Agentic Decision Intelligence
Purpose-built AI
for the business of architecture and design.
Where firm data becomes actionable intelligence and automated workflows.
One system, five decision domains
The questions principals actually ask, and where Norma answers them.
Are we underpricing this work?
Norma prices it three ways, against the market, from your own costs, and against your completed comparables, then triangulates a verdict. A fee below the floor is flagged.
Fees & pricing →Can we bill for this?
Norma reads the contract language, the scope history, and the client request, and returns a defensible read on whether the work is billable.
Contracts & scope →What's our biggest risk this quarter?
Norma reads the contracts you signed and the receivables you are owed: risk findings cite their clauses, and late money carries an escalation verdict.
Cash & collections → Contracts & scope →Where are we leaking margin?
Norma walks the budget to the actual to the transaction, and puts backlog, payroll, and utilization in one read, so the margin question and the hiring question both start from the numbers.
Budget & operations →What is the firm worth?
Norma builds a firm valuation from revenue quality, concentration, margins, backlog, and owner dependency, a number you can explain in the room.
Firm value → The full spineFive domains, one system, from the books to the verdict.
See how the workspace thinks →Why Norma is different
You've always had data.
Now you can talk to it.
Norma is a vertical AI platform built for architecture and design firms. The economics of the profession, the structure of the work, the decisions that recur at firms like yours, encoded in the system.
Your existing tools store data. They are systems of record, and they do their jobs. None of them consolidate your firm's full operating context in one place. Norma does. It is the layer above your stack where the firm's full picture lives, and where the work principals would otherwise hand to internal staff gets done.
The reasoning is deterministic. The same question, on the same firm, produces the same answer with the same chain of logic. Not the probabilistic output of a generalized AI tool. Reasoning a principal can take into a meeting and defend.
The workspace, working
See it, then believe it.
The review reads the report to you.
A verdict first, then the story. Every figure in the prose is drillable to the transactions beneath it, and the review is rejected outright if it states a number the books do not hold.
Every dollar has a name on it.
A contested account splits to the people who spent it, each row carrying its receipt. The remainder nobody can claim stays amber and honest, and the sums foot to the cent.
The contract answers before the argument starts.
Every risk finding cites the clause it stands on, carries its severity, and arrives with the action to take. A finding that cannot cite its clause is dropped.
Three ways to price. One axis. One verdict.
The market band and your own delivery cost drawn on the same line, your number between them, and the verdict computed live at every view.
How the answers hold up
Every answer shows its work.
The reasoning is an engine you can inspect, not a model's guess. That is what decision intelligence means here, across every domain the system runs.
Deterministic engines
The numbers come from engines with fixed, inspectable methods: the same thresholds, the same formulas, the same inputs, the same answer. Open any verdict and the method is on the page.
The AI narrates. It never computes.
The model writes the reading; the engines produce every figure in it. A number the engine did not compute is rejected before it reaches you. The narrative is checked against the data it cites, and a narrative that fails the check is rewritten or refused.
Every figure drills down
A spend total opens to its transactions. A risk finding cites its clause. A budget line opens to the rows that summed into it. Nothing asks to be taken on faith.
Ask, and get evidence
Ask Norma answers in plain language, and every figure in the answer carries a citation to the record it came from. When it cannot cite, it says so instead of guessing.
How it runs
Firm data in. A verdict with its receipt, and a drafted action, out.
-
01
Inputs
Timesheets, AR, pipeline, budgets, contracts, financials.
-
02
Deterministic engines
Fixed, inspectable methods. The same answer every time.
-
03
Verdicts, with receipts
Every figure drills to the source beneath it.
-
04
Drafted actions
Memos, statements, invoices, and follow-ups from the live books.
-
05
The human gate
A queue a person approves. Nothing client-facing sends itself.
Built for architecture firms
Built on the economics of your profession, not adapted from generic finance software.
- Fee setting & pricing floors
- Scope change detection
- Contract risk assessment
- Additional services evaluation
- Budget vs actual & attribution
- Owner budget reports
- Billing & invoicing
- Collections & AR
- Overhead analysis
- Utilization & realization
- Firm valuation
- Cash position
Built for
- Architecture and design firms with roughly 10–100 employees
- Studios with project-based revenue and complex staffing decisions
- Firms dealing with scope creep, weak forecasting visibility, or inconsistent fee discipline
- Principals who want better operating judgment without hiring a full-time CFO
- Finance leaders who need a decision layer above spreadsheets
Not built for
- Solo freelancers who only need invoicing
- Firms only looking for bookkeeping
- Enterprises looking for generic BI dashboards
- Teams unwilling to upload basic financial, project, or contract data
Norma was built by a finance leader with 12+ years inside architecture and design firms, from project controlling to partner-level advisory.
Why it was builtSecurity and trust
Firm workspaces, scoped context, no training on your files.
-
01
Each customer workspace is isolated from others.
-
02
Standard API terms with our AI providers: data you send is not used to train public models.
-
03
Uploads stay tied to your workspace; export and deletion requests are handled in product.
Private early access, from $895 a month for qualifying firms.
See what it includesRequest access
Ready to run a real decision?
Norma is currently available through private early access. Firms request access, complete a fit review, and onboard through a guided data setup process. Prefer to follow the work first? Leave your email below.
Follow the work
Not requesting access today? Leave your email and follow what we ship.