agents/). Keeping those questions separate prevents a tool review from pretending every graph problem has the same answer. ## What is Semantica? **Semantica is a self-hostable Python platform for context graphs, knowledge graphs, deterministic rules and decision provenance.** It sits beside an existing LLM, vector store and agent framework. The product does not need to replace your orchestration layer to record a decision, connect it to causes, search past decisions for precedent or export provenance. | Public fact, checked 14 August 2026 | Buyer implication | | --- | --- | | MIT licence | No licence fee for commercial use, but no warranty or implementation support is included. | | Python package and self-hosted deployment | You can keep the graph in your infrastructure if every configured extraction and model path stays there too. | | Decision records, causal links and precedent search | A decision becomes queryable business data rather than an ephemeral trace. | | Forward chaining, Rete, Datalog and SPARQL | Rule execution and graph traversal can be repeated without asking an LLM to invent the explanation. | | W3C PROV-O, RDF, JSON-LD and other exports | The trail can move into standards-based evidence and graph workflows. | | Databricks, Snowflake, MCP, REST, CLI and editor integrations | A pilot can attach to an existing data and agent stack, but each connector expands the security review. | ## Why are embeddings and logs not enough? A vector store answers a similarity question: which passages resemble this query? A trace answers an execution question: which prompt, model or tool call ran? A decision record must answer a harder business question: which evidence and policy produced this outcome, what alternatives were rejected, who approved it and what happened next? Those layers complement each other. Keep vector retrieval for fuzzy recall and OpenTelemetry-style traces for latency, errors and tool calls. Add a decision graph only where the relationship between evidence, policy, decision and consequence must remain searchable after the original run has disappeared. | Layer | Question it answers | What it does not prove | | --- | --- | --- | | Vector memory | What context is semantically similar? | Which exact facts caused the final decision | | Agent trace | What executed, in which order and how long did it take? | Why a policy permitted the outcome | | Policy engine | Did structured inputs satisfy a versioned rule? | Whether those inputs were complete and correct | | Decision provenance graph | Which evidence, actors, rules and prior decisions connect to this outcome? | That the recorded evidence represents reality | ## How does Semantica record an agent decision? The central API records a category, scenario, reasoning text, outcome, confidence and optional metadata. Additional relationships mark one decision as a cause, influence or precedent for another. Teams can then trace upstream causes, inspect downstream impact, search similar decisions and run policy checks before an outcome stands. Consider a loan workflow. The graph could link the application snapshot, verified income record, policy version, risk tier, model version, human review and final adverse-action notice. Six months later, a reviewer should be able to select one outcome and traverse back to the exact versioned inputs, not regenerate a plausible story from the current model. 1. **Capture the decision boundary.** Define where analysis becomes a recommendation, approval, rejection or action. 2. **Record versioned evidence.** Store stable record IDs, source versions, valid time, permission scope and checksums instead of copying uncontrolled prose. 3. **Attach the policy.** Save the rule-set version, matched rule, failed alternatives, exception and human override. 4. **Connect consequences.** Link the decision to the notice, transaction, case or downstream decision it produced. 5. **Test reconstruction.** An independent reviewer must reproduce the path without relying on the original agent's context window. ## Is Semantica's reasoning really deterministic? **The graph traversal and symbolic rule engines can be deterministic while the complete AI workflow is not.** Given the same stored graph, rule set and query, a forward-chain or causal traversal can return the same path without an LLM. That is useful because the explanation is derived from persisted relationships rather than generated from scratch. The boundary matters. Entity extraction may use patterns, machine-learning models or LLMs. The agent can also write the original `reasoning` field. Deterministic traversal cannot repair a wrong source, stale relationship, biased policy or fabricated rationale. Buyers should therefore test three separate properties: repeatability of the traversal, fidelity of the recorded inputs and validity of the business rule. ## Does PROV-O make a system compliant? No. The [W3C PROV-O Recommendation](https://www.w3.org/TR/prov-o/) provides interoperable concepts for entities, activities and agents, including which activity used or generated an entity and who was responsible. It is a useful evidence schema. It is not a compliance certificate, retention policy, access-control system or guarantee that a regulator will accept a specific export. The legal obligation depends on the use case. The [EU AI Act](https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng) requires high-risk systems to support automatic event logging under Article 12 and gives affected people a scoped right to clear and meaningful explanations for certain individual decisions under Article 86. A provenance graph can help assemble that evidence, but a deployer still has to determine classification, roles, retention, human oversight and the explanation shown to a person. Credit is even more concrete in the United States. The [CFPB's circular on complex credit algorithms](https://www.consumerfinance.gov/compliance/circulars/circular-2022-03-adverse-action-notification-requirements-in-connection-with-credit-decisions-based-on-complex-algorithms/) says a creditor must provide specific and accurate principal reasons for adverse action. “The model is too complex” is not a defence. A stored causal path helps only when its reasons match the factors actually considered. The [NIST AI RMF Core](https://airc.nist.gov/airmf-resources/airmf/5-sec-core/) also treats documentation as support for transparency, human review and accountability. That is the right procurement frame: Semantica can be an evidence substrate inside a governance system, not the governance system by itself. ## Can Semantica keep regulated data inside your infrastructure? It can be self-hosted, and its native Databricks and Snowflake connectors can read governed data where teams already manage it. That reduces the need for a separate SaaS export. It does not automatically mean that no data leaves your environment. Map the complete path before approval: document parsers, embedding models, LLM-based extraction, vector stores, graph backends, MCP clients, REST consumers, backups and telemetry. If any configured provider is remote, relevant inputs may cross that boundary. Test tenant isolation and row-level permissions at retrieval time, not only during ingestion. ## What should security teams know about version 0.6.5? The current [Semantica 0.6.5 security release](https://github.com/semantica-agi/semantica/releases/tag/v0.6.5), published on 11 August 2026, fixed six externally reported vulnerabilities. The release notes include critical missing authentication on Explorer API routes, critical Cypher and SPARQL injection paths, plus RCE, XXE, denial-of-service and ReDoS fixes. That response is evidence of active maintenance, but the underlying findings make a security review mandatory for any high-stakes pilot. 1. **Pin 0.6.5 or newer.** Block older versions in deployment and software-composition checks. 2. **Keep the Explorer private.** Put authentication, authorization, rate limits and network policy in front of every API surface, then verify them independently. 3. **Treat graph labels as untrusted input.** Fuzz document-derived entity types, relationship types, IRIs and property names against the chosen backend. 4. **Separate evidence from presentation.** Keep an append-only or tamper-evident record outside the interface an agent can modify. 5. **Plan for 0.x change.** Pin dependencies, review migrations and rerun the decision-reconstruction suite before upgrades. ## Semantica vs RAG, observability and policy-as-code | If your primary need is | Start with | Add Semantica when | | --- | --- | --- | | Answer questions from documents | Permission-aware RAG with citations | Answers require durable multi-hop facts, conflicts and decision history | | Debug agent runs | Tracing, logs and evaluations | Reviewers need precedent and causal reconstruction across runs | | Block unsafe actions | A narrow policy engine at the tool boundary | Policies, evidence, overrides and consequences must form a queryable record | | Manage enterprise controls | Your GRC and records-management systems | Agent decisions need to feed those systems in a structured provenance format | | Model connected operational facts | SQL or a knowledge graph, based on query shape | Decisions themselves must become first-class, causally linked objects | ## Who should pilot Semantica, and who should skip it? | Situation | Verdict | Reason | | --- | --- | --- | | Repeated credit, claims, eligibility or case-routing decisions | Pilot in shadow mode | The same evidence and policy questions recur, with real review cost. | | Several agents need one versioned decision history | Pilot | Precedent and causal links can compound across workflows. | | Low-risk chatbot or one-off document Q&A | Skip | Traces and cited retrieval are cheaper and easier to operate. | | No stable decision schema or policy owner | Fix governance first | A graph will formalize ambiguity, not remove it. | | Team expects automatic regulatory compliance | Do not buy on that premise | PROV-O export is an evidence format, not a legal sign-off. | ## What does an open-source decision trail really cost? The licence is free. The operating capability is not. Budget for ontology and policy design, source integration, identity resolution, permission enforcement, storage, retention, redaction, tamper evidence, evaluations, security testing, reviewer tooling and incident response. Databricks or Snowflake connectivity saves one integration step, but it does not remove data-quality or access-control work. The commercial metric is not nodes created. Measure reviewer minutes per decision, percentage of material factors with source evidence, time to reconstruct an incident, policy-gate false positives and negatives, export acceptance, and the cost of keeping the graph current. ## How should a two-week Semantica pilot work? 1. **Choose one consequential decision.** Run in shadow mode on 30 to 50 historical cases. Do not begin with automated approval. 2. **Define the required explanation first.** Compliance, operations and engineering should agree on material factors, evidence, policy version, overrides and retention. 3. **Build a baseline.** Measure how long the current logs, database and staff need to reconstruct each case. 4. **Pin and isolate the stack.** Use version 0.6.5 or newer, private network access, test data and the smallest connector set. 5. **Run adversarial cases.** Include missing evidence, conflicting facts, stale policies, unauthorized tenants, malicious labels, human overrides and deleted source records. 6. **Set a kill rule.** Stop if the system cannot reconstruct at least 95% of material factors, if any cross-tenant access succeeds, or if reviewer effort does not improve enough to cover maintenance. If the pilot passes, Wavect's [AI enablement team](/services/ai-enablement/) can connect the decision layer to permissions, evaluations and production workflows. Review our [AI agent evaluation and sandbox security checklist](/blog/ai-agent-eval-sandbox-security-checklist/), inspect the [Twinsoft AI case study](/case-studies/twinsoft-ai/), compare the [technology-stack decision guide](/software-development-guide/how-to-choose-a-tech-stack-for-mvp/), or [book a decision provenance workshop](/contact/). ## Frequently Asked Questions ### What is Semantica? Semantica is an MIT-licensed, self-hostable Python platform for context graphs, knowledge graphs, deterministic reasoning and first-class AI decision records. It exposes capabilities through Python, MCP, REST and a CLI. ### Is Semantica an AI agent framework? Not primarily. It is an accountability and context layer designed to complement an existing LLM, vector store and agent framework. Teams can adopt its decision and provenance components without replacing orchestration. ### Is Semantica's reasoning deterministic? Its symbolic rules and graph traversal can be deterministic for the same stored facts and rules. The complete workflow may still use probabilistic extraction or agent-generated reasoning, so repeatability does not guarantee input correctness. ### Does PROV-O export make Semantica compliant? No. PROV-O provides an interoperable provenance model. Compliance also requires correct use-case classification, access controls, retention, human oversight, evidence quality and an explanation that meets the applicable law. ### Does Semantica replace a vector database or observability platform? No. Vector retrieval remains useful for semantic recall, while tracing remains useful for execution and performance. Semantica adds queryable decision, causal and provenance relationships where those records have durable value. ### Can Semantica keep data inside our infrastructure? It can be self-hosted and connect to Databricks and Snowflake. You still need to inspect every parser, embedding, LLM, MCP, API, graph store, backup and telemetry path before claiming that data never leaves your environment. ### How should a regulated team evaluate Semantica? Use historical cases in shadow mode and score evidence completeness, causal reconstruction, policy accuracy, tenant isolation, tamper resistance, reviewer effort and export usability. Pin version 0.6.5 or newer and require independent security and legal review. ## Final thoughts Semantica targets a real gap between agent observability and accountable decisions. Its first-class decision records, causal graph, policy checks and standards-based provenance make it a credible open-source pilot for teams that must answer why an agent acted. The honest verdict is not ‘install it and become compliant.’ Deterministic paths are only as defensible as the facts, policies and controls recorded underneath them. Pin the security-fixed release, start with one shadow workflow and keep Semantica only if an independent reviewer reaches a source-backed explanation faster and more reliably. ## You may also like.. [**Graph Engineering for AI Agents** Decide when a workflow graph, experiment DAG, vector store, SQL model or knowledge graph earns its operating cost.](/blog/graph-engineering-ai-agents/) [**AI enablement vs generic AI consulting** Compare an implementation pilot on your infrastructure with a strategy-only engagement.](/compare/ai-enablement-vs-generic-ai-consultancy/) AI governance and regulation ## Continue through this cluster Security, policy, compliance and operating controls for responsible AI adoption. [Start with the cornerstone**EU AI Act Cost for a 5-Person Startup**](/blog/eu-ai-act-compliance-cost-startup/) - [AI Agent Contract Signing: eIDAS QES Integration Guide](/blog/ai-agent-eidas-signature-integration/) - [Terafab: Who Controls the AI Stack From Silicon to Orbit?](/blog/terafab-vertical-integration-ai-stack/) - [Can AI Create Viruses? What Stanford Actually Proved](/blog/ai-designed-viruses-stanford-biosecurity/) - [Stripe Billing and E-Invoicing 2027](/blog/stripe-billing-e-invoicing-2027/) - [EU AI Act Article 50 Checklist for SaaS and AI Agents](/blog/eu-ai-act-article-50-checklist/) Inbox, without the noise ## Follow the work that matters to you Get a short email when we publish something new. Follow the whole blog or only the problems you care about. [**Back**](/blog/overview/) [![Kevin Riedl](/img/team/kevin.webp)](/team/kevin-riedl/) [Kevin Riedl](/team/kevin-riedl/) https://linkedin.com/in/wsdt 13 min read · 14 Aug 2026 Last reviewed August 14, 2026 [**Next**](/blog/cloudflare-kitesurf-browser-ai-agents/) New posts by email × × Get new posts by email A short email when we publish. Free, no tracking. ## Structured Data ```json { "@context": "https://schema.org", "@graph": [ { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ], "employee": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "founder": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "legalRepresentative": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "name": "Wavect GmbH", "subjectOf": { "@id": "https://wavect.io/verified-claims.json#dataset", "@type": "Dataset", "creator": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "description": "A machine-readable registry of quantitative and qualitative claims published by Wavect, with review dates, localized page appearances and public third-party citations where available.", "inLanguage": "en", "isAccessibleForFree": true, "license": "https://creativecommons.org/licenses/by/4.0/", "name": "Wavect verified publication claims", "url": "https://wavect.io/verified-claims.json" }, "url": "https://wavect.io/" }, { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "sameAs": [ "https://www.wikidata.org/wiki/Q139796365", "https://www.linkedin.com/in/wsdt", "https://github.com/wsdt" ], "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "sameAs": [ "https://www.wikidata.org/wiki/Q139796367", "https://www.linkedin.com/in/jocr77/", "https://github.com/jo-chris" ], "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/#website", "@type": "WebSite", "inLanguage": [ "en", "de", "es", "zh" ], "name": "Wavect", "potentialAction": { "@type": "SearchAction", "query-input": "required name=search_term_string", "target": { "@type": "EntryPoint", "urlTemplate": "https://wavect.io/search/?q={search_term_string}" } }, "publisher": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "url": "https://wavect.io/" }, { "@id": "https://wavect.io/blog/semantica-ai-agent-decision-provenance/#webpage", "@type": "WebPage", "dateModified": "2026-08-14", "inLanguage": "en", "isPartOf": { "@id": "https://wavect.io/#website", "@type": "WebSite" }, "lastReviewed": "2026-08-14", "url": "https://wavect.io/blog/semantica-ai-agent-decision-provenance/" } ] } ``` ```json { "@context": "https://schema.org", "@type": "BlogPosting", "abstract": "Semantica is an MIT-licensed, self-hostable Python platform that adds a context graph, first-class decision records, causal links, policy checks and provenance exports to an existing agent stack. Its deterministic graph traversal and rules can reproduce an explanation from stored inputs, but they cannot prove that extracted facts or an agent-written reason were correct. Version 0.6.5, released on 11 August 2026, fixed critical authentication and injection flaws, so buyers should pin that version or newer and perform their own security review. Semantica is a credible pilot for consequential, repeated decisions that need evidence and precedent. It is excessive for low-risk chat, simple workflows or teams without a defined decision schema, retention policy and accountable reviewer.", "articleBody": " Blog overview/Business and regulation/AI governance and regulation Semantica Review 2026: Can It Explain Every AI Agent Decision? TL;DR Semantica is an MIT-licensed, self-hostable Python platform that adds a context graph, first-class decision records, causal links, policy checks and provenance exports to an existing agent stack. Its deterministic graph traversal and rules can reproduce an explanation from stored inputs, but they cannot prove that extracted facts or an agent-written reason were correct. Version 0.6.5, released on 11 August 2026, fixed critical authentication and injection flaws, so buyers should pin that version or newer and perform their own security review. Semantica is a credible pilot for consequential, repeated decisions that need evidence and precedent. It is excessive for low-risk chat, simple workflows or teams without a defined decision schema, retention policy and accountable reviewer. Semantica is worth a controlled pilot when an AI agent makes repeated, consequential decisions and a reviewer must reconstruct the evidence, policy and causal chain months later. It records decisions as graph objects instead of leaving them as log lines or transient model context. That is a materially different job from ordinary vector memory. Our commercial verdict is conditional. The software is MIT licensed, but an audit-ready operating model still needs access controls, retention rules, tamper evidence, tested policies and an accountable human reviewer. We reviewed the public Semantica repository and documentation on 14 August 2026. This is an evidence-based buyer review, not a sponsored post, legal opinion or hands-on penetration test. This article owns the product-evaluation question: should you pilot Semantica for AI agent decision provenance? For the broader architecture decision between workflow graphs, experiment DAGs, SQL, vector RAG and knowledge graphs, use our graph engineering guide for AI agents. Keeping those questions separate prevents a tool review from pretending every graph problem has the same answer. What is Semantica? Semantica is a self-hostable Python platform for context graphs, knowledge graphs, deterministic rules and decision provenance. It sits beside an existing LLM, vector store and agent framework. The product does not need to replace your orchestration layer to record a decision, connect it to causes, search past decisions for precedent or export provenance. Public fact, checked 14 August 2026Buyer implication MIT licenceNo licence fee for commercial use, but no warranty or implementation support is included. Python package and self-hosted deploymentYou can keep the graph in your infrastructure if every configured extraction and model path stays there too. Decision records, causal links and precedent searchA decision becomes queryable business data rather than an ephemeral trace. Forward chaining, Rete, Datalog and SPARQLRule execution and graph traversal can be repeated without asking an LLM to invent the explanation. W3C PROV-O, RDF, JSON-LD and other exportsThe trail can move into standards-based evidence and graph workflows. Databricks, Snowflake, MCP, REST, CLI and editor integrationsA pilot can attach to an existing data and agent stack, but each connector expands the security review. Why are embeddings and logs not enough? A vector store answers a similarity question: which passages resemble this query? A trace answers an execution question: which prompt, model or tool call ran? A decision record must answer a harder business question: which evidence and policy produced this outcome, what alternatives were rejected, who approved it and what happened next? Those layers complement each other. Keep vector retrieval for fuzzy recall and OpenTelemetry-style traces for latency, errors and tool calls. Add a decision graph only where the relationship between evidence, policy, decision and consequence must remain searchable after the original run has disappeared. LayerQuestion it answersWhat it does not prove Vector memoryWhat context is semantically similar?Which exact facts caused the final decision Agent traceWhat executed, in which order and how long did it take?Why a policy permitted the outcome Policy engineDid structured inputs satisfy a versioned rule?Whether those inputs were complete and correct Decision provenance graphWhich evidence, actors, rules and prior decisions connect to this outcome?That the recorded evidence represents reality How does Semantica record an agent decision? The central API records a category, scenario, reasoning text, outcome, confidence and optional metadata. Additional relationships mark one decision as a cause, influence or precedent for another. Teams can then trace upstream causes, inspect downstream impact, search similar decisions and run policy checks before an outcome stands. Consider a loan workflow. The graph could link the application snapshot, verified income record, policy version, risk tier, model version, human", "articleSection": "AI Governance", "author": { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "name": "Kevin Riedl", "sameAs": [ "https://www.wikidata.org/wiki/Q139796365", "https://www.linkedin.com/in/wsdt", "https://github.com/wsdt" ], "url": "https://wavect.io/team/kevin-riedl/" }, "citation": [ { "@type": "WebPage", "name": "Semantica repository and documentation", "url": "https://github.com/semantica-agi/semantica" }, { "@type": "WebPage", "name": "W3C PROV-O Recommendation", "url": "https://www.w3.org/TR/prov-o/" }, { "@type": "WebPage", "name": "EU AI Act", "url": "https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng" }, { "@type": "WebPage", "name": "CFPB's circular on complex credit algorithms", "url": "https://www.consumerfinance.gov/compliance/circulars/circular-2022-03-adverse-action-notification-requirements-in-connection-with-credit-decisions-based-on-complex-algorithms/" }, { "@type": "WebPage", "name": "NIST AI RMF Core", "url": "https://airc.nist.gov/airmf-resources/airmf/5-sec-core/" }, { "@type": "WebPage", "name": "Semantica 0.6.5 security release", "url": "https://github.com/semantica-agi/semantica/releases/tag/v0.6.5" } ], "dateModified": "2026-08-14", "datePublished": "2026-08-14", "description": "Semantica is an MIT-licensed, self-hostable Python platform that adds a context graph, first-class decision records, causal links, policy checks and provenance exports to an existing agent stack. Its deterministic graph traversal and rules can reproduce an explanation from stored inputs, but they cannot prove that extracted facts or an agent-written reason were correct. Version 0.6.5, released on 11 August 2026, fixed critical authentication and injection flaws, so buyers should pin that version or newer and perform their own security review. Semantica is a credible pilot for consequential, repeated decisions that need evidence and precedent. It is excessive for low-risk chat, simple workflows or teams without a defined decision schema, retention policy and accountable reviewer.", "headline": "Semantica Review 2026: Can It Explain Every AI Agent Decision?", "image": "https://wavect.io/img/blog/headers/header_semantica-ai-agent-decision-provenance.svg", "inLanguage": "en", "keywords": "AI Governance, AI Agents", "mainEntityOfPage": { "@id": "https://wavect.io/blog/semantica-ai-agent-decision-provenance/", "@type": "WebPage" }, "publisher": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "url": "https://wavect.io/blog/semantica-ai-agent-decision-provenance/", "wordCount": 2520 } ``` ```json { "@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "item": "https://wavect.io/", "name": "Home", "position": 1 }, { "@type": "ListItem", "item": "https://wavect.io/blog/overview/", "name": "Blog overview", "position": 2 }, { "@type": "ListItem", "item": "https://wavect.io/blog/topics/business-regulation/", "name": "Business and regulation", "position": 3 }, { "@type": "ListItem", "item": "https://wavect.io/blog/clusters/ai-governance/", "name": "AI governance and regulation", "position": 4 }, { "@type": "ListItem", "item": "https://wavect.io/blog/semantica-ai-agent-decision-provenance/", "name": "Semantica Review 2026: AI Agent Audit Trails | ", "position": 5 } ] } ``` ```json { "@context": "https://schema.org", "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Semantica is an MIT-licensed, self-hostable Python platform for context graphs, knowledge graphs, deterministic reasoning and first-class AI decision records. It exposes capabilities through Python, MCP, REST and a CLI." }, "name": "What is Semantica?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Not primarily. It is an accountability and context layer designed to complement an existing LLM, vector store and agent framework. Teams can adopt its decision and provenance components without replacing orchestration." }, "name": "Is Semantica an AI agent framework?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Its symbolic rules and graph traversal can be deterministic for the same stored facts and rules. The complete workflow may still use probabilistic extraction or agent-generated reasoning, so repeatability does not guarantee input correctness." }, "name": "Is Semantica's reasoning deterministic?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "No. PROV-O provides an interoperable provenance model. Compliance also requires correct use-case classification, access controls, retention, human oversight, evidence quality and an explanation that meets the applicable law." }, "name": "Does PROV-O export make Semantica compliant?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "No. Vector retrieval remains useful for semantic recall, while tracing remains useful for execution and performance. Semantica adds queryable decision, causal and provenance relationships where those records have durable value." }, "name": "Does Semantica replace a vector database or observability platform?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "It can be self-hosted and connect to Databricks and Snowflake. You still need to inspect every parser, embedding, LLM, MCP, API, graph store, backup and telemetry path before claiming that data never leaves your environment." }, "name": "Can Semantica keep data inside our infrastructure?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Use historical cases in shadow mode and score evidence completeness, causal reconstruction, policy accuracy, tenant isolation, tamper resistance, reviewer effort and export usability. Pin version 0.6.5 or newer and require independent security and legal review." }, "name": "How should a regulated team evaluate Semantica?" } ] } ```