By admin
Subtopic 4: Existing Solution Providers
Our final episode of the series Automations I Used as an Engineer…or Wish I Used. We have covered Python, natural language processing, AI agents, and today it will be existing solution providers.
There’s a growing ecosystem of companies building AI tools specifically for civil engineers. Some offer plug-and-play platforms you can access with a subscription. Others build custom solutions tailored to the needs of your firm. This week, we’re diving into both categories and breaking down companies that fit within each.
A quick overview: we will look at how AI companies build their tools, samples of subscription-based providers and the pros and cons, and then custom providers and the pros and cons. We will wrap up on how you can choose which is right for you.
How Do AI Engineering Companies Actually Build These Tools?
When you hear that a company is using “AI,” it can feel vague — like a buzzword. So, let’s demystify the backend. Most AI engineering tools follow one of three paths:
- Leveraging Large Language Models (LLMs) Through APIs
Many tools — especially those focused on natural language understanding, document processing, or proposal writing — start by integrating existing large language models (like GPT-4, Claude, or Mistral) via API keys. This lets them “plug in” advanced capabilities like summarization, question answering, or translation.
These APIs don’t just return generic chatbot outputs. Smart companies use custom prompt engineering, vector databases, and retrieval-augmented generation (RAG) to make responses specific to their domain — like reading a concrete code clause or extracting project specs from an RFP. Some also apply fine-tuning — training the model further on firm-specific language, past reports, or drawing notes. This results in AI that feels more like a project engineer than a general assistant. - Building Custom Pipelines
When off-the-shelf LLMs aren’t enough, companies may stitch together multiple tools into a pipeline: image recognition → OCR → classification → document rewriting, for example. These pipelines are often hosted in cloud environments like AWS or GCP and built with Python, HuggingFace transformers, or tools like LangChain. They can also pull together multiple LLMs. - Training Machine Learning Models from Scratch
This is rarer — and expensive. But in areas like drawing generation, optimization, or pattern recognition, some companies develop proprietary models from the ground up. These models are trained on thousands of plans, BIM models, or estimation files to spot trends, predict clashes, or generate optimized layouts. Developer tools and libraries like TensorFlow are becoming more widespread and allowing people to generate the solutions much more frequently.
Subscription-Based, Pre-Built Solutions
These are off-the-shelf AI products, developed for wide adoption, where firms pay a monthly or annual subscription to access specific features.
Characteristics:
- Quick to deploy (cloud-based, minimal setup)
- Consistent pricing across clients
- Designed for broad use cases in engineering
- Limited customization
- Best suited for firms that want plug-and-play efficiency
Civils.ai
A research assistant built for civil engineers. It enables fast search and summarization of large technical documents like building codes, standards, and geotech reports. It is a SaaS (software as a service) subscription per user or per team with minimal setup and no customization required, making for quick deployment.
Engineers upload PDFs or access public standards. The AI instantly finds relevant clauses, figures, or tables, with clickable references. Some use cases include geotechnical engineers performing code checks, feasibility reviews, or pre-design research.
Pathw.ai
Pathw.ai is a subscription-based AI platform built specifically to assist structural engineers with steel connection design. Unlike general-purpose AI tools, Pathw.ai focuses solely on automating constructible steel connections.
Its subscription-based model makes it accessible for small and mid-sized firms without requiring custom development. They implement after a structural model is complete, import that model into their software, allow for smart connection design from a custom-built database, and ease the communication process between engineer, drafter, and fabricator.
Custom-Built AI Solutions
These firms develop tailored AI tools or automations specifically for your company’s internal workflow. They start with discovery, then build or fine-tune models and interfaces based on your actual data, processes, and goals.
Characteristics:
- Highly flexible and specific to your company
- Built for unique pain points or workflows
- Slower to deploy, but deeper integration
- Usually one-time dev cost + optional maintenance
- Ideal for firms with complex processes or innovation goals
AECforward
A Paris- and London-based innovation studio building AI “bots” to automate structural engineering and design tasks. AECforward collaborates with engineering teams to develop tools like low-carbon design assistants, FEM simulation bots, or data-mining models that pull insights from past project files. They are ideal for firms exploring sustainable design, parametric workflows, or deep drawing intelligence and typically cater to larger firms. Solutions are built collaboratively and may include R&D-style experimentation.
Struct.digital
Struct.digital is a small UK-based firm that has a couple pre-built solutions but also develops custom solutions for clients. They can automate CAD to spreadsheets, simplify Revit models to wireframe models accurately, and complete tasks using a combination of applications, such as Viktor and Grasshopper. Their small size makes them more attuned to smaller sized engineering companies.
Final Thoughts – Which Path Is Right for You?
Choosing the right solution depends on the type of work you complete — large vs. small projects, residential vs. commercial vs. industrial, etc. — how you complete that work, and your company size. Custom solutions can cost more up front, but be worthwhile in the end, while with existing subscription-based solutions, you know what you are getting; with them typically having free trials but require continual payments for access.
Sidian’s goal is to build tools that make your work more creative, more strategic, and more impactful where there is a gap, or connect you with one of our partners — not replace engineers but empower them.
You can find a deeper breakdown of these tools — including demos, pricing info, and integration guides — on the Sidian website. Let us know if you want to be introduced to any of these teams directly or have us look at your workflows and help you decide where to go. There is a plethora of options to choose from.