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.

Product / KynticAI Scout

The free, open-source layer above any authorised company data set.

Scout is the free, open-source entry point to the Universal Context Layer (UCL): it injects authorised data items through connectors or approved one-off imports, stores those items as relationship sets with attribution paths in PostgreSQL/pgvector, and exposes governed JSON for the LLM that explains what task should happen next.

Start free and open source, put a sovereign relationship memory above company data, then ask it the best way to achieve a goal.

Buyer spark

Start free. Prove the layer. Make the enterprise sale obvious.

Scout gives a buyer the first hit of belief: one authorised source becomes relationship memory, then relationship memory becomes JSON a model can explain.

No vague AI promise: the buyer can inspect the source, selector, relationship fact, and JSON.

The free open-source path makes the first conversation easier because the product is tangible.

When Scout hits proof scale, Fortress and Elite become the natural upgrade instead of a cold upsell.

Plain English

What Scout does and how the relationships support the next task.

What it does

Scout is free and open source. It sits above authorised company data and turns ordinary records, cookies, events, email addresses, products, accounts, and outcomes into relationship sets an AI workflow can use.

How it works

You inject data items, identify the object they belong to, preserve the ordered attribution path, store vectors in PostgreSQL/pgvector, compare similar relationship sets, and expose governed JSON through clean APIs.

Commercial value path

It lets buyers prove one high-value question locally, such as how to convert this email enquiry, before moving into Enterprise/Fortress for the private Rust/LanceDB runtime.

Task moment

An email, a cookie, a web search, and product interest become: here are the top examples, here is the best task, and here is why.

What you get

The concrete deliverables behind Scout.

Free open-source entry point

Scout can be run and inspected without a licence fee so buyers can prove the Universal Context Layer mechanics before Enterprise/Fortress deployment.

Data-item injection

A local data-plane path for authorised connector loads plus approved one-off imports or mapping work, including Claude/Codex-assisted preparation when the customer permits it.

Item attribution paths

Store the ordered path for a customer, email address, cookie, browser event, product, account, or case: what happened, when it happened, and which source proved it.

Relationship-set JSON

REST and GraphQL-shaped output that carries cited data items, attribution path, similar examples, confidence, caveats, and ranked task options for a selected goal.

Proof-scale PostgreSQL/pgvector store

Scout uses PostgreSQL/pgvector for the free open-source path. Use it for developer proof, first-source validation, and lower-load pilots, not high-concurrency or high-performance vector-search installations.

Enterprise upgrade route

A clear path from free open-source Scout into Enterprise/Fortress private connectors, proprietary Rust/LanceDB relationship analysis, LanceDB-backed vector storage, sovereign deployment, and stronger governance.

Example data walkthrough

Sales next-best task from an email enquiry and product-interest trail

Privacy-safe synthetic example backed by a real validation path. Customer pilots replace the sample data with authorised source systems and agreed success measures.

01 / Source row

Scout injects the authorised data items

email = testname@test.com

cookie = web_cookie_4281

event = email_enquiry

web_search = page_a

product_interest = product_b

crm_status = new enquiry

outcome_history = converted / did_not_convert

A normal enquiry becomes an item with identity, event history, and enough context to compare against previous similar journeys.

02 / Relationship facts

The attribution path becomes searchable memory

attributionPath = email_enquiry -> page_a -> product_b

sameEmailPattern = matched

sameProductBrowse = matched

registeredAccount = not_yet

abstractPattern = email enquiry + generic web search

Scout can prove the relationship memory in PostgreSQL/pgvector; Enterprise/Fortress moves the same pattern into the Rust/LanceDB runtime for high-volume comparison.

03 / Money move

The top examples become task JSON

topExample = previous email + page_a + product_b conversion path

option_1 = send follow-up email | estimated_probability = 72%

option_2 = ask user to register account | estimated_probability = 61%

Fortress handoff = JSON to customer LLM

Elite handoff = on-prem open-source LLM

The LLM receives a JSON file with the best examples and turns it into a text explanation of what to do next.

How it works

The private data plane above the systems you already own

Scout is the free, open-source Universal Context Layer (UCL) data-plane path. It injects authorised data items, stores relationship sets, ordered attribution paths, and vectors in PostgreSQL/pgvector, assembles similar examples for the goal, and exposes JSON to consumers while keeping customer data under local control.

Inject

Connectors or one-off imports

Describe databases, APIs, files, customer-approved import packs, and system boundaries while the customer data plane remains the owner of raw records.

Store

PostgreSQL/pgvector proof store

Use repeatable mappings to store data items, identity links, source order, purpose, provenance, vectors, and the path each item took in the inspectable Scout data plane.

Compare

Similar relationship sets

Find same or abstract relationships such as same email, same product browse, same registration path, or email enquiry plus generic web search.

Serve

Top-example JSON

Expose snapshots, top examples, relationship facts, and task-ready JSON through REST, GraphQL, admin UI, webhooks, and downstream model clients.

What this unlocks

The practical moves that make Scout worth paying for

Domain entities

Represent customers, email addresses, cookies, browser events, assets, tickets, deals, cases, products, or any operational object that needs relationship memory.

Attribution paths

Store the order of events and sources for each item so the system knows what happened before the goal was reached or missed.

Selector engine

Turn authorised source fields into relationship facts with repeatable mappings instead of fragile prompt glue.

Clear sizing boundary

Use Scout up to proof, developer, and lower-load pilot scale. Around 100,000 relationship/vector records, 250,000 source events a month, high concurrent search, or any P95/P99 performance requirement, move the buyer to Fortress or Elite.

JSON task handoff

Assemble top-example JSON for a specific question, goal, entity, use case, or agent request, with provenance, confidence, caveats, and task options.

Developer console

Inspect sources, selectors, facts, and API output from a React admin surface built for engineers.

Integration points

Designed to sit inside the enterprise stack you already own

Local systems

SQLite, PostgreSQL/pgvector, SQL databases, REST APIs, GraphQL endpoints, flat files, and developer-controlled test fixtures.

Product operations registration

Register status, entitlement posture, update checks, and aggregate-only usage while operational records remain in the customer data plane.

Enterprise path

Promote proven free open-source Scout deployments into Enterprise/Fortress patterns with the proprietary Rust/LanceDB relationship engine, LanceDB vector database, private connectors, vaults, identity, and restricted-environment planning.

Evidence Results

Scout proves the Universal Context Layer before enterprise rollout.

These examples focus on the free open-source path, PostgreSQL/pgvector proof store, and governed JSON output.

KynticAI Result
Free Open Source

Scout scenario - SQL row, source event, attribution path

Prove the Universal Context Layer free

A buyer can inject authorised rows, browser events, emails, and product signals into Scout before committing to the enterprise runtime.
KynticAI Result
PostgreSQL/pgvector

Scout scenario - lower-load vector proof store

Readable proof before scale

Scout stores relationship facts, attribution paths, and vectors in PostgreSQL/pgvector so the team can inspect what the layer is doing.
KynticAI Result
JSON Output

Scout scenario - API, snapshot, relationship fact, next task

Give the model evidence it can explain

The output is not vague context. It is governed JSON with the item path, examples, confidence, caveats, and ranked task options.