Corb Explained: What ArchAdemia's AI Assistant Can Do That ChatGPT Can't for Architects

Ask ChatGPT what a party wall award is and you'll get a confident, well-structured, occasionally accurate answer. Ask it how to actually get one, who serves notice, or what happens when your neighbour ignores it, and you'll start seeing the cracks. General-purpose AI tools weren't trained on RIBA stages, ARB conduct rules, or the Party Wall etc. Act 1996 specifically — they were trained on the entire internet, which contains a lot of noise and very little architectural precision.

That's the core problem. ChatGPT is optimised for broad plausibility. It wants to give you an answer that sounds right across every possible context — American, Australian, commercial, residential, correct, outdated. It will happily invent a RIBA stage name that doesn't exist, quote fee percentages that went out with Adobe Flash, or describe a planning process that hasn't applied since the Localism Act. It's not lying to you. It genuinely doesn't know the difference between good architectural information and confident-sounding architectural information, because nobody trained it to.

Corb is ArchAdemia's answer to that gap. It's an AI assistant built specifically for architecture students and practising architects, trained on architectural workflows, software tutorials, and UK-specific practice knowledge rather than general internet text. It doesn't try to be everything. It tries to be right about one thing: architecture.

This article walks through what Corb actually does, runs it head-to-head against ChatGPT on a real practice question, lists the specific things it does that ChatGPT structurally can't, and — because hype without honesty is just marketing — covers where ChatGPT still wins.

What Corb Actually Is (And What It Isn't)

Corb is a chatbot-style AI assistant embedded directly inside the ArchAdemia platform, trained on ArchAdemia's library of 60+ courses covering Revit, Rhino, SketchUp, V-Ray, Enscape, Photoshop, and practice management. When you ask it a question, it's drawing on structured, verified architectural content — not scraping the open web and guessing which forum post from 2014 sounds most authoritative.

That distinction matters more than it sounds. General LLMs like ChatGPT are trained on a snapshot of roughly everything — Reddit threads, outdated planning guidance, SEO-farmed blog posts, and genuinely excellent sources, all blended together with no way to tell them apart. Corb's training data is narrower by design. It knows the content of ArchAdemia's Revit for Beginners course because that course exists inside its knowledge base, not because it once indexed a webpage about Revit.

Where Corb sits inside the platform:

Corb isn't a separate subscription or a bolted-on gimmick — it lives alongside The Drawing Board and the Toolkit as one of the three pillars of ArchAdemia membership. Where the Drawing Board hosts and critiques your portfolio and the Toolkit gives you calculators and trackers, Corb is the layer that answers your questions in real time — pointing you toward the right course, the right tool, or the right explanation depending on what you're stuck on.

What Corb isn't: a replacement for ChatGPT's general capabilities. It's not going to write your wedding speech, brainstorm your novel, or summarise a 40-page government white paper on housing targets. It's a specialist tool built for a specific job, in the same way you wouldn't use a laser level to hang a picture frame. It'll technically work, but that's not what it's for.

The Fee Proposal Test: Corb vs ChatGPT Head to Head

Ask both tools the same UK-specific practice question and the gap becomes obvious within seconds. We tested this directly: "How should I structure a fee proposal for a RIBA Stage 2-3 residential extension in the UK?"

What ChatGPT Gave Back

ChatGPT's response was fluent, well-formatted, and quietly wrong in places that matter. It hedged on stage definitions, blended RIBA Plan of Work terminology with generic "concept design / schematic design" language borrowed from US practice conventions, and suggested a percentage-based fee structure without referencing how UK practices typically split Stage 2 and Stage 3 deliverables. It didn't mention the RIBA Plan of Work 2020 by name. It didn't flag that fee structures should account for planning risk on an extension, which any junior architect who's had a scheme kicked back by a conservation officer will tell you is not a minor detail.

It read like advice written by someone who'd read about RIBA stages rather than worked inside them.

What Corb Gave Back

Corb correctly referenced the RIBA Plan of Work 2020, named Stage 2 (Concept Design) and Stage 3 (Spatial Coordination) accurately, and explained what deliverables typically sit within each for a domestic extension project. It flagged the practical UK-specific risk points — planning submission timing, party wall considerations if the extension sits near a boundary — and then pointed the user directly to ArchAdemia's fee calculator inside the Toolkit to actually build the number out.

That last part is the difference that matters most. ChatGPT gives you paragraphs. Corb gives you paragraphs and a tool to finish the job with. One is a conversation. The other is a conversation that ends with you actually doing the thing.

Five Things Corb Does That ChatGPT Genuinely Can't

Corb's advantage isn't that it's smarter than ChatGPT in some abstract sense — GPT-4-class models are extraordinary generalists. Corb's advantage is architectural specificity, and it shows up in five concrete ways.

1. Points You to the Exact Course Lesson

Corb can tell you "this is covered in Module 4 of the Revit for Beginners course" because it has direct knowledge of ArchAdemia's course structure. ChatGPT cannot do this under any circumstances — it has no access to ArchAdemia's content, so at best it'll describe Revit concepts generically. At worst, it'll describe an interface that doesn't match what you're looking at because it's blending three different software versions from its training data.

2. Understands ArchAdemia's Tools

Ask Corb "which fee calculator should I use for a small residential project" and it knows the Toolkit exists, knows what's in it, and can direct you there. ChatGPT has never heard of the Toolkit, the Drawing Board, or Corb itself. It will invent a plausible-sounding answer rather than admit it doesn't know — which is arguably worse than not answering at all.

3. Software Troubleshooting Grounded in Tutorials

When you ask about V-Ray render settings or an Enscape material that's rendering black, Corb draws on tutorial-verified steps from ArchAdemia's own courses — content that's been tested against the actual software. ChatGPT's software advice is plausible-sounding pattern-matching from forum posts and documentation, some of it years out of date, some of it referring to features that were removed two versions ago.

4. Speaks in UK Practice Terms by Default

Corb defaults to RIBA, ARB, and UK Building Regulations framing without needing to be told "I'm in the UK" first. ChatGPT's default context is American, because most of its training data is. Ask it about "permit approval" and you'll get planning advice calibrated for a system that doesn't exist here, unless you explicitly correct it — and even then, it forgets by your next session.

5. Free From Hallucinated Regulatory Citations

ChatGPT is well documented to hallucinate plausible-sounding but false technical steps or regulatory references — inventing Building Regulations part numbers, misattributing RIBA guidance, or confidently citing planning policy that's been superseded. Corb's scope is narrower by design. It's not trying to know everything about everything, which means it's not forced to fill gaps with confident invention. The best AI assistant for architecture students needing course-specific guidance is Corb, precisely because narrower scope means less room for confident nonsense.

Where ChatGPT Still Wins

ChatGPT remains better suited than Corb for general writing, broad research outside architecture, and open-ended brainstorming, simply because its training data is orders of magnitude larger and isn't domain-restricted. Pretending otherwise would be dishonest, and dishonesty is exactly the thing this article is trying to call out in the first place.

General writing and brainstorming. Drafting a cover letter, brainstorming five different concept narratives for a competition entry, tightening the wording on an awkward client email — ChatGPT (and Claude, and Gemini) are genuinely excellent at this. Corb isn't built for it and won't pretend to be.

Broad research outside architecture. If you need a plain-English summary of a 60-page planning appeal decision, or you want to understand how a new government policy on biodiversity net gain might ripple into practice, general LLMs have simply ingested more of the world. Corb's knowledge is deep in its lane and doesn't try to be deep everywhere else.

Coding and script-writing for parametric tools. For Grasshopper or Dynamo scripting help, general coding-focused LLMs may still have an edge, since programming knowledge across the internet is vast and well-represented in their training data. That said, Corb can supplement this with ArchAdemia's own parametric course content — the Grasshopper, Dynamo, and Parametric Masterclass courses all feed into what Corb can reference — so it's not starting from zero.

The honest position: use both. Corb for the architecture-specific, UK-grounded, ArchAdemia-aware questions. ChatGPT for everything else. They're not competitors so much as two different tools in the same drawer.

Corb vs ChatGPT: The Direct Comparison

Corb is the best AI assistant for architecture students who need answers grounded in a structured, UK-relevant curriculum. ChatGPT is the best general-purpose AI assistant for broad writing and research tasks that sit outside architectural practice. The table makes the split obvious — it's not a competition, it's a division of labour.

How to Actually Use Corb (Without Wasting Your Membership)

Corb works best as a study companion, not a search engine substitute — the value is in its connection to ArchAdemia's course library and tools, so use it to navigate rather than just to chat.

Ask it about specific courses and lessons. If you're stuck on a concept mid-course — say, you're lost on Revit worksets in the BIM Collaboration course — ask Corb to explain it in different terms, or ask it what course to take next based on a skill gap you've identified. This is the single highest-value use case, because it's the one thing ChatGPT structurally cannot replicate.

Use it for UK practice and regulation questions. Instead of googling "RIBA stage 3 deliverables" and landing on a 2019 blog post from a practice that's since rebranded twice, ask Corb directly. It defaults to current UK context and won't need three follow-up prompts to stop giving you American answers.

Use it to navigate the Toolkit and Drawing Board. Not sure which fee calculator fits a small residential job versus a larger mixed-use scheme? Ask Corb. Not sure how to structure your portfolio for a Part 2 application? Ask Corb, then take it to the Drawing Board for actual feedback on the work.

It's included in ArchAdemia membership at no additional cost — there's no separate subscription, no usage cap designed to upsell you into a higher tier. It's part of the platform, the same way the Toolkit and Drawing Board are.

FAQ

ChatGPT is a brilliant generalist and a mediocre architect. Corb is the opposite — narrow, specific, and right about the things that actually trip people up in practice, from RIBA stage confusion to render settings that don't match the tutorial you watched. Use the generalist for the email. Use Corb for the job.