In-Depth Review: Allsite.ai
What It Is
A Civil 3D plugin that sends a site model to cloud processing and returns finished grading, road, and pipe network design as native Civil 3D objects. Level AI handles roads, retaining walls, lot grading, drainage, and detention. Service AI routes and sizes sanitary, storm, and water networks with clash avoidance.
Who It's For
Civil engineers and land developers doing detailed site design for subdivisions, master-planned communities, and commercial sites, who already work in Civil 3D 2025 or later on Windows.
Why It Matters
Most generative site tools stop at feasibility and massing, and hand you a concept someone then redraws in Civil 3D. Allsite.ai produces the Civil 3D objects themselves: profiles, COGO points, 3D polylines, grade breaks, a proposed surface, pipe networks, and structures. The output is detailed design work rather than a study, and it comes back into the model the engineer is already working in.
Problem Solved
Grading a site is an iteration problem. Each change to a road profile moves earthwork volumes, which moves the detention sizing, which moves the pipe inverts, and an engineer walks that loop by hand for every option a developer asks to see. Allsite.ai runs the loop as an optimization and returns the result with cut and fill quantities and a priced schedule of quantities as CSV. Trench spoil from Service AI feeds back into Level AI so grading balances against trench volumes rather than ignoring them.
Audience Fit
Primary Users:
- Civil Engineer
- Land Developer
- Site Designer
- Developer
Ideal Fit:
- Firms running Civil 3D 2025 or later on Windows
- Subdivision and master-planned work where lot yield and earthwork balance drive the deal
- Teams testing several site layouts before committing to one
- Projects where detention sizing and pipe routing have to agree with the grading
Key Differentiator
Returns construction-grade Civil 3D objects rather than a feasibility concept
Workflow Fit
Project Phases Covered:
- Feasibility
- Schematic
- Design Development
Where It Plugs In:
Installs as a Windows plugin inside Civil 3D. You prepare the model, clear the validation messages, run the job to the cloud, then load the results back as native objects. Runs take about 30 minutes on smaller jobs and more than two hours on projects over 100 acres, per the vendor.
User Experience
Speed: Cloud processing rather than local. About 30 minutes for smaller jobs and over two hours past 100 acres, per the vendor
Interface: Runs inside Civil 3D as Prepare, Review, Run, and Results, with validation messages graded as info, warning, or error
Learning Curve: Moderate. The work is in preparing a model the optimizer can solve, and the vendor's own troubleshooting docs describe over-constrained models as the usual reason a run fails
Collaboration: Single-model workflow. The vendor publishes no multi-user or shared-review features
Integrations: Autodesk Civil 3D 2025 and later, and ArcGIS Pro. Autodesk Forma and Bentley OpenRoads Designer are listed as coming soon.
Bonus Insight
The viability number decides whether a result is usable. Below 90 percent viability the tool enters what the vendor calls What-If Mode, skips the grading optimizer, and returns a surface its own documentation describes as a rough draft that may contain irregular break-lines and other inaccuracies, and advises against sending outside the office. At 90 percent and above it runs the full optimization. Two things are worth confirming before you buy. The company publishes no list of the jurisdictions, codes, or roading standards it designs to, despite saying designs meet local rules, so ask which apply to your region. And automated drawing production, the Drawing AI module, is limited to New Zealand and Australia and requires an ArcGIS Pro license. The strongest evidence that this works at scale is the Amoruso project in Roseville, California, a 400-acre community of 1,500 lots that Brookfield and the engineering firm Wood Rodgers graded with Level AI. Esri and AEC Magazine reported it independently of each other, describing 11 miles of roads, 23 detention ponds, and 650,000 cubic yards of earthworks balanced in days. The company's broader claims of 100x speed-ups and over $1 million in earthworks savings appear only on its own site. Allsite.ai was founded in 2021 by Sam Blackbourn, a chartered civil engineer, and George Stirling, splits operations between New Zealand and the United States, has disclosed no outside funding, and prices by quote, with usage metered in tokens.
Need help implementing Allsite.ai?
Our advisory team has hands-on experience across 520+ AEC tools. Book a free AI strategy session to get specific recommendations for your firm.