Inject
Any data item
Connectors or approved one-off imports load email addresses, cookies, events, CRM rows, products, support, billing, documents, and outcomes.
Store
Attribution paths
Scout stores each item's relationship set, vector, and ordered attribution path in customer-owned PostgreSQL/pgvector. Fortress and Elite move high-load memory into Rust/LanceDB, not KynticAI product operations.
Compare
Rust + LanceDB
Enterprise contains the proprietary Rust relationship, weighting, traversal, and LanceDB path for fast similarity analysis across millions of relationship sets.
Output
Top-example JSON
The engine creates JSON with the best examples, weights, confidence, caveats, and ranked task options for the question being asked.
Explain
LLM task brief
Fortress sends JSON to your chosen LLM. Elite can use KynticAI's on-prem open-source LLM path with no third-party token charges.
Proof & Validation
This page shows the KynticAI product story in motion: Scout injects authorised data items and preserves attribution paths, Fortress adds the Rust/LanceDB relationship engine, and the output becomes top-example JSON the approved LLM can turn into a useful next-task explanation.
Buyer spark
The exciting part is not a badge on a page. It is seeing the whole chain click: authorised data item, attribution path, relationship comparison, top-example JSON, and a useful next task.
A buyer can see the product mechanics before they share sensitive operational data.
An investor can see the engine story without needing to read a repository.
A technical team can point to the exact JSON contract that makes the LLM useful.
Product Proof
The public proof is designed to answer the first real question: does the product chain make sense before any customer shares sensitive systems or records?
KynticAI's product operations surface supports reviewed commercial workflows: accounts, contacts, licences, entitlements, data-plane registration, aggregate usage, downloads, support, audit, health, OpenAPI, GraphQL, and lead handling.
Source: Product operations readiness material
The public contact path supports product updates, customer proof requests, and investor access requests with KynticAI team follow-up.
Source: Contact-form safety checks and hosted request flow
The website, product pages, scenario pages, assumption-based demo, waitlist path, and investor materials are in place for serious first conversations.
Source: IONOS go-live checks and current marketing site state
The Enterprise/Fortress evidence runtime proof is ready: a synthetic source can move through embedding, vector write, relationship analysis, and L3 evidence synthesis for technical review.
Source: Enterprise runtime notes, ready validation path, and L0-to-L3 benchmark material
The Enterprise core engine benchmark proof is ready: a privacy-safe synthetic, production-shaped proof lane passed with 100,000 vector seed rows, 101,000 final rows, 500 measured vector/search samples, and 19/19 ENT-008 checks.
Source: ENT-041/ENT-008 benchmark proof report and ready validation path
Scout provides the free open-source Universal Context Layer (UCL) path for source registration, selector shaping, snapshots, relationship facts, JSON output, APIs, and developer-facing integration behaviour.
Source: Scout integration docs and local package validation
Importance Engine and Clarity Engine are ready separate optional product lanes: Importance supports signal weighting, positive-response agents, and forensic pattern matching, while Clarity resolves ambiguous intent before model or agent routing.
Source: Importance Engine and Clarity Engine ready product proof
The stored case studies now show the product chain end to end: Scout captures the evidence in the PostgreSQL/pgvector proof path, Fortress compares relationship sets in the Rust/LanceDB runtime, and Elite receives governed JSON to produce ranked textual recommendations.
Scope: realistic synthetic demo evidence generated from the stored runtime pack. Customer deployments use authorised customer data, agreed source boundaries, and measured outcomes.
10
case studies run
34
Elite output prompts
960
relationship records
PASS
JSON validation
Case-study outputs
Logistics / supply chain
Cold-chain lane failure risk and control-tower intervention
High
Outcome: lane_stabilised
Action: Move the affected load to the contingency carrier, pre-alert the receiving dock, and start a control-tower exception bridge.
Legal / compliance
Privilege-safe matter escalation under deadline pressure
High
Outcome: privilege_safe_escalation_completed
Action: Escalate to the matter owner with a privilege-safe issue summary, deadline map, and outside-counsel question list.
Manufacturing / field operations
Asset downtime risk and predictive maintenance action
High
Outcome: downtime_prevented
Action: Reserve the critical spare, schedule a planned intervention window, and dispatch the qualified technician before automatic shutdown.
Education / university operations
Cohort progression signal and student-success support plan
High
Outcome: cohort_support_plan_started
Action: Start a cohort support plan with advisor outreach, assessment-deadline triage, and targeted workshop invitations.
Technical Workflow
01
Connectors or approved one-off imports load email addresses, cookies, browser events, web searches, registrations, CRM rows, support, usage, billing, product, document, and outcome items.
02
For each item, free open-source Scout stores the relationship set, vector, and ordered path of what happened, when it happened, and which source proved it inside the customer-owned PostgreSQL/pgvector proof store.
03
Enterprise/Fortress moves high-load relationship memory into the Rust engine and LanceDB vector database, linking entities, events, recency, contradiction, abstract patterns, weights, traversal, confidence, and caveats across millions of relationship sets.
04
For the question being asked, the output is governed JSON containing the top examples, attribution path, cited facts, similar successful patterns, estimated probabilities, confidence, and task options.
05
Fortress sends top-example JSON to the customer's chosen LLM, such as ChatGPT Enterprise or an internal model. Elite can use KynticAI's open-source on-prem LLM model without third-party token charges.
06
The LLM turns the JSON into a text explanation of the best task to achieve the goal, with human review and the outcome feeding future relationship analysis.
Relationship-Analysis JSON Example
The public schema shape shows what the LLM receives: data items, attribution path, relationship set, top examples for the question, recommendation options, estimated probabilities, confidence, next action, caveats, and provenance. Fortress sends that JSON to the customer's own model; Elite can use the KynticAI open-source on-prem LLM model, so no third-party LLM token charge is created by the explanation step.
{
"schema": "kynticai.relationship_analysis.example.v1",
"privacySafeSyntheticExample": true,
"validationPath": "public shape backed by local validation path; customer pilots replace this with authorised data",
"subject": {
"entityType": "contact",
"emailAddress": "testname@test.com",
"crmContactId": "crm_contact_synth_4821"
},
"rawDataBoundary": {
"rawOperationalDataLocation": "customer-owned data plane",
"controlPlaneReceives": [
"licensing metadata",
"update metadata",
"aggregate usage",
"health state",
"support and commercial records"
],
"controlPlaneDoesNotReceive": [
"connector credentials",
"raw customer records",
"prompt payloads",
"source exports"
]
},
"authorisation": {
"purpose": "sales_next_best_action",
"role": "revenue_operations",
"auditId": "audit_synth_2026_06_17_001",
"retentionPolicy": "pilot scoped"
},
"sourceSignals": {
"emailEnquiry": "new inbound enquiry",
"cookie": "web_cookie_4281",
"webSearch": "page_a",
"productInterest": "product_b",
"crmStatus": "new enquiry",
"accountRegistration": "not yet registered",
"supportHistory": "none for this contact",
"productUsage": "not yet active",
"billingStatus": "not a customer yet",
"wonLostSaleOutcome": "similar converted and non-converted journeys available for comparison"
},
"attributionPath": [
{ "order": 1, "item": "emailAddress", "value": "testname@test.com", "event": "email_enquiry_received" },
{ "order": 2, "item": "cookie", "value": "web_cookie_4281", "event": "web_search_page_a" },
{ "order": 3, "item": "product", "value": "product_b", "event": "product_interest_recorded" }
],
"relationshipSet": {
"subject": "testname@test.com",
"relationships": [
"same_email_inquiry_pattern",
"same_product_browse_path",
"email_enquiry_plus_generic_web_search",
"not_yet_registered_account"
],
"storedIn": "Scout PostgreSQL/pgvector proof store; Enterprise/Fortress Rust/LanceDB private runtime for high load"
},
"relationshipWeights": [
{ "signal": "emailEnquiry", "weight": 0.21, "direction": "positive" },
{ "signal": "pageASearch", "weight": 0.18, "direction": "positive" },
{ "signal": "productBInterest", "weight": 0.24, "direction": "positive" },
{ "signal": "noAccountRegistrationYet", "weight": 0.09, "direction": "caveat" }
],
"similarOutcomePatterns": {
"converted": "previous contacts with email enquiry + page A search + product B interest converted more often after fast follow-up",
"notConverted": "previous contacts with email enquiry only often cooled without account registration or timely response"
},
"topExamplesForQuestion": [
{
"question": "How do we convert similar email enquiries to a sale?",
"examplePath": "email enquiry -> page A search -> product B interest -> fast follow-up -> account registration -> sale",
"matchStrength": 0.78
},
{
"question": "How do we get similar users to search the site again?",
"examplePath": "email enquiry -> product B answer -> guided page A link -> second product search",
"matchStrength": 0.64
}
],
"recommendation": {
"summary": "Send a focused follow-up email and offer a simple account registration path.",
"valueTarget": "increase conversion probability",
"similarSuccessfulPattern": "email enquiry + page A search + product B interest",
"confidence": 0.78,
"options": [
{ "task": "send_follow_up_email", "estimatedProbability": 0.72 },
{ "task": "ask_user_to_register_account", "estimatedProbability": 0.61 }
],
"nextAction": "Send a relationship-backed follow-up that references product B and offers the shortest account registration path.",
"caveats": [
"synthetic public example",
"human review required",
"not an outcome promise"
]
},
"llmHandoff": {
"target": "Fortress: customer-owned LLM such as ChatGPT Enterprise; Elite: KynticAI open-source on-prem LLM model",
"instruction": "Explain the best task to do next using the relationship analysis, caveats, and source trail."
}
}Validation Paths
These validation paths help technical buyers decide where to start. Customer-specific connector credentials, operational records, and deployment details stay inside the agreed walkthrough.
Open-core demo path
Available for technical review
Workflow automation slice
Hosted-validated and available for technical review
Private connector families
Validated and ready for private technical walkthrough
Document and object-store expansion
Hosted-validated and ready for controlled review
Clean Boundaries
The product workflow, source categories, runtime evidence shape, JSON handoff, and privacy-safe examples that explain how KynticAI works.
Connector-specific evidence, deployment topology, run logs, and the customer-approved data-plane details needed to scope a serious rollout.
Raw operational data, connector credentials, prompt payloads, and relationship facts stay inside the customer-controlled data plane by default.
Next Step
Evidence Results
These examples show how source items become relationship paths, JSON output, and task explanation.
Proof scenario - stored JSON, relationship sets, generated Elite output
Inspect the evidence chain
“The proof page shows how stored relationship sets become top-example JSON and generated task recommendations.”
Proof scenario - Scout, Fortress, Elite, connector path
See the full product chain
“Scout proves the data plane, Fortress proves private relationship analysis, and Elite proves JSON-to-text task explanation.”
Proof scenario - customer-owned data plane, product-operations metadata
Trust the boundary before the pilot
“The proof story shows where customer evidence stays private and what minimal commercial metadata KynticAI product operations need.”