# How an Architecture Firm Replaced Rate Intuition With a Shared Spreadsheet in One Quarter

**By Dan Cumberland** · Published May 10, 2026 · Categories: AI Strategy

> No, AI will not replace architects.  The American Institute of Architects (AIA), NCARB, and academic experts at Yale and the Royal Institute of British...

## Will Architecture Be Replaced by AI?  The Short Answer

No, AI will not replace architects\.  The American Institute of Architects \(AIA\), NCARB, and academic experts at Yale and the Royal Institute of British Architects \(RIBA\) all agree: AI augments architectural practice but cannot hold the seal, the liability, or the human judgment that licensure requires\.

NCARB— which regulates architect licensure in all U\.S\. jurisdictions— put it plainly in its January 2024 position statement[1](/blog/blog-will-architecture-be-replaced-by-ai#ref-1) through architect Dity Ayalon, AIA:

> "AI will not replace architects, but architects who use AI will replace those who don't\."

That same statement establishes the doctrine that matters most to anyone running a firm\.  Architects retain "responsible control" over all AI\-assisted work and remain accountable for any technical submission under their seal[1](/blog/blog-will-architecture-be-replaced-by-ai#ref-1), regardless of how much underlying labor an AI tool produced\.  AIA's March 2025 study[2](/blog/blog-will-architecture-be-replaced-by-ai#ref-2) frames the same conclusion at the industry level\.

The role is shifting— that part is honest\.  But the question every principal is asking right now— "will architecture be replaced by AI?"— is the wrong question\.

## The Better Question Every Architecture Principal Should Be Asking

The question that actually matters inside a firm is which decisions should still run on the principal's intuition, and which should now run on data\.

Most "will AI replace architects?" articles stay in the design lane\.  That's the wrong lane\.  The leverage question for a firm in the $5M to $30M range is operational: pricing, utilization, scope, change\-order capture\.  These are the decisions that determine whether a firm runs at typical margins or hits the 21\.4% top\-quartile mark[3](/blog/blog-will-architecture-be-replaced-by-ai#ref-3), and they're almost always made on intuition\.

**The leverage isn't in the design tools\.  It's in the decisions\.**

AI doesn't have to design a building to change a firm's economics\.  It just has to surface the patterns in the firm's own project history\.  Average net billings per employee at U\.S\. architecture firms ran $143,000 in 2023[4](/blog/blog-will-architecture-be-replaced-by-ai#ref-4)\.  A 22\-person firm sitting on twenty years of project data is sitting on a pricing model it can't yet see\.  A [decision framework for founders weighing AI investments](/blog/ai-decision-framework-founders) starts here\.

The risk isn't replacement\.  It's getting out\-competed by firms that moved operational decisions onto data while you were still arguing about whether AI could draw a building\.

## The Shared Spreadsheet Story— How One Firm Replaced Rate Intuition

A representative engagement looks like this\.  A mid\-sized commercial\-interiors firm— in the range of 22 staff and roughly $6M in net billings, anchored against the AIA per\-employee benchmark[4](/blog/blog-will-architecture-be-replaced-by-ai#ref-4)— was losing somewhere between 8 and 15 percent on every fixed\-fee project\.  Within one quarter, the principal replaced their pricing intuition with a shared spreadsheet of the firm's last 30 projects, and the next round of fee proposals priced with discipline for the first time in the firm's history\.

This is a composite drawn from how the pattern plays out across founder\-led professional services firms in this revenue range\.  No invented client\.  No borrowed numbers\.  Just the shape of the work\.

The firm wasn't underpricing\.  It was pricing without a memory\.  Twenty years of intuition lived inside one principal's head, and every project re\-litigated assumptions that should have been settled long ago\.  You can't read the label from inside the bottle\.

**The setup\.**  The firm bills predominantly via stipulated\-sum fee[5](/blog/blog-will-architecture-be-replaced-by-ai#ref-5)— the contract type 92% of architecture firms use, where the price is fixed and any blown\-budget hours come out of margin\.  The principal couldn't articulate why small tenant\-improvement projects systematically bled margin\.  They just knew it in their gut\.

**The spreadsheet\.**  They pulled the last 30 projects out of Deltek Vantagepoint and a side QuickBooks file\.  Six columns:

- Phase \(programming, schematic, design development, construction documents, construction admin\)
- Fee type \(stipulated sum, hourly, percentage of construction cost\)
- Planned vs\. actual hours per phase
- Number of scope changes during the project
- Final realized margin
- Square\-footage class

No new software\.  No consultants\.  Just data the firm had already paid to collect, finally in one view\.

**The AI layer\.**  ChatGPT and Claude both read the dataset and surfaced patterns the principal couldn't see\.  One emerged within an hour: tenant\-improvement work under 8,000 square feet ran roughly 22% over budget on construction administration, every time, regardless of client\.  Not magic\.  Structured pattern recognition the human eye misses across 30 projects\.  In practical terms, this is AI as intellectual augmentation— surfacing what the principal half\-knew but had never been able to prove\.

**The principal's reaction\.**  Disbelief, then validation\.  Most patterns matched gut hunches the principal had never trusted enough to act on\.  The goal wasn't to eliminate intuition\.  It was to make intuition legible\.

```html-table
<table><thead><tr><th>The Old Pricing Conversation</th><th>The New Pricing Conversation</th></tr></thead><tbody><tr><td>"This feels like a $180K project."</td><td>"Last 6 comparable projects landed at $14/SF + a 1.3x multiplier for tenant-finish complexity."</td></tr><tr><td>"We always lose money on small TIs."</td><td>"Sub-8,000 SF TIs run 22% over on CA.  Price for it or pass on it."</td></tr><tr><td>"I'll know the right fee when I see the scope."</td><td>"Stipulated-sum projects with >3 scope-change touchpoints lose ~12% margin."</td></tr><tr><td>"Trust me, I've been doing this twenty years."</td><td>"The spreadsheet says trust the spreadsheet."</td></tr></tbody></table>
```

The firm moved from "what does this feel like?" to "$/SF plus complexity multiplier," anchored in phase\-by\-phase data from comparable past projects\.  Within one quarter— the next round of fee proposals— fees were higher on the right projects, lower on the wrong ones, and the principal stopped re\-debating pricing on every call\.  The pricing model became the firm's source of truth\.

## What AI Can and Cannot Do Inside an Architecture Firm

AI can compress drafting, code research, schedule generation, marketing copy, transcription, and increasingly, design iteration\.  It cannot hold the architect's seal, carry professional liability, or sit across from a worried client at week six of a renovation\.

The day\-to\-day usage data shows the division\.  79% of architects use chatbots, 45% use grammar and text analytics, and 25% use transcription or meeting assistants[6](/blog/blog-will-architecture-be-replaced-by-ai#ref-6)\.  Generative AI— the kind of model behind chatbots and image generators— has settled into the text\-heavy parts of the job first\.  The higher\-stakes design and judgment work has not\.

On the design side, the firms moving fastest are doing specific things\.  Gensler is using AI to reduce administrative load and free designers for early\-stage concept work\.  Zaha Hadid Architects is using AI\-based simulation to shorten design feedback cycles from days to hours[7](/blog/blog-will-architecture-be-replaced-by-ai#ref-7)\.  Both are augmentation patterns\.

But here's what AI cannot do, drawn from Yale's faculty[8](/blog/blog-will-architecture-be-replaced-by-ai#ref-8) and NCARB's responsible\-control doctrine[1](/blog/blog-will-architecture-be-replaced-by-ai#ref-1): hold the seal, carry liability, build trust with a hesitant client, resolve stakeholder conflict, or make value judgments about how people will inhabit a space\.

```html-table
<table><thead><tr><th>AI Replaces or Compresses</th><th>AI Cannot Replace</th></tr></thead><tbody><tr><td>Drafting and basic CAD work</td><td>Holding the architect's seal</td></tr><tr><td>Code research and compliance lookups</td><td>Carrying professional liability</td></tr><tr><td>Schedule generation, RFI responses</td><td>Building trust with a hesitant client</td></tr><tr><td>Marketing copy, proposal boilerplate</td><td>Resolving stakeholder conflict</td></tr><tr><td>Meeting transcription, note synthesis</td><td>Judging how a space will be inhabited</td></tr><tr><td>Early-stage design iteration</td><td>Standing in front of a worried client at week six</td></tr></tbody></table>
```

Honest acknowledgment: junior\-architect tasks are compressing fast\.  Firms that don't restructure how they train and bill at the entry level will feel the pressure first\.

## The Adoption Curve Is Steeper Than It Looks

The two big industry surveys tell different stories that are both true\.  AIA's March 2025 study found 8% of firms have formally implemented AI and another 20% are working on it[2](/blog/blog-will-architecture-be-replaced-by-ai#ref-2)\.  Deltek's May 2025 study, drawing on data from nearly 700 architecture and engineering firms[3](/blog/blog-will-architecture-be-replaced-by-ai#ref-3), found 53% of A&E firms now use AI tools— up from 38% the prior year\.

Both numbers are accurate\.  AIA measures formal firm\-level initiatives; Deltek measures any AI\-tool usage\.  The gap between 8% and 53% is the story— the difference between "we have a strategy" and "people are already using ChatGPT to draft proposal language whether you've talked about it or not\."

A few stats worth sitting with:

- **Adoption velocity:** 38% to 53% in a single year[3](/blog/blog-will-architecture-be-replaced-by-ai#ref-3)\.  If the trajectory holds, the industry crosses majority adoption inside 18 months\.
- **Economic backdrop:** A&E operating margin hit a 10\-year high of 21\.4%[3](/blog/blog-will-architecture-be-replaced-by-ai#ref-3) in the same year AI usage climbed 15 points\.  Causation isn't proven, but the firms running ahead of the curve are pulling away\.
- **Age skew:** 67% of architects under 35 use AI image generators; 41% of those over 50 do[6](/blog/blog-will-architecture-be-replaced-by-ai#ref-6)\.  The principals making firm\-wide tech decisions are often the slowest to use the tools themselves\.
- **Optimism:** 84% of architects believe AI can save time on manual tasks[2](/blog/blog-will-architecture-be-replaced-by-ai#ref-2)\.  The bottleneck isn't belief\.  It's operational structure\.
- **Concerns:** 90% of architects flag accuracy, security, and data transparency as concerns[2](/blog/blog-will-architecture-be-replaced-by-ai#ref-2)\.  Real concerns— worth budgeting for the [hidden costs of AI projects that show up after the demo](/blog/hidden-costs-ai-projects)\.

About a third of firms are already running AI in day\-to\-day work[4](/blog/blog-will-architecture-be-replaced-by-ai#ref-4)— the experimentation is happening inside your firm whether the principal blesses it or not\.  A shared spreadsheet of the firm's own project history sidesteps most of those concerns: the data is yours, the pattern surfacing is auditable, and the architect stays in the loop on every interpretation\.

## Three Operational Moves to Make This Quarter

Three operational moves can be made by any architecture firm this quarter, without licensing new software, hiring a consultant, or touching a single drawing\.

The data already exists in your firm\.  Project files, time sheets, fee proposals, change orders\.  AI's job is to surface what's already in your history that no one has had time to look at\.

**1\. Pull your last 30 projects into a single spreadsheet\.**

Phase, fee type, planned vs\. actual hours per phase, scope\-change count, final margin, square\-footage class\.  Use whatever your firm already has— Deltek Vantagepoint, Monograph, BQE Core, even QuickBooks plus Smartsheet\.  Without it, AI has nothing useful to surface\.

**2\. Use a chatbot to interrogate the dataset\.**

Upload the spreadsheet to ChatGPT or Claude and ask the questions you've never had time to answer: "Where am I systematically wrong about hours?"  "Which project types bleed margin?"  "What's my actual cost\-to\-completion ratio by phase?"  The principal already knows the answers in their gut\.  The AI gives them the receipts\.  This is the "intuition becomes legible" move— and it matters because AI's biggest economic impact is on knowledge work[9](/blog/blog-will-architecture-be-replaced-by-ai#ref-9), and pricing decisions are knowledge work\.

**3\. Re\-price the next three proposals using the surfaced patterns\.**

Use a $/SF plus complexity\-multiplier framework anchored in the data\.  Track which proposals win and at what margin\.  One quarter of feedback loops is enough to validate the model or kill it\.  This is where you start [measuring whether your AI investments are actually working](/blog/measuring-ai-success)— before the next budget cycle, not after\.

If you can't pull your last 30 projects into a spreadsheet this week, the problem isn't AI\.  The problem is you don't yet have a firm that can run on data\.

## FAQ— Will Architecture Be Replaced by AI?

A few more questions architecture firms keep asking— answered briefly, with sources\.

### Will AI replace architects?

No\.  AIA, NCARB, and academic experts at Yale agree AI augments architects but cannot hold the seal, carry liability, or stand in for the standard of care that licensure requires[1](/blog/blog-will-architecture-be-replaced-by-ai#ref-1)[2](/blog/blog-will-architecture-be-replaced-by-ai#ref-2)[8](/blog/blog-will-architecture-be-replaced-by-ai#ref-8)\.  The risk is being out\-competed by architects who use AI on their highest\-leverage decisions\.

### How many architecture firms use AI right now?

About 53% of architecture and engineering firms now use AI tools, up from 38% the year before[3](/blog/blog-will-architecture-be-replaced-by-ai#ref-3)\.  Within architecture specifically, 8% have formally implemented AI and another 20% are working on it[2](/blog/blog-will-architecture-be-replaced-by-ai#ref-2)\.  Both numbers are climbing\.

### What does AI actually do in an architecture firm today?

Most usage is text\-based\.  79% of architects use chatbots, 45% use grammar and text analytics, and 25% use transcription or meeting assistants[6](/blog/blog-will-architecture-be-replaced-by-ai#ref-6)\.  Design\-side use is growing but still concentrated in larger firms like Gensler and Zaha Hadid Architects[7](/blog/blog-will-architecture-be-replaced-by-ai#ref-7)\.

### Are architects still legally responsible if AI does the work?

Yes\.  NCARB's January 2024 position statement is explicit: architects retain "responsible control" over all AI\-assisted work and remain fully accountable for any technical submission under their seal[1](/blog/blog-will-architecture-be-replaced-by-ai#ref-1)\.  The seal carries the same weight whether the underlying drafting was done by a human, an AI, or both\.

### Where is the fastest AI win for a small architecture firm?

Operations— pricing, utilization tracking, and scope management\.  The data already exists in the firm's project history; AI's job is to surface patterns the principal hasn't had time to see\.  The payback shows up in the next quarter's P&L, not five years from now\.

## The Architect's Real Job, Now and Next

AI will not replace architects\.  It will replace the architect's intuition about pricing, utilization, and scope— the part of the job most architects would happily hand off\.

What's left is the part that was always the real work\.  The seal\.  The judgment\.  The relationships\.  The conversation with a client at week six who has just realized they hate the layout\.  These get more valuable as the operational layer compresses, not less\.  Firms that move operations onto data this year are pulling away from firms still pricing on gut feel, and the gap compounds quarterly\.

If pulling your last 30 projects into a structured view feels like the right next move but not a lift your team can take on this quarter, that's the kind of work [Dan Cumberland Labs is built to help with— mapping where AI actually moves the needle in a founder\-led firm](/services/ai-strategy) without buying tools you don't need\.  This is the same conversation behind [whether to bring in a fractional AI officer or a one\-time consultant](/blog/what-is-a-fractional-ai-officer)\.

Domain expertise paired with AI is what produces the firms that pull away\.  Domain expertise alone is no longer enough\.

**AI amplifies the architect's value\.  It doesn't replace it\.**

## References

1. National Council of Architectural Registration Boards \(NCARB\), "AI as an Enhancer for the Architect, Not a Replacement" \(January 2024\) — [https://www\.ncarb\.org/blog/ai\-enhancer\-the\-architect\-not\-a\-replacement](https://www.ncarb.org/blog/ai-enhancer-the-architect-not-a-replacement)
2. American Institute of Architects, "New Research Explores Perceptions and Opportunities of Artificial Intelligence in Architecture" \(March 2025\) — [https://www\.aia\.org/about\-aia/press/new\-research\-explores\-perceptions\-and\-opportunities\-artificial\-intelligence](https://www.aia.org/about-aia/press/new-research-explores-perceptions-and-opportunities-artificial-intelligence)
3. Deltek, "What the 46th Annual Deltek Clarity AE Study Reveals About the Architecture and Engineering Industry" \(May 2025\) — [https://www\.deltek\.com/en/about/media\-center/press\-releases/2025/what\-the\-46th\-annual\-deltek\-clarity\-ae\-study\-reveals\-about\-the\-industry](https://www.deltek.com/en/about/media-center/press-releases/2025/what-the-46th-annual-deltek-clarity-ae-study-reveals-about-the-industry)
4. American Institute of Architects, "AIA Firm Survey Report 2024" \(November 2024\) — [https://www\.aia\.org/resource\-center/aia\-firm\-survey\-report\-2024](https://www.aia.org/resource-center/aia-firm-survey-report-2024)
5. American Institute of Architects via Neumann Monson, "Architecture Fees: Hourly vs\. Percentage vs\. Fixed" \(2024\) — [https://neumannmonson\.com/blog/architecture\-fees\-hourly\-percentage\-fixed](https://neumannmonson.com/blog/architecture-fees-hourly-percentage-fixed)
6. American Institute of Architects, "Architects are excited about the potential of AI, but concerns abound" \(2025\) — [https://www\.aia\.org/aia\-architect/article/architects\-are\-excited\-about\-potential\-ai\-concerns\-abound](https://www.aia.org/aia-architect/article/architects-are-excited-about-potential-ai-concerns-abound)
7. ArchDaily, "How Top Firms See AI Shaping Architecture's Workflows" \(2025\) — [https://www\.archdaily\.com/1036357/how\-top\-firms\-see\-ai\-shaping\-architectures\-workflows](https://www.archdaily.com/1036357/how-top-firms-see-ai-shaping-architectures-workflows)
8. Yale News, "How might AI affect architects? A Yale expert weighs in" \(April 2025\) — [https://news\.yale\.edu/2025/04/23/how\-might\-ai\-affect\-architects\-yale\-expert\-weighs](https://news.yale.edu/2025/04/23/how-might-ai-affect-architects-yale-expert-weighs)
9. McKinsey & Company, "The economic potential of generative AI: The next productivity frontier" \(June 2023\) — [https://www\.mckinsey\.com/capabilities/tech\-and\-ai/our\-insights/the\-economic\-potential\-of\-generative\-ai\-the\-next\-productivity\-frontier](https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier)


---

Source: https://dancumberlandlabs.com/blog/will-architecture-be-replaced-by-ai/