Vertallax is purpose-engineered for commercial GCs — a bitemporal, event-driven intelligence platform built around a canonical operating model that compresses institutional knowledge into daily operational clarity.
All data — structured, semi-structured, unstructured, external — flows through a single import spine into a canonical operating model that surfaces awareness, enables decisions, and captures execution. The loop closes through continuous learning.
Three interlocking architectural primitives give Vertallax its structural integrity. Together they ensure that every piece of data — regardless of origin, type, or timing — connects to a single, consistent operating model.
118 registered event types flow through a single temporal spine. Every state change — whether an Opportunity stage transition, a Task completion, or a Project milestone — writes a typed event record with actor, timestamp, and full payload. Nothing is deleted; history is immutable.
The spine is the single source of truth for everything the firm has ever done, believed, or decided. Pipeline Replay reconstructs any past state by scrubbing the event log.
118 Event Types Immutable LogA single event structure captures any object type. An Opportunity state change, a ResourceCapacity update, and a PriceTransaction all resolve to the same event schema — with a discriminator field for the object type and a JSONB payload for type-specific data.
This means the spine never needs to be extended for new object types — only the payload schema changes. The query interface is uniform across the entire domain model.
JSONB Payload Typed DiscriminatorA canonical object layer — Company, Person, Project, Opportunity, Resource — anchors all enrichment, activity, and intelligence to stable, normalized identifiers. Every import, integration, and AI response resolves to a canonical record.
Canonical objects survive source system changes. A Procore project and an imported bid tab both resolve to the same canonical Project ID, enabling cross-domain analytics without duplication.
UUID Primary Keys Cross-domain IdentityEvery external data source — regardless of format, structure, or origin — passes through the same six-step import pipeline into the canonical schema. Data doesn't just arrive; it transforms, validates, and becomes operational intelligence.
Live API connections with change-data capture (CDC) replication. Delta sync on each run — only changed records are pulled. Conflict-free merge on write with versioned audit trail. Supports Procore, Sage, Viewpoint, and custom REST endpoints. Schema mapping is firm-specific and persisted.
Upload-based ingestion with intelligent header detection and column mapping. PriceMappingRules handle bid tab variations across different owners and estimators. Tabular data is normalized to BossCostConcepts and benchmarked against ConceptBenchmarks on load.
NLP-based entity extraction identifies canonical references (Company, Person, Project) within free-text documents. Extracted entities link to canonical records; the original document is stored with provenance. Enables search and retrieval across unstructured operational history.
External feeds are normalized and linked to canonical Company and Person records. LinkedIn enrichment via Proxycurl surfaces relationship depth and firm affiliations. Economic indices feed into Price Intelligence benchmarks. All external data is timestamped for temporal attribution.
The Vertallax data model is not a snapshot — it is a live digital twin of your firm's operational reality, structured hierarchically from raw inputs through derived intelligence, with full temporal precision and version history at every layer.
Every record carries two timestamps: valid time (when the fact was true in the real world) and transaction time (when it was recorded in the system). This enables exact reconstruction of past states from any future vantage point — including retroactive corrections that preserve the original record.
The platform maintains a continuously-updated digital representation of your firm — every pursuit, project, resource, relationship, and price signal is modeled as a live object with current state and full history. The twin is queryable at any past moment via Pipeline Replay, and projectable forward via the Intelligence Engine.
Every record carries a version counter. Updates increment the version and preserve the prior state in the event log. Corrections are recorded as new transactions against the original valid-time period. This means the system always knows what was believed at any point in time — not just what is believed now.
Vertallax is built to work alongside the systems you already run — not to replace them. Preconstruction, project management, and accounting platforms stay where they are; Vertallax creates a synchronized digital twin around them.
Each enterprise system has its own data model, timing, and level of API maturity. We treat them that way. Some integrations are deep and event-driven, some are scheduled API syncs, and some are lighter reference updates — all designed around how that system actually works.
The goal is not forced uniformity, but trustworthy interoperability between tools that were never designed to talk to each other.
Wherever possible we integrate through existing APIs and SDKs rather than brittle, one‑off data jobs. That means respecting rate limits, pagination, and change‑data capture semantics so we only move what changed, when it changed.
Behind the scenes, a dedicated integration spine tracks last‑sync state, schema mappings, and error handling so that operational systems remain the source of record for execution.
The Vertallax twin does not assume instant truth. Different systems publish on different clocks, and we model that explicitly: every imported change carries both valid time and transaction time so we know what was true, when, and according to which system.
Robust infrastructure guards data and system integrity while keeping the twin in step with reality, even when updates arrive minutes or hours apart.
Vertallax does not try to be your precon platform, project management system, or accounting ledger. Those systems are optimized for execution. Our job is to ingest, align, and model what they know so leaders can see the whole firm at once.
Where Vertallax goes deep is in pipeline tracking and customer tracking. Opportunities, pursuits, contacts, and relationships all resolve to canonical records so that every system’s view of a client rolls up into one coherent picture.
AI means everything and nothing. Every vendor claims it. Most mean one thing — a language model wrapped in a chat interface. Vertallax takes a different position: AI is a category of tools, not a single product feature. We deploy whichever type creates the most value for the specific problem in front of us. Sometimes that’s a large language model. Sometimes it’s a deterministic rule engine. Sometimes it’s a natural-language interface. Sometimes it’s a great workflow that gets out of the way.
Digest and synthesize unstructured data at scale. Used in Vertallax for document ingestion, market signal interpretation, and contextual summarization. Powered by Anthropic Claude.
Translate plain English into structured queries and actions across firm-scoped data. The Just Ask layer removes the need to navigate forms, menus, and reports just to reach an answer.
Autonomous tool loops that execute multi-step workflows, surface anomalies, and trigger actions without waiting to be asked. Not a chatbot. An operator working across your firm’s canonical model.
Go/No-Go scoring, operating posture, margin gates, and capacity thresholds. Rules-based intelligence informed by firm history and ML signal — deterministic where determinism is right.
Task-specific ML trained on domain data — win probability, pricing benchmarks, capacity forecasting. These models compound in value as firm data accumulates over time.
Sometimes the right answer is a well-structured formula, not a neural network. We know the difference. We use both, and we choose based on reliability, auditability, and operational value.
Natural language access to the entire firm’s operational data via the
Just Ask interface at
/admin/ask/.
Ask about pipeline status, resource availability, project health, and
pricing benchmarks — get precise, sourced answers from live firm data.
The AI doesn’t search once — it reasons across a loop of tools: opportunity lookup, stage query, project status, resource capacity, price intelligence, pipeline summary, and firm context. Multi-step inference, not single-pass retrieval.
Describe a new pursuit in plain English and the system populates
canonical fields via the
get_opportunity_defaults()
cascade — stage, owner, lead type, budget range, and market sector —
sourced from firm history and the canonical operating model.
Every AI response is grounded in structured records — not hallucinated summaries. Answers reference canonical IDs, field values, and timestamps. The AskLog model captures every query, tool call, and response for audit and analytics.
The Intelligence Engine surfaces Go/No-Go signals, at-risk pursuits, expiring opportunities, resource overload warnings, and margin anomalies — presented as actionable intelligence in Just Ask and the Daily Plumb feed.
Every decision, correction, and enrichment captured becomes signal for the next query. The longer a firm uses Vertallax, the sharper its intelligence — because the context is your firm’s operational history, not generic benchmarks.
Anthropic, OpenAI, or models running behind your firewall — Vertallax is not locked to a single provider. The model serves the firm, not the other way around.
AI is only one part of the system. Sometimes the highest-value move is a clearer workflow, a stronger rule, or a better decision surface. The point is not to use AI everywhere — it is to apply the right kind of intelligence where it creates real operational leverage.
An ergonomic, intuitive interface is not cosmetic — it is a core design imperative. The goal is to make complex operational systems feel obvious, connected, and low-friction, even when the machinery underneath is doing difficult work on the user's behalf.
The interface should feel legible at first glance. Clear hierarchy, familiar patterns, disciplined typography, and purposeful iconography reduce cognitive load so the user can focus on the work, not on how to operate the software.
Simplicity is not the removal of capability. It is the careful shaping of complexity so that the next action feels natural.
The UI should not feel like a collection of disconnected modules. Pipeline, customer, project, resource, and intelligence workflows should resolve into one continuous operating picture, with each screen reinforcing the same canonical model underneath.
When the model is coherent, the interface can feel calm — because the user is never being asked to mentally stitch the business together.
The system should help the user know what matters next. AI is not just a chat layer; it is a prompting layer — surfacing missing context, highlighting likely actions, and reducing the need for users to hunt through the interface for what they should do.
A good enterprise interface does not wait passively. It assists, suggests, and clarifies without becoming intrusive.
The visible layer should feel composed and direct: fewer decisions, clearer pathways, stronger defaults, and less clutter. Icons, layout, and language all work together to make repeated daily use feel fast, calm, and reliable.
Much of the system's success is in what the user does not have to see: reconciliation logic, temporal alignment, identity resolution, exception handling, integration tracking, and context assembly across multiple enterprise systems. The interface feels simple precisely because the complexity has been absorbed below the surface.
Vertallax tracks four distinct dimensions of firm activity — not just what happened operationally, but how the platform itself is being used, how data quality evolves over time, and the health of every connected integration.
Every user action that modifies operational data writes to the activity log. The DeliverableActivity and TaskTransaction models capture who did what, when, against which canonical record. Activity feeds are surfaced in the Opportunity Intelligence Panel and Pursuit Hub.
The Firm Health & Engagement score (0–100) measures how actively and completely a firm is using the platform. Sub-scores across time-to-assign, task on-time rate, deliverable completion, and opportunity data staleness roll up to a single adoption health indicator.
Every connected data source has a tracked integration state: last-sync timestamp, sync health score, record count, error log, and drift detection. Integration tracking surfaces in the admin dashboard and triggers alerts when source systems change schema or fail to sync.
Every canonical record carries a Data Completeness score — computed from required fields, enrichment depth, and freshness. Staleness flags trigger when records haven't been updated within the expected window. Completeness reporting surfaces gaps for remediation.
The Optimization Toolkit is a set of decision-support modules that surface improvement opportunities across the firm's core operational dimensions. It draws on the bitemporal event store and the canonical operating model to identify patterns, inefficiencies, and untapped leverage points.
Unlike static reporting, the Toolkit is prescriptive — it tells you not just what happened, but what to change and why.
Data Hound runs at firm onboarding and on-demand thereafter. It performs a comprehensive assessment of all available data — connected integrations, uploaded files, manually entered records — and produces a Data Readiness Report that tells you exactly what you have, what's missing, and what to do next.
It's how Vertallax turns a blank firm into an operational intelligence platform from day one.
Firm-level isolation is structural — enforced at the ORM layer, not the application layer. No query runs without passing through FirmScopedManager. No cross-firm data is possible by design.
Every database query filters through FirmScopedManager at the Django ORM level. There is no application-layer bypass, no raw SQL risk — firm boundaries are enforced at the query construction layer, not at the presentation layer.
Enforced at Query Layer118 registered event types capture every state change with actor identity, timestamp, and full payload. Nothing is deleted from the event spine — only superseded. Full reconstruction from any historical point is always possible.
Immutable Event LogThe Just Ask AI operates with firm-scoped read-only tools. The AI cannot query across firm boundaries. All tool calls are logged to the AskLog audit model. API keys are stored as environment variables — never hardcoded.
Tool-Scoped AccessGeneric platforms are built for everyone, which means they're optimized for no one. Vertallax is built exclusively for commercial general contractors — which means the terminology is already right, the integrations are already mapped, and the risk model reflects how GC firms actually operate. You're not starting from a blank canvas. You're starting from a running position.
Every generic platform promises customization. What they deliver is configuration — a blank data model you spend six months teaching your language, your workflow, and your risk thresholds. By the time you go live, you've rebuilt half the platform yourself. And when your firm changes, you start over.
Vertallax doesn't need to be taught what a pursuit is. Or what Go/No-Go means. Or why bid due dates matter. That knowledge is already in the model.
Industry specialization means your firm doesn't start from zero. Preloaded terminology. Pre-mapped integrations. A risk spine calibrated to how commercial GCs actually win and lose work. Industry presets that give the platform Day 1 value before your firm has entered a single record.
The longer you run Vertallax, the more it learns about your firm specifically. But it starts informed — not blank.
Pursuit. Bid Due. Go/No-Go. Buyout. Closeout. Precon Hours. The platform already speaks GC. No translation layer between your language and the system's language — and no custom field sprint before your first report runs.
Go/No-Go scoring, operating posture, margin gates, and capacity thresholds are structural — built into the data model, not bolted on as custom fields. The platform understands risk because it was designed around the decisions that create it.
Procore. Sage. Viewpoint. ProjectMark. Mapped and normalized — not negotiated per implementation. Your enterprise systems connect to a canonical model that already knows what each field means and where it belongs.
Your firm starts from a running position, not a blank slate. Sector benchmarks, pricing reference points, and scoring baselines are seeded from industry patterns — so the platform is useful before your data accumulates.
Compounding Over TimeSectors, delivery methods, trade codes, and project types are already structured and named correctly. No configuration sprint before you can run a meaningful report. No debate about what "sector" means in your system vs. ours.
Every pursuit, huddle, and field note makes the next decision cheaper to get right. The ML layer compounds firm-specific history into sharper scoring and tighter benchmarks. The longer you run Vertallax, the harder you are to compete against.
ML-PoweredVertical focus is not a market limitation — it is the source of the product's value. A platform that knows your industry doesn't need to be taught your problems. It arrives with them already modeled, already named, and already wired into the decisions that matter. That is what purpose-built means. Not configured. Not adapted. Built.
See how Vertallax compresses decades of institutional intelligence into day-one operational clarity.
Coming Soon!