REASONING ENGINES

Three engines. One context. One answer.

Most enterprise questions are not answerable from one kind of data. Flexday reasons across documents, structured records and imagery inside a single solution, and cites all three in the same answer.

ENGINE

SpeciphicAsk

Retrieval and reasoning over unstructured content: contracts, policies, manuals, tickets, correspondence, archives. Built for the case where the answer is spread across nine documents and one of them contradicts the others.

  • Permission-aware retrieval that respects source-system entitlements
  • Citation to page, clause and paragraph, never an unsourced claim
  • Conflict detection across documents, surfaced rather than averaged
  • Multilingual, with translation that keeps the legal position intact
ENGINE

StructIQ

Reasoning over structured data, warehouses, ERPs, operational databases, exports. Questions become ontology-resolved queries, not guessed SQL against column names it has never seen defined.

  • Ontology-resolved query generation, not raw text-to-SQL
  • Row-level lineage on every figure returned
  • Measure and calendar definitions taken from Layer 3, so numbers reconcile
  • Refuses and escalates when the data can't support the question
ENGINE

SpeciphicView

Image and video intelligence for the physical world: inspection imagery, scanned documents, proof-of-delivery photographs, plant and line footage, and long-form video archives.

  • Neural inspection and defect classification on production lines
  • Scanned-document understanding where OCR alone fails
  • Video indexing, search and summarization across large archives
  • Confidence thresholds that route uncertain frames to a human
WHY IT MATTERS THAT THEY'RE ONE SYSTEM

“Why is this shipment still unbilled?”

Answering that properly means reading the delivery record (StructIQ), checking whether the proof-of-delivery image is legible and signed (SpeciphicView), and finding the accessorial authorization buried in an email thread (SpeciphicAsk). Three products would give you three answers and a reconciliation meeting. One solution gives you a cited answer and a next action.

StructIQ · delivery recordSpeciphicView · POD imageSpeciphicAsk · authorization email
ENGINE SPECIFICATIONFLEXDAY.AI
SPECIPHICASK
Permission-aware retrieval, clause-level citation, cross-document conflict detection, multilingual reasoning
STRUCTIQ
Ontology-resolved querying, row-level lineage, Layer 3 measure definitions, refusal on insufficient data
SPECIPHICVIEW
Defect and inspection classification, scanned-document understanding, video indexing and search, confidence routing
SHARED
One context window across engines, one citation model, one audit trail, one evaluation harness
MODELS
Claude by default; per-solution model configuration including customer-hosted endpoints

Most real questions need all three.

Tell us a question your team currently answers by opening four systems.