Architecture firms forget every project the day it finishes.

The detail that caused an RFI goes back into the library unfixed, and turns up on the next job with the same problem attached.

In today’s episode of Bricks & Bytes, Ross Wagner from Pirros joined us to unpack why it keeps happening, and what an AI project hub actually changes.

What we learnt:

* Why firms forget every project the day it finishes, and how a bad detail carries its RFI onto the next job

* What Pirros does as an AI project hub, and how its AI agent Mira works alongside designers inside Revit

* Why quality review belongs in the model as well as the drawing set

* The shift in construction tech upstream into design, and the move from point tools towards connected platforms

* Lessons from PlanGrid on hiring, building an AEC sales team and go to market through community

Full episode live on Youtube and Spotify. Link in the comments!

#aec #construction #constructiontech #ai #bricksandbytes

Our Sponsors:

This episode is brought to you by our friends at Miter. If you haven’t heard of Miter, you’re missing out. Today, 2 of every 100 construction workers in the U.S. are paid through their AI-powered platform that connects payroll, HR, field operations, and job costing. And that number is growing fast.

Miter is on a mission to defeat what they call “the Frankenstack”, the tangle of 1980s software, spreadsheets, and paper that every construction back office is familiar with.

Learn why thousands of contractors are running everything from prevailing wage payroll to time tracking to per diems on Miter at miter.com (M I T E R .com)

https://www.miter.com/?utm_source=bricks-and-bytes&utm_medium=podcast&utm_campaign=bb-sponsorship-q4-2026&utm_content=bb-podcast

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

Chapters

0:00 Teaser

0:38 Intro: Ross Wagner, Head of Strategy at Pirros

1:05 Quickfire: are contractors taking control of design?

2:53 Why Ross joined Pirros after PlanGrid and Autodesk

4:06 How to stop knowledge leaving your firm

4:40 What is Pirros? The AI project hub for architects

5:30 Why AEC firms forget every project the day it finishes

7:38 Sponsors

10:43 The biggest problems inside architecture firms

12:20 Startup vs Autodesk: the “say no with pleasure” advice

14:46 Builders vs optimisers: architect and engineer co-founders

16:17 How to measure the ROI of AI in design

18:31 Why AI shouldn’t design construction details

20:15 Mira: an AI agent that works inside Revit

21:55 What a Head of Strategy actually does

22:51 How to know if your strategy is working

25:00 Competitive moat in the age of AI: why context beats code

27:12 How to hire great people (the “no jerks” rule)

28:44 Growing 5x and lessons from PlanGrid

29:56 What’s the end game: acquisition or IPO?

31:02 Construction tech trends: the shift upstream to design

31:57 Platforms vs point solutions

32:46 The three eras of AI: chat, copilot and coworker

33:46 Is AI drawing review enough? Drawings vs BIM models

36:53 How to build a construction tech sales team

37:50 Hiring your first salesperson

38:34 Advice for founders: start with the boring workflow

41:11 Doug Leone’s startup orchestra lesson

41:37 Go-to-market that works in AEC: community

42:43 Software fatigue and change management

44:06 Lessons from Autodesk vs startups

45:49 What’s next for Pirros: quality review and MCP

This week's briefing covers three things that matter more than the headlines suggest.

First, America. The Architecture Billings Index came in at 44.7 in September, down from 47.2 in August and the largest share of firms reporting falling billings since May. Signed design contracts fell to 43.7. Enquiries are still positive at 53.2, which means clients are asking but not committing. AIA describes the index as a leading indicator for non-residential construction nine to twelve months out. What architects are not signing now is work you will not be bidding next autumn. On the other side, JLL puts 317 million square feet of industrial space under construction in the US, up 28.7% on a year ago and the most since early 2024. Same country, same week, two completely different economies.

Then the knowledge problem. There is a chief estimator somewhere who is 63 years old, 30 years in the job, and retires in 18 months. None of what he knows is written down. This week Owen sat down with Zaid Khan and Bogdan Arsintescu, who helped build the Google Knowledge Graph in 2011, to answer one question. What actually is a knowledge graph and why does construction need one. The answer starts with a three-hundred-page spec, an addendum in a different folder, and a door schedule that was never updated. A chatbot finds the spec and answers with confidence. A knowledge graph finds the conflict between all three and cites the clause. The pitch stripped of the buzzwords is this. Become the project historian. Keep the context when the people leave.

Then the Bricks and Bytes Council. Last Tuesday Owen sat in a room with leaders from some of the largest design, engineering and contracting businesses in the UK and the US. Everything is confidential but five themes came through. Design teams are materially smaller than a decade ago for the same work. Contractors are now running AI across drawing sets before they go out, so designers are building the same checks into their own process to catch problems first. Door hardware schedules and lighting schedules, handed to the most junior person in the office, are driving more RFIs than structure on some jobs. Architecture may be going the way of book publishing, more buildings, more architects, a profession hollowing out in the middle. The line that came up more than once from more than one person in the room: we did not even need a tool to solve that. We just had to change the way we think.

This week's episode is sponsored by Buildots, the construction intelligence platform turning site data into actionable insights for over 100 of the largest owners and contractors. Find out how to join them at https://buildots.com/.

Drop your answers to this week's questions in the comments of this week's LinkedIn post.

Twelve months ago, Zaid Kahn said the words "knowledge graph" to us in a Boston office.

We nodded. We had no idea what he meant.

So we brought him back, with Bogdan Arsintescu, who sat on the original Google Knowledge Graph team and later ran graph systems at LinkedIn and Microsoft.

The ask was simple. Explain it like we're five.

• A knowledge graph is a set of facts. Subject, predicate, object. "Door 105 has a 45 minute fire rating."

• An ontology is the rulebook. Everyone agreeing what every symbol on a drawing sheet means before anyone reads it.

• A chatbot context window is a drawing table. If the addendum is in the filing cabinet, it doesn't exist.

Bogdan's line of the episode: "Knowledge Graph is just the bullet points."

We still don't fully know what ontology means. But we're closer.

#constructiontech #ai #aec #preconstruction #bricksandbytesOur Sponsors:Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.coArchdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

Chapters

00:00 Teaser

01:07 What Is a Knowledge Graph? (Explained Like You're Five)

04:10 Knowledge Graph vs Ontology: The Difference Explained

07:56 Sponsor

10:57 Why Construction Needs Knowledge Graphs (Spec vs Drawing vs Addendum)

14:58 Why ChatGPT Can't Do This: Context Windows Explained

19:11 Knowledge Graphs vs LLMs: Why You Need a Vertical AI

26:43 Preconstruction Risk: Why Bids Go Wrong

29:53 The Future of AI in Construction

32:10 How Neuron Factory Works With Suffolk and GCs