Vertical AI estimating, managed end to end

Your estimating logic, running as an engine.

Pinnacia builds you a Razorback engine — a private AI estimator trained on your own documents, rules, and pricing history. We deploy it, run it, and keep it accurate as a managed service.

Every client gets their own engine. Razorback is the platform we build it on, not a shared application you log into alongside your competitors.

Your own engine, not seats Human review on every bid Your logic stays portable

Razorback engine

Document-to-estimate pipeline A five-stage vertical schematic. Stage one: intake and OCR of plans, scopes, specs and photos. Stage two: extraction into structured line items. Stage three: your estimating logic and pricing structure. Stage four: a human review gate, highlighted in engine crimson, where confidence or dollar risk requires approval. Stage five: a priced, reviewable estimate leaves the engine. 01 / INTAKE + OCR Documents in PLANS · SCOPES · SPECS · PHOTOS 02 / EXTRACTION Structured line items 03 / ESTIMATING LOGIC Your rules, your pricing 04 / REVIEW GATE Human approval LOW CONFIDENCE OR HIGH VALUE 05 / OUTPUT Priced, reviewable estimate
Fig. 01 — Intake to estimate inside your Razorback engine. Crimson marks the engine running: active flow, and the review gate. Nothing leaves the engine without a named human approving it.

The bottleneck

Estimating is still a manual bottleneck

Bid volume keeps climbing. The number of people who can price work accurately does not. Everything in between gets absorbed by your most expensive, least replaceable staff.

  1. 01

    Every bid starts from scratch

    Someone opens the plans, reads the scope, cross-checks the specs, squints at site photos, and rebuilds takeoff detail that your company has already priced dozens of times. The work is repetitive, but it still requires judgment, so it lands on senior people.

  2. 02

    Two estimators, two numbers

    Assumptions live in individual heads. Waste factors, crew rates, exclusions, and how aggressive to be on a given client all shift by who picked up the file. Leadership finds out at the margin review, not at bid time.

  3. 03

    Slow turnaround costs winnable work

    Bids get declined, deprioritized, or submitted thin because the queue was full. The lost revenue never shows up in a report — it shows up as the bids you never sent.

  4. 04

    Software standardizes workflow, not judgment

    Traditional estimating platforms organize the process. They do not capture your company's judgment, your local market playbook, or the reasons your best estimator prices a specific job the way they do. That knowledge stays undocumented and walks out the door at retirement.

The strategic point

As everyone adopts the same off-the-shelf AI tools, access to AI stops being an advantage. What is left is proprietary workflow, proprietary data, and better decisions at the moment a number gets committed.

The Razorback engine

A private estimator, built around how you actually win work

Pinnacia encodes your estimating process into six connected layers. Each layer is configured to your documents, your rules, and your thresholds — not to a generic industry template. Razorback is the platform; the engine that comes out of it is built for one company only, yours.

  1. Document intake and OCR

    Plans, scopes, specs, photos, and supporting documents read in the formats you already receive.

  2. Scope extraction

    Documents converted into structured line items that match your takeoff and quoting format.

  3. Estimating logic

    Your rules, your assumptions, your pricing structure — applied the same way on every job.

  4. Exception handling

    The edge cases your best estimator already knows, written down and applied consistently.

  5. Confidence scoring

    The engine flags what it is not sure about instead of quietly guessing.

  6. Review workflow

    Human oversight routed in wherever confidence is low or dollar risk is high.

Engine layers

Six-layer Razorback engine diagram Six stacked layers labeled L1 through L6: intake, extraction, estimating logic, exceptions, confidence, and review. A crimson bracket spans the confidence and review layers, marked as human oversight. An arrow on the left shows flow from documents at the top to a released estimate at the bottom. DOCUMENTS → ESTIMATE L1 Intake + OCR L2 Scope extraction L3 Estimating logic L4 Exception handling L5 Confidence scoring L6 Review workflow OVERSIGHT
Fig. 02 — The six layers of a Razorback engine. It escalates, it does not decide. Bids are released by your team, not by the model.
Human in the loop, by design

A Razorback engine never submits a bid unsupervised. You set the confidence floor and the dollar thresholds that force a review, and every released estimate carries a named approver.

The offer

What you're actually buying

Two layers. One builds your Razorback engine around your process. The other keeps it accurate as your work, crews, costs, and clients change.

Layer 01 / Initial build

Pinnacia builds your engine around your process

  • Workflow discovery with the people who actually estimate
  • Data and document mapping across your historical bids
  • Rules capture — assumptions, waste factors, exclusions, approval logic
  • Prototype Razorback engine running on your real documents
  • Testing against jobs you have already priced
  • Deployment and staff onboarding
Layer 02 / Ongoing management

Pinnacia runs it and keeps it sharp

  • Maintenance and monitoring of the deployed workflow
  • Tuning as pricing, scopes, and crews change
  • Quality review against approved estimates
  • Support for your estimating team
  • Performance monitoring and reporting
  • Model and vendor updates handled on our side
Read this twice

Pinnacia is a long-term operating partner, not a one-time implementation vendor. The maintained Razorback engine is the product. We are not selling access to AI — we are selling a maintained operating system for how your company estimates work.

Fit

Who it's for, and who it isn't

We would rather disqualify early than build a Razorback engine on a process nobody inside the company owns.

Strong fit

  • Mid-market operators with meaningful, recurring bid volume
  • Historical estimate data that is messy but recoverable
  • Leadership willing to co-design the workflow rather than delegate it
  • An internal owner who can answer scope and pricing questions
  • Estimating capacity that is already the constraint on growth

Not a fit

  • Companies shopping for a cheap, generic bid tool
  • Teams with no historical estimate data to learn from
  • Buyers unwilling to assign an internal owner to the project
  • Organizations that want the engine to price and submit without review

Verticals Pinnacia builds for

  • 01 Commercial construction and subcontract bidders

    Plan-and-spec bid packages, invitation-to-bid volume, tight submission windows.

  • 02 Specialty and service contractors

    High-frequency scoped work where quoting speed decides who gets the job.

  • 03 Restoration and claims-adjacent estimating firms

    Photo-and-scope documentation, carrier-facing line item discipline, defensible pricing.

  • 04 Manufacturing and fabrication quoting teams

    Drawing-driven quotes, material and labor build-ups, repeat configurations.

  • 05 Distribution and wholesale bid desks

    Large multi-line RFQs, substitutions, and margin rules applied under time pressure.

Engagement

How the first engagement runs

Four stages, one deployed workflow at the end. Durations below are indicative and get firmed up in discovery.

  1. Stage 01

    Discovery

    Pinnacia sits with your estimators and leadership, walks real bids end to end, and identifies the workflow with the highest return.

    [1–2 weeks]
  2. Stage 02

    Rules capture and mapping

    Assumptions, pricing structure, exclusions, and approval thresholds get documented and mapped against your historical documents.

    [2–3 weeks]
  3. Stage 03

    Prototype and testing

    We run your engine against jobs you have already priced and tune until the output holds up to your chief estimator's review.

    [3–5 weeks]
  4. Stage 04

    Deployment, onboarding, tuning

    The workflow goes live with your team trained on the review process, then moves into continuous monitoring and tuning.

    [2–3 weeks, then ongoing]

The first deployed workflow is scoped to land inside the first 90 days. Expansion into additional workflows, divisions, or estimation domains happens after that first workflow is holding in production.

Architecture and ownership

You own the logic, not a vendor

Every Razorback engine is deliberately modular. Ingestion, orchestration, knowledge storage, business rules, output generation, and dashboards are separated, so Pinnacia can replace any single backend provider without rebuilding the business logic that makes your estimates yours.

Modular architecture and ownership stack A stack of six client-owned modules on the left, marked in Pinnacia slate blue: dashboards and reporting, output generation, business rules, knowledge storage, orchestration, and ingestion. On the right, three dashed crimson boxes represent the interchangeable backend providers inside the Razorback engine, connected to the ingestion, orchestration, and knowledge layers and labeled as replaceable without touching the business logic. CLIENT-OWNED LOGIC REPLACEABLE Dashboards + reporting Output generation Business rules Knowledge storage Orchestration Ingestion PROVIDER A PROVIDER B PROVIDER C SWAP WITHOUT REBUILDING RULES
Fig. 03 — Slate blue is yours and stays yours. Crimson is engine plumbing Pinnacia swaps out underneath you. Your rules, schemas, and process intelligence are not interchangeable.
  • Your rules stay documented and portable. Assumptions, thresholds, and pricing logic live as structured configuration you can read, export, and take with you.
  • Your data schema is yours. Line item structures, cost codes, and scope taxonomies are modeled to your business, not to a provider's proprietary format.
  • Backend providers are implementation detail. When a better or cheaper option appears, Pinnacia swaps it underneath. Your workflow does not change.
  • No black box. Every estimate your engine produces can be traced back to the document that triggered it and the rule that priced it.
  • Exit is not a hostage situation. If the relationship ends, your rules, schemas, and historical outputs go with you.

Pricing structure

One published price, and it is the first step

We publish the price of the blueprint because it is the only number we can know before we have seen your documents. Everything after it is quoted from what the blueprint finds, at a fixed price, once.

Step one
Estimating workflow blueprint

We interview your estimators, read your real bid documents, and map how work actually moves from invitation to submitted number — including the exceptions your senior people carry in their heads and have never written down.

  • A BPMN process chart of your current estimating workflow
  • Every decision point, handoff, rework loop, and bottleneck, named
  • Which stages an engine can carry and which stay with your people
  • The rules, thresholds, and pricing logic we would need to encode
  • A scoped, fixed-price proposal for the build
$4,500 fixed Credited in full against the build if you proceed within 90 days

Step two
The build

Rules capture, workflow configuration, testing against your historical bids, and deployment. Priced as one fixed number, not an hourly estimate that moves.

The range across engagements is wide, and honestly so — a two-format subcontractor and a multi-division fabricator are not the same job. The blueprint is what closes that gap before you are asked to commit to anything.

Fixed price Quoted from the blueprint

Step three
Running it

Maintenance, tuning, quality review, monitoring, support for your team, and model and vendor updates as the underlying tooling changes.

Compute and processing costs are included in the monthly figure. You get one predictable number, not a bill that moves with how many jobs you bid.

Monthly, flat Quoted alongside the build
Start with a blueprint

No seat licenses. No per-user pricing. No usage meter. You are paying for a built and maintained engine, and the monthly is what keeps it accurate.

Questions we get

Straight answers before the call

Estimating software gives everyone the same structure to work inside. It still waits for a person to read the documents, decide the scope, and apply judgment. A Razorback engine encodes that judgment — your assumptions, your pricing structure, your exceptions — so the first pass is produced for you and your estimators move to review and strategy. It can sit alongside your existing platform rather than replace it.

Historical bid packages and the estimates you produced from them, your current pricing structure or cost basis, and access to the people who make the judgment calls. It does not need to be clean. Messy but recoverable is normal, and organizing it is part of the build.

Your team does. Your engine produces a structured, priced estimate with confidence flags, and your estimator approves, adjusts, or rejects it. You set the thresholds that force review — typically a confidence floor plus a dollar-value trigger. Nothing goes out under your name without a human approving it.

Corrections are the feedback loop. When your reviewer changes a line, Pinnacia traces it back to the rule or extraction step that caused it and fixes the cause, not the symptom. That tuning is what the retainer pays for. Expect the first weeks after deployment to involve more correction than month six.

You own the inputs that matter: your data, your documented rules, your schemas, and the outputs your engine produces. Pinnacia builds, hosts, and maintains the Razorback engine itself under the engagement, and it is yours alone — engines are not shared between clients. Specific ownership and exit terms are written into the agreement so there is no ambiguity later — see the engagement terms for the current framework.

Because a free proposal would be a guess, and you would be the one carrying the risk of it being wrong. The work in a blueprint is real work — sitting with your estimators, reading your actual bid packages, and finding the judgment calls nobody has ever written down. That is what lets us quote a fixed price for the build instead of an hourly range that moves once we are inside.

The chart and the process map are yours to keep either way, whether or not you build anything with us. If you do go ahead within 90 days, the $4,500 comes off the price of the build in full.

That is expected, and it is why a Razorback engine is modular. Backend providers sit behind your business logic, not inside it. When we swap one out for cost, accuracy, or availability reasons, your workflow and rules stay where they are. You should barely notice.

The heaviest lift is discovery and rules capture, when we need real access to the person who knows how you price. Pinnacia does the documentation work. Plan on roughly [X hours/week] from an internal owner during the build, dropping to [X hours/month] once the workflow is live. You make the calls; we write them down and encode them.

Usually the highest-volume, most repetitive bid type you handle — the one where the documents look similar every time and your estimators are doing recognizable work over and over. It is rarely your most complex job. We pick it together in discovery based on volume, repeatability, and how much senior time it consumes.

Next step

Your estimating logic, running as an engine.

The Razorback engine, built and maintained by Pinnacia. A discovery call is a working conversation: your bid volume, where the queue backs up, and what the first workflow would be. If it is not a fit, we will say so on the call.

Direct

hello@pinnacia.com [confirm live inbox]

Response within [1 business day].

Seven fields, one screen. We use this to decide whether a call is worth your time, and nothing else. See the privacy policy.