15+ construction & engineering skills
9 cross-discipline conflict categories
Every finding cites exact clause + quoted text
Isolated per-project workspaces

Workspace Templates

Upload your spec. Get actionable findings in minutes.

Each workspace template packages domain-specific skills, CSI MasterFormat reference data, and evidence-based validation for construction document workflows.

How it compares

Why AEC teams choose Rakenne over manual review and generic AI

Generic AI can summarize a spec section. It can't systematically cross-reference HVAC against Electrical across 300 pages and cite every clause.

Manual Review
Generic AI (ChatGPT)
Cross-discipline consistency
Manual Review 3-5 days per review — depends on coordinator experience
Generic AI (ChatGPT) May find keyword matches; misses semantic conflicts
Evidence quality
Manual Review Human reviewers cite correctly — when they find issues
Generic AI (ChatGPT) May hallucinate section numbers or paraphrase
Standards validation
Manual Review Rarely done systematically
Generic AI (ChatGPT) May know some standards but not withdrawal dates
Procurement traceability
Manual Review Manual extraction from Part 2 of each section
Generic AI (ChatGPT) Lists materials but misses vendor quals and submittals
Change impact tracing
Manual Review Relies on individual expertise — inconsistent coverage
Generic AI (ChatGPT) May guess at impacts without reading dependent sections
RFI generation
Manual Review 30-60 minutes per RFI — manual citation collection
Generic AI (ChatGPT) Can draft but often lacks proper spec references

See it in action

The agent finds contradictions humans miss

Skills systematically compare requirements across discipline pairs — checking voltage, materials, fire ratings, clearances, and scope boundaries. When a conflict is found, the agent cites both clauses and explains why they contradict.

Cross-discipline contradiction detection

In this example, the agent checks a 320-page project manual and finds a critical voltage mismatch between the HVAC and Electrical specifications that would have caused equipment failure during construction.

Rakenne Agent
You
Check this project manual for cross-discipline contradictions between HVAC and Electrical.
Agent
I’ve parsed 14 CSI divisions and built a cross-reference index with 47 internal references. Analyzing HVAC (Division 23) against Electrical (Division 26)…
Agent

[CRITICAL] — Voltage mismatch for AHU-1

Section 23 73 00, Paragraph 2.1.A:

“Air Handling Unit AHU-1: 480V, 3-phase, 60 Hz. Motor: 25 HP.”

Section 26 24 16, Paragraph 2.2.C, Panel Schedule MDP-M:

“Circuit 4 — AHU-1: 208V, 3-phase, 40A”

Conflict: AHU-1 requires 480V per HVAC spec; Electrical panel allocates 208V. Equipment will not operate on incorrect voltage.

Agent
Found 11 findings total: 3 Critical, 5 Warning, 3 Info. Full report written to output/spec-consistency-report.md. Want me to generate RFIs for the critical findings?

Beyond specification validation

15+ skills across construction contracts, structural engineering, and building compliance

The construction skill library covers contracts, material specifications, building codes, and regional structural standards. [Browse all infrastructure skills →](/skill-tags/infrastructure/) or start with a workspace template.

Go deeper

Tutorials and guides

Step-by-step guides showing real dialog, findings, and how skills work together for specification validation.

FAQ

Common questions from construction professionals

The agent checks 9 conflict categories systematically — voltage/power mismatches, material conflicts, fire rating inconsistencies, dimensional clearances, performance contradictions, submittal conflicts, scope boundary gaps, environmental conditions, and seismic design category alignment. A human reviewer typically focuses on their own discipline; the AI checks across all discipline pairs. That said, the agent is a first pass — it reduces the manual review burden but does not replace professional judgment. Use it as you’d use a junior coordinator: check the findings, add your expertise.

Yes. The agent detects whether section numbers follow MasterFormat 2016 (six-digit, e.g., 23 05 00) or MasterFormat 1995 (five-digit, e.g., 15820) and applies the correct discipline mapping. Many renovation projects, institutional owner master specs, and older project manuals still use the legacy numbering.

The agent uses a section-pair chunking strategy for large specs. Rather than loading 500+ pages at once, it builds a cross-reference index first, then loads and compares related section pairs. This means a 1,000-page project manual is analyzed as a series of targeted comparisons — each discipline pair checked independently. Processing time scales with the number of discipline pairs, not the total page count.

The built-in standards watchlist covers ASTM (A36, A992, A572, A615, C150, C33, E84, E119, C920), ANSI/ASHRAE (90.1, 62.1, 55, 15/34), NFPA (13, 72, 70/NEC, 101, 90A), IBC/IRC/IECC, ACI (318, 301, 530→TMS 402/602), AISC (360, 341, 303), and AWS (D1.1, D1.4). It flags withdrawn editions, missing edition years, and commonly confused standards. For standards not on the watchlist, the agent states that status could not be verified rather than guessing.

No. Your specification documents, findings, and RFIs are processed only to deliver the service and are never used to train foundation models. Each project gets an isolated workspace — no cross-tenant or cross-project access. Construction specs often contain proprietary design details and client-sensitive information; Rakenne’s architecture isolates every project.

Yes. Every skill is written in plain text — you can fork a skill, add conflict patterns, modify the standards watchlist, or adjust the RFI template to match your firm’s format. The Skill Workshop template lets you author and test custom skills interactively. You could add a check for your firm’s specific standards (e.g., a utility company’s supplementary specs) without any programming.

InspectMind focuses on drawing-side plan checking — coordinating architectural, structural, and MEP drawings for conflicts. Rakenne focuses on the specification text — the written requirements that govern what gets built. These are complementary: InspectMind catches conflicts visible in drawings; Rakenne catches conflicts in the written requirements. Document Crunch focuses on contract risk — identifying liability, insurance, and payment terms. Rakenne’s spec consistency check focuses on technical contradictions — voltage mismatches, material conflicts, fire rating inconsistencies. Again, complementary: Document Crunch asks ‘is this contractually risky?’; Rakenne asks ‘does this technically work?’

Ready to check your first specification?

Upload a project manual, run the consistency check, and see findings with evidence in minutes.

Ready to let your expertise drive the workflow?

Stop wrestling with rigid templates and generic chatbots. Describe your process, let the agent handle the rest.

Get Started Free — No Sign-Up