For many architects and BIM managers, the treasure trove of past project drawings often feels more like a messy attic than a handy library of lessons. Critical design knowledge is commonly buried in decades-old project folders and scattered PDFs, making it hard to reuse past insights. According to a Metropolis magazine article, even seasoned architects admit that
“it’s hard to find information on particularly what works and what doesn’t”
in previous project documents.
This article explains how Tektome’s KnowledgeBuilder uses AI to turn your piles of old drawings and files into a searchable knowledge base, enabling faster and smarter reuse of design know-how. It automatically processes all your past unstructured data in drawings and transforms them into structured, usable knowledge. In doing so, it pulls out the insights hiding in those files, such as common quality mistakes to avoid, standard design solutions worth reusing, and the hard-won “experienced know-how” of veteran staff. The result is a living repository of design intelligence that every team member can tap into.
Using Natural-Language Prompts to Extract Custom Attributes
One of KnowledgeBuilder’s strongest features is extracting custom attributes from drawings using plain-English prompts. No manual tagging or scripts – just say, for example, “extract drawing title, date, scale and purpose,” and the AI saves those as structured data in the database. With simple natural-language instructions you can structure data from plans, elevations, sketches and PDFs exactly as you need. Architects and BIM managers can define fields like drawing type, issue date, phase, scale, author – without code. Non-technical teammates can set up the rules too, so everyone contributes to a well-organised knowledge base.
Extracted attributes are stored as structured data, building a rich, searchable index of the firm’s assets. Crucially, the AI is layout-agnostic: the same prompt works across different title blocks and formats. This goes far beyond traditional OCR. Rather than pulling raw text from fixed positions, KnowledgeBuilder interprets the intent of your prompt. Ask for a “project completion date” and it will find it in a header, footer or paragraph alike – where older methods fail if the layout changes. You spend less time tweaking templates or hunting through folders.
In short, KnowledgeBuilder converts drawing archives into a structured design database you control: architects configure what to extract as needs evolve, while the system automates the heavy lifting so you can focus on using the information.
Retrieving Drawing-Specific Insights Across Design Stages
With your past drawings now enriched with searchable data, KnowledgeBuilder makes it easy to retrieve insights at any stage of a new project. Different phases of design have different information needs – here’s how structured knowledge reuse helps during each:
Early Stage – Finding Similar Projects
In the early conceptual stages, you might be gathering references and lessons from previous jobs. KnowledgeBuilder lets you filter and find similar projects by key attributes like location, building type/usage, or completion date.
For example, if you’re planning a new office in London, you can quickly pull up all past projects with attributes of offices in similar regions. Because the system has extracted attributes like project location, function, and finish year, you can identify relevant precedents in seconds.
This helps you spot patterns and applicable solutions from the past. Teams can proactively learn what worked (and what didn’t) on comparable projects before design gets too far. In fact, by learning from past data, common issues can be spotted early – KnowledgeBuilder easily surfaces documented challenges and veteran insights to mitigate risks on current projects.
Early-stage design benefit greatly from this “look-back”: you avoid past mistakes and don’t have to reinvent standard design elements from scratch when the answers might already exist in your archives.
Mid Stage – Pinpointing Precise References
As the project moves into detailed design and documentation, the questions get more granular. At this mid stage, KnowledgeBuilder’s structured metadata enables fine-grained searching within and across projects.
You can zoom in on exactly the reference you need. Say you recall a particular facade detail or a clever HVAC layout from a past project – using the indexed attributes, you can filter that project’s documents by drawing name, drawing type, or scale to find the precise sheet in seconds.
For instance, you might filter for all “Section” drawings in Project X to find a certain cross-sectional composition of the building. Or you might search within a single project for drawings labelled as “For Construction” to review final design solutions, versus earlier schematic sketches. Because KnowledgeBuilder captured properties like drawing purpose, issue stage, and scale, you can slice through a huge drawing set with ease. This is incredibly useful for design development and coordination, when you need quick access to specific technical solutions. It saves you from manually combing through hundreds of PDFs.
Moreover, since the data extraction is done uniformly, you can even perform cross-project queries – e.g. filter all steel connection detail drawings across multiple past projects – if you’re looking for best practices. By reusing prior details and approaches that worked well, teams can prevent reinvention of solutions that the firm has already mastered once. Mid-stage design becomes more efficient because you’re building on a foundation of proven knowledge.
Later Stage – Semantic Search for Specific Details
In later project stages (or even during maintenance and post-occupancy analysis), you often need to retrieve highly specific information that isn’t neatly filed under a standard attribute. This is where KnowledgeBuilder’s semantic search capability shines. Because it’s powered by AI, you can search the content of drawings and documents in natural language, not just via predefined tags.
For example, you could ask the system to find “all instances of a particular HVAC equipment model used in our projects” or “conference room layouts with movable partition walls.” The AI will scan through the textual content of drawings, annotations, and even scanned markups to find relevant matches, even if the exact keywords differ. This goes beyond simple keyword search – it understands your query’s intent.
If you search for a concept like “plant room ventilation design,” you might discover a detail in a 2015 project where a clever venting solution was annotated on a drawing. Or you could search across projects for a term like “rainwater harvesting tank” and get results from various drawings and reports that mention it.
Essentially, you can retrieve very granular, technical insights (MEP equipment names, room configurations, code compliance notes, etc.) across your entire project library by simply asking in plain language. This semantic search is powerful for quality assurance and learning in late stages – we have a whole article written about this powerful feature here.
All this means that in later stages of design, even highly detailed veteran knowledge (say a senior engineer’s handwritten red-line notes on a print) becomes accessible and actionable to the wider team.
Summary
By structuring your unstructured drawing data, KnowledgeBuilder turns your firm’s archive into a goldmine of reusable design intelligence. The benefits to architects and BIM managers are significant. Better knowledge transfer is one major advantage – newer team members can quickly learn from past projects, closing the experience gap. Teams also make fewer repeated mistakes because common issues and their solutions are readily available for review. When a potential pitfall comes up, there’s a good chance someone in the firm encountered it before – and with KnowledgeBuilder, you can find out what they did about it.
Additionally, the firm avoids duplicating effort: why redraw a standard detail or re-research code interpretations when you can prevent reinvention by retrieving what already exists? Perhaps the most tangible day-to-day benefit is the faster retrieval of design assets. What used to take hours of rummaging through old folders can now take minutes or seconds with a smart search.
To summarise, KnowledgeBuilder helps architects and BIM managers work smarter: valuable design knowledge becomes easier to find, learn from, and apply, resulting in more informed design decisions and higher-quality outcomes on current projects. The days of critical insights gathering dust in the archive are over – your past drawings can now actively inform your future designs.
Book a demo with us today to see how KnowledgeBuilder can help you work smarter and more efficiently.
