What Makes Tektome Decision-Support, Not an AI Checker

Every major AEC project now has access to an AI checker. Upload a drawing set, wait a few seconds, get a verdict against a code reference. It looks like assurance. It isn’t.

When “checked” isn’t the same as “assured.”

There is a precise difference between checking and assurance, and it sits exactly where project risk lives.

Checkers verify against external codes. Tektome assures against your own requirements.

A code checker answers one question: does this drawing comply with the relevant regulation? That is a legitimate and useful question. It is also a narrow one.

Your project’s requirements are not a building code. They are the layered, firm-specific accumulation of your client brief, your development standard, your specialist consultant outputs, your planning conditions, and, post-Building Safety Act 2022, the dutyholder obligations your organisation has explicitly accepted. No external code database contains that. No off-the-shelf checker can check against it.

Tektome checks against your requirements. It reads your brief, extracts and manages your requirements, and then reviews incoming submissions, drawings, specifications, model data, against what you committed to deliver. That is the distinction that matters when the signature is yours.

Graded confidence, not a verdict.

Code checkers return a result: pass, fail, or flag. That binary forces the evidence into a shape it doesn’t naturally have.

Design information is rarely all-or-nothing. A requirement may be partially addressed in the drawings but not yet confirmed in the spec. A structural assumption may be consistent with the brief but unverified against the latest ground investigation. The honest answer in both cases is not a tick or a cross: it is a calibrated position.

Tektome expresses confidence as a grade backed by evidence:

• What has been confirmed, with citation to the source that confirms it
• What is partially addressed and what remains open
• What has not yet been reviewed at all, the honest “not checked” that a checklist buries

That graded view is what Dame Judith Hackitt’s 2018 Independent Review of Building Regulations and Fire Safety identified as structurally absent from industry practice: the ability to see where confidence is low, not just whether a gate was passed. A false all-clear is more dangerous than an honest amber.

A false all-clear is more dangerous than an honest amber.

The compounding record that checkers lack.

Here is what no checker, however capable, can replicate: the record of how your firm has resolved judgment over time.

Every time a senior reviewer decides how to interpret a client brief clause, how to weight a structural query against programme risk, or how to handle a gap in the specification, that is a piece of resolved judgment. In most firms it lives in someone’s head, or in an email thread, or it walks out the door when the reviewer leaves.

Tektome captures it. Each review cycle adds to a firm-specific record: which requirements were raised, how they were addressed, what evidence was cited, what confidence was reached. The next project on a similar brief starts from that accumulated position, not from zero.

Keeping a human signing.

None of this removes the professional. It is designed to keep a professional meaningfully in the loop.

The problem with checkers that produce a verdict is that they create the conditions for rubber-stamping. The output looks authoritative; the reviewer’s job becomes approving the AI’s finding rather than exercising independent judgment.

Tektome’s graded, evidence-linked output is designed for the opposite dynamic. A reviewer sees exactly which requirement was reviewed, which document was cited, what confidence was reached, and what remains open. They are being asked to review evidence and hold a position, not ratify a conclusion.

Under the Building Safety Act 2022 dutyholder regime, that distinction is not cosmetic. A signing professional carries personal competence liability. The question is not whether a system produced a pass: it is whether the professional can demonstrate that the review was real, evidence-backed, and appropriately calibrated. Tektome is built to make that demonstration possible. No checker is.

Why this reaches the risk budget.

AEC tools budgets are small. Risk budgets are not.

A code checker is bought from the tools budget: it reduces hours, speeds up a workflow, and produces a consistent output. That is genuine value at its own level.

Tektome is bought from the risk budget. The question it answers is: if this scheme has a defect, a Gateway delay, or a dutyholder liability event, can you demonstrate that your assurance process was robust enough to have caught it? That question is worth a different order of magnitude.

Mid-tier owner-developers running higher-risk residential, life sciences fit-out, or data centre schemes are already living this. Programme delays at Gateway 2 cost capital. A defect that reaches site costs more. The compounding liability of a signature on a scheme that was checked but not assured costs most of all.