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 moves 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, importance bands, confidence, caveats, and ranked task options for the question being asked.

Explain

LLM task brief

Fortress sends JSON to your approved model boundary. The proposed Elite path shows the executive walkthrough from safe discovery to scoped pilot and outcome review.

Product / KynticAI Scout

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

Scout is the free, open-source entry point to KynticAI Context Engine: 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.

Sales path / Scout proof

The Scout demo question: which next task follows this enquiry?

Scout should sell itself in one walkthrough. The buyer brings one enquiry, sees the relationship path, inspects the JSON, and understands why Fortress becomes the next commercial step when the proof needs scale.

Question: this person emailed us, looked at the pricing page, then read integrations. What should the sales team do next?

1. Data in

Scout starts with a proof-sized set of authorised records.

email = maya.chen@example.invalid

event = inbound_email_enquiry

web_path = pricing -> integrations -> security

crm_stage = new_opportunity

outcomes_available = converted, registered, no_response

2. Relationship path

The enquiry becomes a journey the buyer can inspect.

identity = email + cookie + CRM lead

path = email_enquiry -> pricing -> integrations -> security

matched_examples = similar technical-buyer journeys

caveat = procurement owner not yet known

3. JSON handoff

Scout returns a small task brief instead of a dashboard.

top_example = converted technical enquiry

option_1 = send integration proof

option_2 = ask user to register account

review = confirm buyer role

Output

Illustrative LLM task brief from Scout JSON

Next task: send the integration proof and security note before asking for budget.

Registration path: offer a trial workspace only after the buyer confirms their role.

Review note: procurement owner is missing, so keep proposal language out of the first reply.

Scout makes the first proof tangible: one email enquiry becomes a ranked sales action with source trail, caveat, and next-step options.

Data in, data out

Scout shows the simple version: business data in, next-task JSON out.

A buyer should be able to understand Scout in one minute. It takes approved data from one workflow, joins the pieces into a relationship path, then returns a JSON brief a person or model can use.

In

Approved records enter

Start with a small set of permitted CRM rows, emails, web events, support tickets, product signals, or outcomes.

Map

Scout links the journey

Scout connects each signal to the account, customer, case, product, or event it belongs to.

Find

Similar paths are compared

The system looks for previous journeys that had a similar pattern and a known outcome.

Out

A task brief comes out

The output is inspectable JSON: what happened, which examples matter, what to do next, and which caveats need review.

Example input

Sales enquiry proof

email enquiry from buyer@example.com

visited pricing page and integration page

CRM status = open opportunity

previous similar enquiries with converted and lost outcomes

Example output

object = buyer enquiry

matched_path = enquiry -> pricing -> integration interest

next_task = send technical follow-up with integration proof

caveat = confirm procurement owner before proposal

Buyer result

The buyer sees Scout turn scattered source records into a concrete next sales task without needing a heavy enterprise deployment first.

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 or the proposed Elite route becomes a natural next conversation 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 Context Engine 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 approved AI-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 | priority = high

option_2 = ask user to register account | priority = medium

Fortress handoff = JSON to approved model boundary

proposedEliteRoute = executive walkthrough where approved

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 Context Engine 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, human decision, or AI 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

Context Engine turns source-traced relationships into governed JSON.

These examples focus on Scout proof, Fortress private runtime, proposed Elite executive walkthroughs, and governed JSON output.

KynticAI Result
Free Open Source

Scout scenario - SQL row, source event, attribution path

Prove Context Engine 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.