Estym8 DocsEstym8 Home
← All docs

Estym8: Vision, Solution & Value Proposition

Product context: Estym8 is built from scratch as an AI-first construction preconstruction platform—not a legacy takeoff stack with AI bolted on. AI runs across the product: bid-package ingestion and classification, multi-model takeoff and vision, plan intelligence, cross-file synthesis, Estee, estimate-to-submittal draft review, and optimization recommendations. Canonical framing: AI-first positioning.

Goal

Use AI to deliver incredible savings on time and money when estimating from full plan sets and bid packages (multi-trade; MEP-forward depth today), and use AI to analyze every project and estimate and make super intelligent recommendations for optimization—to save money, charge appropriately, reduce risk, and win more work. We don't just automate counting; we deliver plan intelligence (conflicts, code, RFI, optimization) that manual or simple tools can't. Every touchpoint should feel like I didn't know software could do that.

If you are new to construction bidding: read Estym8 for new readers first—it defines GCs, subs, takeoffs, plan sets, and why speed and accuracy matter—then continue here for product detail.

Built from scratch, AI everywhere

Estym8 was authored greenfield as an AI-first preconstruction platform. We did not fork legacy estimating software and sprinkle models on top—the classify → plan → takeoff → intelligence → export loop is model-native end to end (text, vision, and structured outputs across Grok, Claude, OpenAI, and supporting extractors). AI is not one feature; it is how we ingest bid folders, quantify scope, surface conflicts and code gaps, draft RFIs, power Estee, and run estimate-to-submittal draft spec review. See AI-first positioning for the layer-by-layer map.

Value proposition (crafted)

Primary (elevator):
Estym8 turns plan sets and bid packages into bid-ready multi-trade takeoffs in minutes instead of days—and flags conflicts, code issues, and missing clarity with draft RFIs and optimization recommendations—so you bid smarter, save money, and win more. (Today the deepest automation is still MEP; other trades share the same pipeline where drawings support them.)

Alternates:

  • AI-forward: "AI-powered plan-set takeoff that thinks like a senior estimator: plan intelligence, draft RFIs, and optimization recommendations so you bid smarter and win more."
  • Time + intelligence: “Fast plan-set takeoff from your PDFs and folders, plus AI plan QA and draft RFIs so you don’t leave money or clarity on the table.”
  • Risk + upside: “AI takeoff that also finds plan conflicts and code gaps and tells you when to file an RFI—with the draft written for you.”
  • One-liner: “Bid-ready multi-trade counts and materials from your plan set, with AI-driven conflicts, code checks, and draft RFIs built in.”

Solution (what we do)

  1. Fast, accurate takeoff from plan PDFs across 6+ disciplines on one folder pipeline
    AI analyzes plan sets and produces bid-ready device counts, materials, raceway/wire takeoff, and assemblies. The same classify → plan → execute loop covers electrical, mechanical, plumbing, civil, architectural, structural and adjacent trades — coverage and depth vary by sheet quality and trade. Upload plan PDF stays the simple path (one file, one full takeoff). Optional: a symbols-only pass on that same surface stores a SymbolTakeoff row with in-app PDF + detection overlays so you can verify CV counts before promoting to an estimate (PROJECT_LEVEL_AND_FOLDER_WORKFLOW.md). Upload folder (plans & docs) ingests whole bid packages: we classify every file (drawings vs spreadsheets vs supporting docs), store an AI project overview (game plan, estimate strategy, file roles), and run the execution plan on eligible sheets—including multi-file discipline bundles when the plan groups them.

  2. Verbatim "As Printed on the Sheets" harvest + code-edition mismatch alerts
    Every project that has at least one harvested PDF gets a verbatim As Printed on the Sheets pass: title block, applicable code editions, exception clauses, printed allowances, calculation rules, and cross-sheet references lifted directly from each PDF (one focused pass per document) so every claim traces back to a sheet, not to model inference. When sheets disagree on code edition (e.g. one sheet cites NEC 2017 and another cites NEC 2020, or NFPA / IBC editions don't line up), code-edition mismatch alerts appear on the project page so the user can RFI before bid day.

  3. Plan intelligence, cross-file intelligence & risk surfacing
    We don't just count. We:

    • Detect inconsistencies across the plans (e.g. schedule vs plan, sheet-to-sheet conflicts, missing refs).
    • Flag anything that may go against local codes (NEC, IBC, energy, seismic, etc.) and surface it in concerns and "things to be aware of."
    • Answer: Are there conflicts in the plans? and Should we file an RFI? When the answer is yes, we create draft RFIs for the user (subject, question, background, reference sheets) so they can send them immediately.
    • Run a post-folder cross-file intelligence pass that compares the harvested digests of every bundle in the run and surfaces consistency findings, coordination topics, and RFI seeds across disciplines — distinct from (and in addition to) the per-folder AI project overview.

  4. Optimization recommendations
    Every estimate and summary should include actionable recommendations: value engineering, where to charge more, where to save cost, buffer and risk tradeoffs, and bid strategy—so users can optimize for profit and competitiveness.

  5. Multi-discipline narrative report (Markdown / DOCX / PDF)
    A single downloadable report stitches the verbatim plan harvest, every completed takeoff, the conflict log, the draft RFIs, and a bid-bucket reconciliation that compares AI counts against engineer-printed grand totals from the schedules into one decision-ready package per project. Available as Markdown, DOCX, or PDF; produced by lib/reports/narrative-report.ts and exported from the project page.

  6. Estee (assistant)
    Public Estee on the marketing site answers product and workflow questions only—it does not read or change private estimates. Signed-in Estee stays grounded in your run: "Should I file an RFI?", conflicts, discipline filters, and the rest of the existing Q&A — including the harvest data (codes, exceptions, allowances, code-edition mismatches) and cross-file intelligence findings. For estimate-scoped chat, Estee can also propose a small set of structured updates (e.g. summary material overage) with a preview; you confirm before anything is written—so RAG and chat behavior stay, and saves stay intentional.

  7. Estimate-to-submittal bridge (in product)
    Reusable submittal catalog (products + PDFs), packages on estimates with optional links to BOM and workbook quantity rows, Markdown register and compliance matrix, merged PDF export, and draft spec AI cross-check—so the same pursuit can move from counted / priced toward GC-facing documentation without retyping everything in a separate silo.

  8. Consistent "blow their mind" experience
    From first upload to export, the product should feel like having a senior estimator plus a code and plan QA expert in their corner—powered by AI. Fast takeoff across all major disciplines, AI-surfaced concerns, verbatim harvest with code-edition alerts, AI-drafted RFIs when needed, cross-file synthesis, and a single decision-ready narrative report on demand. Users should repeatedly think: I didn't know this was possible.

How we reflect this in product

  • Master prompt & outputs:

    • Concerns list includes plan conflicts, cross-sheet inconsistencies, and potential local code violations.
    • RFI recommendation is explicit: when to file an RFI, plus a structured draft RFI (e.g. rfiSuggestions) so the user can copy or send.
    • Summary recommendations include optimization (save money, charge more, value engineering, risk vs buffer).

  • UI:

    • Concerns & risks are prominent.
    • “Request for Information (RFI)” section appears when we recommend an RFI, with draft text and reference sheets.
    • Summary/recommendations highlight optimization and next steps.
    • Estee: floating chat on project/estimate views; confirm-before-save when applying assistant-proposed estimate edits (no silent mutations).

  • Roadmap:

    • Deeper trade verticals and MEP extensions (concrete, drywall, doors/windows, air balance, HVAC recommendations, cross-discipline consistency depth)—same goal: one plan set / bid package, maximum intelligence and time savings; see product roadmap.
    • Estimate-to-submittal (catalog, packages, register/compliance exports)—bid documentation on the Estym8 side; GA-track, evolving in product today.
    • Bodi (portfolio): ICT / low-voltage project brain and BOD—complementary product; plan intelligence from Estym8 can feed Bodi’s sources/claims on shared pursuits (integration doc, investor overview).

  • RFI at plan set level:
    RFI is about the plan set, not a single estimate. We store rfiSuggestions in project plan metadata when we process the PDF, and show the “Request for Information” card on the project page (plan set level) as well as on the estimate view. Any estimate that uses that plan set shares the same RFI recommendations.

Other ideas, recommendations & suggestions

  • Single source of truth for “plan set”: Treat the project (or the Takeoff tied to one PDF) as the plan-set scope. All plan-level outputs—locale, demographics, elevation/survey, codes awareness, RFI suggestions, concerns that are plan-wide—live in plan metadata and are shown on the project page. Estimates built from that takeoff then focus on bid-specific choices (buffers, edits, versions).

  • “Plan health” score: One headline metric per project (e.g. “Plan health: Good / Review recommended / RFI suggested”) derived from concerns + missing symbols + rfiSuggestions length, so users see at a glance whether the set is clean or needs follow-up.

  • Export draft RFIs: Allow “Export RFIs” to download a PDF or Word-ready file with all draft RFIs for the plan set (subject, question, background, ref sheets) so they can attach to an email or upload to a project portal.

  • Optimization as a first-class section: Give “Optimization & recommendations” its own card or tab (from summary.recommendations + thingsToBeAwareOf), not buried in the full report, so “save money / charge more” is impossible to miss.

  • Chat: “Should I file an RFI?”: In the estimate/project chat, when the user asks “Should I file an RFI?” or “Any conflicts?”, the assistant should explicitly point to concernsList, rfiSuggestions, and plan consistency and say “Yes—here’s why” or “No—here’s what we checked.” Confirm-before-save applies when the user asks Estee to change stored estimate data (preview + explicit yes/no), not for ordinary Q&A.

  • Re-run plan QA only: Optional “Re-analyze plan set” that re-runs only the consistency/code/RFI part of the prompt (no re-count) to refresh concerns and RFI suggestions after design revisions, without regenerating the full takeoff.

Implementation details for the above (plan health, export RFIs, optimization section, Estee RFI handling, re-analyze stub, duplicate-RFI fix) are documented in the repository (docs/PLAN_SET_AND_VALUE_PROP_IMPLEMENTATION.md).

← Back to all docs