The AI platform for German architecture and engineering firms.
What is not connected
has to be checked by hand.
Every building project produces thousands of documents over the years, spread across drawings, models, standards, minutes and email, with nothing holding them together.
Knowledge is not linked
Decisions, exceptions and the reasoning behind them sit in email, in minutes and in the heads of individual colleagues, not in the drawing.
Every change is manual work
When a drawing changes, someone has to work out by hand what that means for dozens of documents and standards. What gets missed often surfaces on site.
Generic AI does not know your drawings
ChatGPT and its peers are trained on text, not on geometry. A drawing with dimension chains needs mathematics, not language, plus the standards and the particulars of your project.
The answer sits between the documents.
This is the grind that costs designers time every day: opening three documents for one answer and joining them up in your head. A tool that reads each document on its own does not take that away. Neither does an AI you hand the whole stack to. Without the connections in between, it is guessing.
- Am I working with the current drawing revision or an old one?
- Does the escape route length still hold after the partition wall moved?
- Where does this variation order depart from the original contract?
- Is this extra work covered by the fee or is it a variation?
- Which rule applies when ASR and LBO contradict each other?
- Does this window satisfy ASR, LBO and the escape route width at once?
- How do I prove later that this decision was taken that way?
- Does this GAEB item still match the changed model?
- How did a colleague solve this on the last comparable project?
That connection is what we build. Every answer sourced down to the page number.
One building. One knowledge graph.
Your systems stay where they are, we connect to them. Nothing is copied, nothing is moved. We work out what ties your documents together. Everything else is built on that.
- CADDWG
Working drawings, sections, details
- Tendering and ERPGAEB
Bills of quantities, costs, variation orders
- EmailMSG
Correspondence, approvals, agreements
- BIM modelIFC
Components, spaces, quantities
- Project filingPDF
Minutes, official notices, drawing revisions
- Further data+
Any system your office already runs
Build your own agent on the structure, not just on the file store.
A SharePoint or Copilot connector only finds what is written somewhere as text. Which component belongs to which drawing, which standard applies to it, which email carries the decision: none of that is in any file. We work that connection out from your project. Your AI reaches it through an open interface, and you build the workflow yourself.
Use our ready-made checks.
Revision check, coordination check, code compliance check, site record check and whatever comes next, with an interface, without you building anything yourself. You ask the question, we show the place in the drawing that has a consequence.
Learn moreRun ABL-14
- Type
- Extract air run
- Nominal size
- 425 × 350 mm
- Air volume
- 432.5 l/s
Every drawing becomes structure.
The connection needs components. They are rarely in a database, they are on the drawing: as a line, as a dimension chain, as a label. We read the drawing itself, from PDF, scanned ones too, and turn every run, every room and every dimension into a component with values. An existing model makes us more precise. It is never a precondition.
- The drawing is the source, not just its title block: geometry, dimension chains, labels
- A 2D PDF is enough, scanned as well. The trades owe each other the drawing anyway, not the model
- Your model is a good source, not a preferred truth
- Your model does not tell you when it is out of date. We check it against what is drawn.
IFC · DWG · DXF · PDF, scanned as well
One knowledge graph. Every answer on it.
We build the knowledge graph once for your project. After that every answer stands on the same understanding, instead of starting again at the single document each time.
Why was the opening at grid line C-4 changed?
- Found the component at grid line C-4 in the drawing, compared index b and c
- Pulled every document that refers to this component
- Found the decision and its reasoning in the minutes
- Checked it against the approval in the correspondence
Gesucht nach „opening grid line C-4"
- Minutes, week 11Item 4
- Approval 14 MarM. Kowalski
- Ventilation_L2_index-c.dwggrid line C-4
The opening was reduced from 80 to 60 centimetres on 14 March at the request of the structural engineer. The reason was the bearing pressure at the corner of the downstand beam.
Decided in the site meeting in week 11, approved by email the same day. The ventilation drawing has carried the change since index c.
Why was the opening at grid line C-4 changed?
The question nobody prepared for. The answer sits in three documents that know nothing of each other, and Folaint puts them together, with the reference.
A clash is not a finding.
Clash detection sees bodies. We see meaning.
You probably already have checking software. Clash detection reliably finds where two components touch, and it should carry on doing that. What it does not find: whether the damper that moved still closes the fire compartment, whether the window is large enough under both rules, what the variation order changes in the contract. Those are questions of meaning, not of geometry. Folaint sits alongside your checking software, not in its place.
Reports that two components clash
Sees bodies
Phrases and judges in one step
A statement without a source
Maximum compliance as the only goal
Folaint enables AI to understand your project like your best employees.
We don’t train your AI. We enable it.
How can we help?
Ask us about AI-assisted design work and document research, tell us about your project, or let us show you Folaint in a demo.
Get in touch