SteelFlo vs SketchDeck LIFT: Full Comparison
SteelFlo and SketchDeck LIFT are the two AI takeoff tools built specifically for structural steel — not generalist construction takeoff with a steel checkbox. The short version: LIFT is the stronger pick if your shop runs Tekla PowerFab and bids exclusively North American AISC work; SteelFlo is the stronger pick if you want transparent pricing you can see before a sales call, a free trial you can run today, built-in estimating, or drawings from outside the US. Both put a human reviewer in the loop, which is the only honest way to run AI on bid documents.
We compared these two on connection detection specifically back in April. This post is the full picture: who each tool is for, detection scope, standards coverage, pricing, accuracy claims, and revision handling.
Who are SteelFlo and SketchDeck LIFT built for?
Both companies looked at the same problem — estimators spending 4-8 hours per bid counting members off PDFs — and built AI to do the counting. They diverge on the shop they imagined using it.
SketchDeck LIFT is built for the North American fabricator already living in Tekla PowerFab or Strumis. Its pitch is drawing-to-model: upload 2D structural drawings, get member and connection detection, and push an estimate-level 3D model into your existing MIS. If your workflow ends in PowerFab, that native export is a genuine advantage — no CSV massaging, no re-keying.
SteelFlo is built for the small-to-mid fabricator who prices work in a spreadsheet or their own rate sheet, bids whatever crosses the desk — including international drawings — and wants one tool from PDF to priced estimate. Upload a blueprint, the AI boxes every steel label on every page, you review the boxes, and you export a BOM or price it directly in the built-in estimating engine.
Neither positioning is wrong — a 40-person fabricator with a PowerFab seat has different needs than a 6-person shop where the owner does estimates at night.
Detection scope compared: members, connections, copes
Both tools detect the core of a steel takeoff. Here's the scope side by side, based on published documentation from both:
| Detection capability | SteelFlo | SketchDeck LIFT | |---|---|---| | Member labels (W, HSS, C, L, plate, pipe) | Yes | Yes | | Cold-formed / CFS sections | Yes | Not published | | Connection types (shear, moment, base plate) | Yes | Yes | | Bolt count, diameter, grade | Yes | Yes | | Cope dimensions | Yes | Yes | | Plate details (shear tabs, end plates, gussets) | Yes | Partial | | Weld specs (fillet, CJP, PJP) | Yes | Not published | | Camber and finish callouts | Yes | Camber yes | | Schedule table parsing | Yes | Yes | | 3D model generation from 2D | No | Yes | | Scanned/raster drawing handling (OCR) | Yes | Not published |
Two rows deserve emphasis.
3D model generation is LIFT's headline feature and SteelFlo doesn't do it. If an estimate-level model is part of your bid process — for connection pricing, erection sequencing, or client presentations — that's a real point for LIFT.
Raster handling matters more than most fabricators expect. Plenty of bid sets arrive as scans — old drawings, faxed addenda, flattened PDFs with no text layer. SteelFlo runs OCR on pages with a missing or unreadable text layer, so scanned sheets still produce detections. LIFT's public documentation doesn't address scanned drawings.
Every detection in SteelFlo is a bounding box on the actual drawing — 7 boxes labeled W10X15 means 7 pieces, each one visible and clickable on the sheet. There's no inferred quantity math; if it isn't boxed, it isn't counted. That transparency is what makes the review step fast: you're confirming things you can see, not auditing a black box.
How many steel standards does each support?
This is the widest gap between the two tools.
| Standard | Region | SteelFlo | SketchDeck LIFT | |---|---|---|---| | AISC | US / Canada | Yes | Yes | | BS / IS | UK / India | Yes | Not published | | AS/NZS | Australia / NZ | Yes | Not published | | EN (IPE, HEA, HEB) | Europe | Yes | Not published | | GB | China | Yes | Not published | | CFS / SSMA | US cold-formed | Yes | Not published |
SteelFlo carries 4,500+ steel sections across these 6 standards and detects the standard automatically from the drawing — you don't tell it whether the sheet is UB/UC or W-shapes; it figures that out and validates every designation against the right section database. A misread like "W12X2G" fails validation and gets flagged instead of silently entering your BOM.
LIFT's public material is AISC-centric. That's a reasonable choice for their market, but it means a Sydney fabricator with AS/NZS drawings, a UK shop mixing UC sections with European IPE imports, or a US fabricator who occasionally gets an EN-spec industrial job is outside LIFT's published coverage.
If you never see a non-AISC drawing, this section doesn't matter to you. If you see even a few per year, it's the difference between one tool and one tool plus a manual process.
Pricing and trial: $399/mo and 3 free takeoffs vs LIFT's model
| | SteelFlo | SketchDeck LIFT | |---|---|---| | Published pricing | Yes — $399/mo Pro | No — demo/sales call | | Free trial | 3 free takeoffs, no card, no call | Demo-gated | | Time to first result | Minutes after signup | After sales process | | Built-in pricing engine | Yes | No — export to PowerFab/Excel | | Export formats | CSV, highlighted PDF, priced estimate | Tekla PowerFab, Strumis, Excel |
SteelFlo publishes its price: SteelFlo Pro is $399/month, and every account gets 3 free takeoffs before paying anything. You can sign up this afternoon, upload a live bid set, and see the detections on your own drawings before talking to anyone.
LIFT follows the enterprise-sales model: book a demo, talk to sales, get a quote. Plenty of good software sells that way, but it means you can't evaluate it on your own drawings this week and can't budget for it without a call.
The pricing-engine row is the workflow difference underneath the price difference. LIFT hands its output to PowerFab or Excel for pricing. SteelFlo's detections flow into its own estimating engine — material rates, labor, hardware — so the same tool that counted the steel prices it. Shops already paying for PowerFab won't care. Shops pricing in Excel will.
Accuracy claims compared — why human verification is the honest frame
Every AI takeoff vendor publishes an accuracy number, and most of those numbers are meaningless without the fine print. Here's the fine print.
SteelFlo's published figure is 95-99% accuracy — and that is explicitly the accuracy of AI detection plus human verification combined, not raw AI output. The AI finds and boxes every label; the estimator reviews the boxes, confirms the trusted ones in bulk, and inspects the flagged ones. The 95-99% is what lands in the final BOM after that review. We wrote up the full breakdown — detection completeness, designation correctness, aggregation — in how accurate is AI for steel takeoffs.
Why frame it that way? Because raw AI accuracy on construction drawings is not high enough to bid from unreviewed, for any vendor. A single missed beam line item can eat the margin on a job; any tool that implies you can skip review is selling you risk.
To LIFT's credit, they operate the same way — AI detects, a human confirms. The practical comparison is not "whose AI is smarter" but how fast is the review step: SteelFlo triages detections into trusted (bulk-confirm) and flagged (inspect first) buckets, shows every box on the source sheet, and validates designations against its section databases so misreads surface before you price them. Judge both tools by how long review takes on your drawings — which loops back to the trial question, because you can only measure that on a tool you can actually run.
Revision handling and workflow differences
Steel bids don't end at Rev 0. Addenda land, the EOR reissues sheets, and the takeoff you did Tuesday is stale by Friday.
SteelFlo's workflow: each takeoff is a standalone run — upload the revised set, get a fresh detection pass, review, export. Because detection runs in minutes, re-running a revised set is cheap; you re-review the sheets that changed instead of re-counting the job by hand. Reviewed takeoffs and priced estimates stay in the project dashboard, so comparing Rev 1 against Rev 0 is a matter of opening both.
LIFT's workflow: revisions flow through the model — a revised drawing set updates toward a revised model, which then re-exports to PowerFab. If the model is the deliverable, this is coherent. It also means the MIS remains the system of record, with LIFT as the front end.
The deeper workflow difference is scope. LIFT is a detection-and-modeling layer that assumes a downstream MIS. SteelFlo is detect-review-price in one flow: cut list, nesting-aware order sheet, and a priced estimate off the same reviewed data. Which is better depends entirely on whether you already own the downstream tools.
FAQ
Does SketchDeck LIFT have a free trial?
Not a self-serve one. LIFT is demo-gated — you book a sales call to see it, and pricing is quoted rather than published. SteelFlo gives you 3 free takeoffs on signup with no card and no call, so you can test it on your own drawings the same day.
Which tool handles international drawings?
SteelFlo. It detects and validates sections across 6 standards — AISC, BS/IS, AS/NZS, EN, GB, and CFS/SSMA — with automatic standard detection, backed by a 4,500+ section database. LIFT's published coverage is North American/AISC. If your work includes UK, Indian, Australian, European, or Chinese drawings, SteelFlo is currently the only steel-specific AI tool that publishes support for them.
Which tool detects connections and copes?
Both. LIFT was early to AI connection detection and remains strong at it; SteelFlo detects connection types, bolts, copes, welds, and plate details, and feeds confirmed hardware into its pricing engine as line items. The connection-by-connection comparison has the full table.
Is SketchDeck LIFT or SteelFlo better for a small fabrication shop?
For most small shops, SteelFlo — the published $399/mo price, free trial, and built-in pricing engine mean no MIS dependency and no procurement process. LIFT makes more sense for shops already running Tekla PowerFab or Strumis, where its native export and 3D model output plug into an existing workflow.
What are the alternatives to both?
Beam AI (service-based, most traction), generalist tools like Bluebeam and PlanSwift (manual counting, no steel AI), and newer entrants — see our rundown of Ferra alternatives and the full steel takeoff software landscape. LIFT and SteelFlo remain the two steel-specific AI options.
The honest close: if you're a PowerFab shop bidding AISC work and want a 3D model from your 2D drawings, book the LIFT demo. If you want to know in the next hour whether AI takeoff works on your drawings — US or international — run your own drawing through SteelFlo free. 3 takeoffs on the house, no card, no call.