Orion feature reference

One system to discover, quantify, verify, and deliver project opportunity.

Orion combines 3-minute opportunity screening, multimodal document intelligence, specialist project analysis, deterministic calculation traces, visual evidence review, and governed deliverables in one construction-native workspace.

Request a walkthrough Explore Orion

3-Minute screening Multimodal plan intelligence Scale-calibrated takeoffs Audit-ready deliverables

Capability index

The complete customer workflow in one view.

This index shows what the user receives and which inspectable record travels with the capability. The detailed sections below explain operation, evidence, and boundary.

Orion customer-visible capability inventory
Stage	Capability	User-visible result	Evidence or state retained
Intake	Multi-PDF upload	Validated project source set	File state, hash, page count, project link
Intake	PDF merge workflow	Named merged PDF for download or ingestion	Input order and generated file
Intake	Processing-mode choice	Fast screen or full analysis	Mode, credits, approval, and status
Organize	Project Hub	Screening and full-ingestion lanes	Project/document lifecycle and actions
Discover	Query Architect	Reusable Boolean strategies	Targets, refinements, instructions, and library metadata
Discover	Master Document Filter	Governed pre-filter	Saved revision, profile hash, and segments
Discover	Connected source hubs	Attributed discovery and screening queue	Source/company/project attribution
Screen	Opportunity scoring	Pursue, review, or reject	Three component scores, reasons, signals, and pages
Ingest	Multimodal page processing	Text and visual evidence index	Page assets, chunks, components, provenance, and failures
Analyze	Project Analysis V2	Typed need-to-report workflow	Needs, evidence, measurements, warnings, and stage log
Analyze	Tenant analysis skills	Domain-specific methods	Enabled skills, dependencies, and variables
Measure	Accuracy profiles	Ballpark, takeoff, or quote-ready path	Scale, formula, assumptions, confidence, and review state
Collaborate	Project-aware chat	Contextual follow-up investigation	Project, workflow, attachments, and report action
Verify	Evidence viewer	Finding beside source	Document, page, region, image/PDF, and highlight
Verify	Methodology and calculation trace	Inspectible analysis basis	Methods, measurements, assumptions, validation, and questions
Report	Qualified Opportunity Report	Modular opportunity dossier	Sources, sections, warnings, actions, and appendix
Report	Version history	Preview and restore-as-new	Version number, parent, source message, change summary
Deliver	Engineer verification packet	Source-linked review PDF	Contents, contacts, checklist, BOM, citations, and context pages
Operate	Member dashboard	Usage and seat visibility	Credits, reservations, policy, roles, and ledger
Operate	Tenant and platform administration	Governed users and knowledge	Permissions, libraries, document state, and vector diagnostics

Intake and organize

Bring project records into a controlled workspace, select the depth of processing, and keep lifecycle state visible after upload.

Multi-PDF project intake

Upload multiple project PDFs into an existing project, create a new project, or let screened document sets be grouped from title, project-number, and filename signals.

PDF validation and duplicate-content checks
Existing, new, or generated project grouping
Per-file acceptance and processing feedback

Result, record, and boundary

User result: A registered source set with project, document, and processing identity.
Boundary: Upload acceptance is not analysis completion; Project Hub remains the lifecycle record.

Fast screening or full analysis

Choose the lighter opportunity-screening lane or the comprehensive ingestion and Project Analysis V2 path for each project-document upload.

Zero Analysis Credits for screening
Credit estimate and confirmation for full analysis
Promotion from screening to full analysis

Result, record, and boundary

User result: A deliberate processing path matched to the review decision.
Boundary: Screening prioritizes documents; full analysis is a separate promoted workflow.

Project Hub and document operations

Separate screened candidates from full-ingestion documents and manage the record through search, filters, status, activation, regrouping, rescreening, retry, promotion, and deletion.

Screening and full-ingestion lanes
Queued, running, ready, inactive, and failed states
Multi-select regrouping and controlled document actions

Result, record, and boundary

User result: A searchable pipeline record from intake through analysis readiness.
Boundary: Permanent deletion and role-gated operations require deliberate user action.

Discover and screen

Translate field knowledge into repeatable discovery logic, then explain why a candidate deserves deeper attention.

Query Architect

Build Boolean discovery strategies from selected applications, products, competitors, suggested terms, custom keywords, and reviewer instructions.

Generate, edit, refine, save, and combine strategies
Favorites, filters, sorting, and list/grid library views
Readable strategy descriptions beside the query

Result, record, and boundary

User result: A reusable and reviewable discovery strategy-not a hidden search prompt.
Boundary: Source access and returned results depend on the authorized source workflow.

Master Document Filter

Curate the tenant-wide Boolean filter applied before new discovery exports, ingestion, and screening, with revision protection and preserved context segments.

Generated baseline and validation
Protected context terms and segment preview
Revision/profile conflict protection

Result, record, and boundary

User result: A governed pre-filter shared by the approved discovery workflow.
Boundary: Edits affect new imports; previews reflect the last saved revision.

Explainable opportunity screening

Rank candidates against configured products, applications, incumbents, commercial language, design signals, exclusions, quantity evidence, and semantic page matches.

Pursue, review, or reject decision
Technical Fit, Pursuit Readiness, and Evidence Confidence
Reasons, strongest pages, candidate products, and verification requirement

Result, record, and boundary

User result: A score with visible decision drivers and source-page evidence.
Boundary: The score prioritizes review; it is not a probability of winning.

Analyze and measure

Turn the selected project into typed needs, targeted evidence retrieval, specialist analysis, calculations, and reconciled material recommendations.

Evidence-first Project Analysis V2

Resolve one authorized project snapshot, extract typed needs, retrieve evidence per need, run specialist analyses, validate results, reconcile BOM rows, and assemble the core report.

Stable source order and typed schemas
Parallel need-specific retrieval with exact fan-in
Explicit no-evidence and no-opportunity outcomes

Result, record, and boundary

User result: A structured evidence record instead of one free-form model response.
Boundary: Direct attachments must be ingested before they can support report citations.

Configurable civil analysis skills

Apply tenant-enabled skill modules for audit, scale calibration, civil quantities, order quantities, QA, and discipline-specific takeoff methods.

Roadway, earthwork, utilities, and linear features
Geogrid, erosion control, and channel linings
MSE walls, sheet piling, shoring, and concrete structures

Result, record, and boundary

User result: Domain methods composed into only the specialist roles that need them.
Boundary: The catalog is broader than any single tenant or project configuration.

Accuracy profiles and calculation engine

Support BALLPARK, TAKEOFF, and QUOTE_READY accuracy profiles with deterministic local arithmetic, scale/geometry analysis, confidence calculation, and explicit input gaps.

Source, scale, formula, units, and assumptions retained
Cross-source agreement and conflict handling
UNKNOWN and needs-input states instead of silent values

Result, record, and boundary

User result: A quantity with an inspectable basis and verification status.
Boundary: The selected profile and required human approvals remain visible.

Verify and collaborate

Inspect the physical source, watch the workflow, ask follow-up questions in project context, and separate evidence from unresolved judgment.

Citation preview and document viewer

Group evidence by document/page/region, preview source imagery or text, and open the selected evidence in a zoomable drawing/PDF viewer.

Rendered-image and original-PDF modes
Bounding-box highlights when available
Zoom, pan, magnifier, reset, and keyboard controls

Result, record, and boundary

User result: A direct route from the finding to the original source context.
Boundary: Exact coordinates depend on the indexed evidence region; page-level citations remain the base locator.

Project-aware chat and workflow selection

Ask follow-up questions in the context of an analysis-ready project, choose the interaction workflow, attach supporting files, and route report actions back into the conversation.

Automatic, interaction, collaborative, and project-analysis modes
Multiple attachments, pasted images, focus topics, and keywords
Specialist-targeted follow-up prompts

Result, record, and boundary

User result: A continuing investigation connected to the selected project and report.
Boundary: Report follow-ups return to Orion chat; they are not automatically sent emails or CRM updates.

Live stage and methodology visibility

Follow elapsed time, progress, stage status, units, methods, measurements, assumptions, validation updates, and client-facing working summaries as analysis runs.

Scope, measurement, evidence, engineering, BOM, report, and review stages
Detailed agent/stage process log
Methodology record without exposing private chain-of-thought

Result, record, and boundary

User result: A visible account of what the workflow did and where review is needed.
Boundary: The methodology record is an operational review aid, not hidden model reasoning.

Report, export, and operate

Package the result for commercial and technical review, preserve revisions, export evidence, and govern users, credits, and knowledge sources.

Qualified Opportunity Report

Render project intelligence, candidate BOM, summary, detailed opportunities, strategy, calculation trace, evidence review, needs review, and appendix as one modular report.

Workflow version, warning count, and source count
Pursuit, product-match, and quantity confidence kept distinct
Recommended actions and material-gap drill-downs

Result, record, and boundary

User result: A dossier that keeps evidence, calculation, uncertainty, and action together.
Boundary: Sections may be empty when the evidence does not support them.

Version history and deliverable exports

Preview saved reports, restore an earlier report as a new current version, copy Markdown, and export report, chat, Deep Research, or engineer-verification PDFs.

Restore-as-new version lineage
Engineer packet with contents, citations, context pages, and review checklist
PDF and structured output paths

Result, record, and boundary

User result: A shareable review package with source context and preserved revision history.
Boundary: Export content follows the active report/export path; temporary version preview is not itself a revision.

Member, usage, and platform administration

Track credit balance and reservations, operator seats, usage policy, member roles, tenant libraries, document activation, access requests, and vector-health diagnostics.

Owner, Tenant Admin, Operator, and Viewer roles
Separate tenant-owner and platform-admin authority
Product and engineering-manual knowledge libraries

Result, record, and boundary

User result: An operating layer for the people, budget, and knowledge behind the workflow.
Boundary: Platform administration remains separately gated from tenant ownership.

Interface examples

Four places where inspectibility changes the work.

Illustrative product views show the interaction model: author repeatable discovery logic, understand why a project was screened, open the physical source beside a finding, and inspect the quantity formula and approval state.

Orion Query Architect showing selected applications and products beside generated discovery strategies. — Illustrative strategy-authoring view. Visible names, terms, and selections are demo content rather than a customer configuration.

Field knowledge becomes repeatable discovery logic.

Query Architect turns application, product, competitor, and commercial-language context into reviewable strategies that teams can save, refine, and reuse.

Screening explains the rank.

Review component scores, decision drivers, signals, cited pages, and the recommended promotion path before spending full-analysis credits.

TECHNICAL FIT94.0

PURSUIT READINESS88.5

EVIDENCE CONFIDENCE92.0

Applicable product and application language located.
Commercial flexibility signal appears in the source set.
Page-locatable evidence improves actionability.

RECOMMENDED NEXT STEPReview the strongest cited pages, then promote for full analysis.

Illustrative interface and scores-not a customer result.

ORIGINAL REQUIREMENTPavement stabilization detail

The source identifies a reinforcement layer and base condition for qualified technical review.

Current materialNamed system in source

Review pathPotential alternative evaluation

SURFACEBASESUBGRADE

TYPICAL PAVEMENT DETAILIllustrative source page

Evidence stays beside the claim.

Move from a report finding into the rendered page or original PDF, inspect the highlighted region, and keep unresolved technical judgment visible.

Quantities show their basis.

Keep source, scale, dimensions, formula, allowance, units, confidence, and review state available to the estimator or technical reviewer.

SourceSchedule of values · Item 31-14

Documented area6,895 SY×Configured allowance1.05=Order quantity7,240 SY

Formula6,895 × 1.05 = 7,239.75 → 7,240 SYReview state Allowance requires approval

Illustrative values only. Verification status and accuracy profile control how a quantity may be used.

Input and output contract

From approved source material to governed deliverables.

Orion’s value comes from keeping what entered the workflow, what was derived, and what still requires approval visible in the same system.

APPROVED INPUTS

Project and tenant context

Plans, specifications, addenda, and geotechnical records
Schedules, details, tables, and quantity sources
Product catalogs, engineering manuals, and approved-equal context
Qualification thresholds, exclusions, accuracy profile, and review rules

REVIEWED OUTPUTS

Evidence and action artifacts

Screening decision and Qualified Opportunity Report
Scale, measurement, BOM, and calculation traces
Visual citations, evidence review, and engineer packet
Approved API, webhook, MCP, JSON, CSV, and PDF handoffs

Features FAQ

Questions about modes, quantities, methodology, and outputs.

The feature set is broad, but each capability still has an explicit processing state, evidence record, and user responsibility.

Is fast screening the same as full analysis?

No. Screening is a lighter, zero-Analysis-Credit prioritization lane that produces decision scores, reasons, signals, and evidence pages. Selected documents can then be promoted into full multimodal ingestion and Project Analysis V2 with the normal estimate, permissions, and confirmation path.

What can the quantity engine show?

Depending on the selected accuracy profile and available evidence, Orion can show documented quantities, independently measured quantities, formula inputs, scale basis, pixel trace, assumptions, conflict checks, confidence methodology, and needs-input questions. BALLPARK, TAKEOFF, and QUOTE_READY have different rigor and review requirements.

Does Query Architect search sources by itself?

Query Architect authors and manages the discovery strategy. Approved source connectors or source hubs execute that strategy within their own authorization, company/project scope, sync, filter, and batching rules.

What does the methodology log contain?

It contains client-visible methods, measurements, assumptions, validation activity, stage status, progress, and concise working summaries. It is not a transcript of private model chain-of-thought.

Which deliverables can users take out of Orion?

The current product supports Markdown copy, qualified-report PDFs, engineer verification packets, Deep Research/report/chat PDF paths, downloadable source attachments, and documented JSON/CSV/PDF integration outputs for approved production workflows.

Are all analysis skills active for every customer?

No. The checked-in catalog defines available skill modules and dependencies. Each tenant enables the relevant disciplines, estimating variables, and specialist roles for its products and operating method.

See Orion in context

See the full feature set against your project workflow.

Bring a representative plan set, your product and engineering context, and the outputs your commercial and technical reviewers need to trust.

Request a walkthrough Read the technical reference