Will AI replace CAD drafters? CV skills to add

What did Mach9's $34M raise actually change for drafters?

Nothing overnight — but it confirms where the money is going. Mach9, a startup that set out in 2021 to build better, faster and cheaper surveying hardware, has raised $34 million after pivoting to AI CAD software instead. That's the whole story in one line: investors decided the bottleneck in surveying wasn't the sensor, it was the weeks of human drafting that turn raw field data into drawings a client can use. If you draft, detail or survey for a living, those production hours are the target. Not your judgement, not your site visits, not your code checks. Read the raise as a price signal on your time rather than an obituary for your trade.

Hardware is capital-hungry and slow to sell; software that compresses drafting hours is neither. A survey firm's margin lives in production time, so a tool that pulls linework and features out of a point cloud and hands back a near-finished drawing more or less sells itself. That's why the pivot read as sensible to investors rather than scattershot. It also describes the market you're now competing inside: scanners, mobile mapping rigs and drones already collect far more data than any team can draw by hand, and the backlog sitting in those point clouds is exactly what automation was built to clear. Collection got cheap years ago. Drawing is what's being industrialised this cycle.

So the honest answer to "will AI replace CAD drafters" is no, not as a job title — but yes, it will take over a slice of what many drafters get paid for, and that slice is the easiest part to describe on a CV. "Produced plan and profile sheets in Civil 3D" is a sentence a software licence can now answer. That's the problem. If your CV describes only production, it reads like a cost line. If it describes decisions, standards, checks and client-facing deliverables, it reads like the person who runs the software. Same career, different document. The rest of this article is about which sentences to write.

Which drafting and surveying tasks is AI software already doing?

The tasks going first are the repeatable ones with a clear right answer: feature extraction from lidar point clouds, first-pass linework and breaklines, surface generation, sheet set-up and annotation, quantity takeoffs, clash detection, and as-built comparison against a design model. Anything that amounts to tracing a known object in captured data, or applying a company standard the same way a hundred times, is a candidate for automation. Pavement edges, kerbs, signs, poles, manhole rims, overhead lines — software recognises them and draws them in minutes rather than days. Routine detailing and library-based drafting sit in the same bucket. None of this requires a decision about what the client should build; it requires speed and consistency, which machines do well.

Script-driven and parametric work has been eating the same hours for years, which is why the shift feels familiar to anyone who's used Dynamo in Revit, Grasshopper in Rhino or Subassembly Composer in Civil 3D. Template-driven corridor models, automated sheet generation and rule-based annotation removed a lot of manual clicking long before machine learning turned up. What's new is that automation now copes with unstructured input — a messy scan of a real street instead of tidy design intent. That's a bigger leap than it sounds. The work that used to be protected by "the data's too dirty for software" is no longer protected by that argument, and anyone planning a career around dirty data should know it.

Here's the contrarian part: the 90%-done first pass is a trap for firms and an opportunity for you. Automated extraction grabs the easy geometry fast, then leaves a long tail — occluded utilities, a kerb that's actually a driveway apron, two record drawings that contradict each other, a feature code that doesn't match the client's standard. Cleaning and verifying a machine-made drawing is skilled work, and it isn't the same skill as drawing from scratch. Firms that bought the software quickly discovered they still need someone who can audit the output, defend it in review and stamp it. If that's what your week looks like, say so in those words, because it's the part nobody can licence.

What parts of surveying and drafting still need a human?

Liability, judgement and the field. Boundary determination, easement interpretation, licensed sign-off and anything ending in a seal stay human, because professional responsibility can't be assigned to software. The field side holds too: setting control, staking, verifying utility locates, deciding where the scanner needs a second set-up, and noticing that the as-built doesn't match the records. Client coordination is the other durable piece — turning a half-formed request into a scope, chasing a civil designer about a conflict, running a drawing through QA against one agency's standard. Automation produces geometry. It doesn't accept responsibility for that geometry, and the whole professional structure of surveying is built on someone doing exactly that, in writing, with their name attached.

Data-quality judgement is the underrated one. Point clouds lie in predictable ways: parked cars hide kerbs, vegetation swallows ditch lines, wet pavement scatters returns, and registration error creeps along a long corridor. Knowing which parts of an automated extraction to trust — and which to go back and measure — is a learned skill that sounds unglamorous and is worth real money. Standards work is similar. Every DOT, council and utility client has its own layering, feature codes, naming and delivery format, and reconciling machine output with that rulebook takes someone who's been burned by it before. Write it down. "Audited automated extraction against client CAD standards" beats "proficient in AutoCAD" every time.

Two more things hold. Existing conditions are messy in ways training data isn't: historic structures, partial demolition, undocumented services, hand-drawn 1970s record sets. And cross-discipline coordination is a conversation, not a file — someone decides whether a clash matters, who moves, and what gets issued. Make those the spine of your CV instead of a footnote. A useful test for every bullet you've written: could a licence key do this unattended by next year? If yes, compress it into a tools line. If no, give it a number, a client type and an outcome. Most drafters already have the right experience and the wrong document, and that gap is fixable in an afternoon.

Which skills should a drafter or surveyor put on a CV in 2026?

Lead with reality capture, automation and data skills, then back them with the licences and standards work that proves accountability. Concretely: point cloud registration and processing in Trimble Business Center, Leica Cyclone, Autodesk ReCap or CloudCompare; scan-to-BIM and automated feature extraction; UAV photogrammetry with the relevant operator certificate; mobile mapping; Civil 3D, MicroStation and OpenRoads; Revit with Dynamo; GIS work in ArcGIS Pro or QGIS; and Python for the repetitive glue nobody else wants to write. Add digital delivery — common data environments, model coordination, LOD and deliverable specifications, CAD standards management. On the survey side, state licensure plainly: surveyor-in-training, licensed, and which jurisdictions. That single line does work no keyword can.

A skills list gets skimmed. Evidence gets you called. Turn each tool into a sentence with volume, standard and outcome attached: "Processed 40 km of mobile lidar corridor; automated extraction plus manual verification cut deliverable turnaround from six weeks to nine days." "Wrote Dynamo scripts that generated 120 sheet sets to agency layering standards, saving roughly 30 hours per project." "Checked and signed off automated utility extraction across 15 intersections; caught three misclassified structures before issue." Those bullets say you run the automation rather than compete with it. Keep every number defensible — you'll be asked about it in the interview, and an inflated figure is easier to spot than people assume.

Titles matter as much as skills, because recruiters and keyword filters search for what the market calls the role. "CAD Drafter" is a shrinking search term; "Scan-to-BIM Technician", "Reality Capture Specialist", "Geospatial Data Technician", "Digital Delivery Coordinator" and "BIM Coordinator" are growing ones. You don't need to invent a title you never held — keep your real one, then add a short positioning line underneath using the language the advert uses. If you're starting from a file you last touched years ago, it's usually faster to build the CV around those keywords from scratch than to keep patching headings that no longer match the job you want.

How do you reposition your CV before the automation reaches you?

Rebuild it around judgement, standards and data, and aim it at one specific job advert instead of sending the same file everywhere. Start by deleting the production-only bullets — anything that just names software and a drawing type — and replace each with a decision you made, a standard you worked to, or a number you moved. Then put a three-line summary at the top saying what you capture, what you process and what you're accountable for. Keep the layout single-column with real text headings, because engineering and survey firms screen through applicant tracking systems that parse badly when you use two columns, icons or text boxes. Design flair wins you nothing here and costs you parsing.

Run the check before you apply, not after three rejections. Upload your current CV to HRLens and the free analysis gives you a score out of 100, five category scores — experience, tech stack, impact and ownership, clarity and structure, and ATS compatibility — your strengths, and a visual read on the layout. Paste the job description into the same analysis and it lists the skills the advert asks for that your CV doesn't show, which for drafters is almost always reality capture, scripting or standards work they do weekly and never wrote down. Check your CV free, then decide what actually needs rewriting rather than guessing.

If you've been at the same firm since before mobile mapping was routine, don't append the new role to a 2016 file — rebuild the whole document around how the work is delivered today. The paid Full Analysis at $3.99 adds the fix list, the blunt "why you might be rejected" read and your blind spots, and every analysis generates ATS-friendly rewritten CVs in six templates that mirror the job's keywords. One closing opinion: the drafters who struggle over the next three years won't be the ones whose tasks got automated, because everyone's did. They'll be the ones whose CV still describes the tasks of 2015.

Frequently asked questions