AECOSYSTEM
Studio · AI for construction documents

AI that reads your tenders, specifications and drawings, and shows its sources.

We build AI for construction documents: systems that read tenders, specifications, contracts and reports. They find the requirements and risks, and they answer your team’s questions with a citation to the exact clause.

AECOSYSTEM builds AI document intelligence for contractors, EPC firms, consultants and developers.

Answers you can check.

Project knowledge is spread across specifications, contracts, tender documents, minutes and drawings. An AI answer is only useful if your team can see where it came from. Every system we build cites the page and clause behind each answer, and says so when the documents do not contain one.

What we build.

  • 01

    Project knowledge bases

    Your PDFs, Word files and spreadsheets ingested into one searchable project memory.

  • 02

    Cited question and answer

    Plain-language questions answered with the exact source.

  • 03

    Tender analysis

    Scope, quantities, deadlines, eligibility and risks drawn out of a tender.

  • 04

    Specification and contract review

    Requirements, obligations and conflicts listed for review.

  • 05

    Report generation

    Executive, technical, compliance and risk reports drafted from the documents.

  • 06

    Document control

    Versions, permissions and audit trails for every file.

What we have built that is relevant.

AECOS Work, our project-intelligence application, ingests PDF, DOCX and XLSX files, answers with citations to the exact clause and drafts executive, technical, compliance and risk reports. It is in development, and its predecessor, AECOS Insights, processed large sets of construction documents with source references.

Who it is for.

How a project runs.

  1. Stage 01

    Discover.

    We learn how your projects, sites and teams actually work before we suggest any software.

  2. Stage 02

    Specify.

    We write down what the system must do, in plain language and in detail, so that everyone agrees on it before we build.

  3. Stage 03

    Build.

    We build in short stages that you can see and use, starting with the part that saves the most time.

  4. Stage 04

    Launch.

    We roll the software out to the people who will use it, on the devices they already carry.

  5. Stage 05

    Support.

    We keep the system running, measure how it is used and improve it as your company changes.

The technology we use.

We choose widely used, well-documented tools that we already run in our own products, so the system is dependable today and maintainable for years.

AI
  • Gemini
  • OpenAI
  • Claude
  • Docling
  • LangChain
  • pgvector
Data
  • PostgreSQL
  • pgvector
  • Python
Web
  • TypeScript
  • Next.js
  • React
  • Node.js
Hosting
  • Vercel
  • Google Cloud
  • Docker
  • On-premises on request

Frequently asked questions.

Is our data used to train public AI models?

We design every system so that your documents stay within the environment agreed with you, and we choose AI providers and settings that do not train on your data.

Can it read scanned documents and drawings?

It can read scanned text and the text and tables in drawings. Interpreting the geometry of a drawing is more limited, and we tell you where the limits are.

What happens when the answer is not in the documents?

The system says so instead of guessing.

Next sheet

Have a project in mind?

Tell us about your company, the problem and the people who will use the software. We will reply with a clear next step.