The Real QuestionFig. 01 · Provenance of an Answer
Can a General-Purpose AI ToolProduce Regulatory Evidence?
Banks ask whether general-purpose AI tools can replace a purpose-built compliance evidence system. This is not a question of capability. It is a question of architecture.
The AnswerNo, It Cannot
Fig. 01 · Provenance of an AnswerOne Demand · Two Architectures
The Examiner’s Demand
“Show Me Your Record for This Obligation.”
BSP.MORB.2023.Sec921.d.p22.OBL1 · Face-to-Face Contact · BSP MORB Part IX
General-Purpose AI
It Improvises an Answer
Input · A Compliance Question
01
Searches for Fragments
Finds pieces that seem related, not the full obligation.
02
Generates a New Answer
A different answer every time you ask.
03
Nothing Is Stored
The session ends and it is gone.
04
Cannot Be Traced
No source text, no audit trail.
No Artifact
Session Ends
Dead End
Nothing stored or versioned.
No ReplayNo Trail
ProfytAI Evidence System
It Maintains the Register
Input · The Full Regulation
01
Retrieves the Stored Obligation
Every obligation preserved word-for-word.
02
Returns Verbatim Output
The same obligation ID returns the same text.
03
Stored and Versioned
Persistent, reproducible, retrievable.
04
Fully Traceable
Source text, mapped controls, evidence.
Audit-Ready
Artifact Stored
BSP.MORB.2023.Sec921.d.p22.OBL1 · v1 · p. 22
Versioned, reproducible record.
VerbatimControlsEvidence
It Vanishes When the Session EndsOnly One Leaves a Record
The Examiner’s Demand
“Show Me Your Record for This Obligation.”
BSP.MORB.2023.Sec921.d.p22.OBL1 · Face-to-Face Contact · BSP MORB Part IX
General-Purpose AI
It Improvises an Answer
Input · A Compliance Question
01
Searches for Fragments
Finds pieces that seem related, not the full obligation.
02
Generates a New Answer
A different answer every time you ask.
03
Nothing Is Stored
The session ends and it is gone.
04
Cannot Be Traced
No source text, no audit trail.
No Artifact
Session Ends
Dead End
Nothing stored or versioned.
No ReplayNo Trail
ProfytAI Evidence System
It Maintains the Register
Input · The Full Regulation
01
Retrieves the Stored Obligation
Every obligation preserved word-for-word.
02
Returns Verbatim Output
The same obligation ID returns the same text.
03
Stored and Versioned
Persistent, reproducible, retrievable.
04
Fully Traceable
Source text, mapped controls, evidence.
Audit-Ready
Artifact Stored
BSP.MORB.2023.Sec921.d.p22.OBL1 · v1 · p. 22
Versioned, reproducible record.
VerbatimControlsEvidence
These Are Not Two Versions of the Same Tool.
General-Purpose AI is a search and summarisation assistant. ProfytAI is a compliance evidence system. They solve fundamentally different problems.
The Examination TestFig. 02 · Examiner Scorecard
Three Questions Every Regulator Asks.General AI Fails All Three.
Standard examination questions. Any compliance tool that cannot answer them is a liability, not an asset.
Examiner Question
General-Purpose AI
Chat Assistants
ProfytAI
Evidence System
Q01
“Show me the exact regulatory text your control satisfies.”
Every examiner asks this first. The answer must be verbatim. Paraphrasing is not acceptable.
Cannot Comply
Paraphrases source text. The regulator's exact words are not preserved.
Verbatim Source Text
Stored word-for-word at ingestion. Retrievable on demand.
Q02
“Reproduce this compliance output exactly as it was generated.”
Reproducibility is a core examination requirement. If the output changes on re-run, it is not evidence.
Cannot Comply
Probabilistic by design. Every run produces a different output.
Reproducible on Record
Once generated, the output is pinned as a versioned record and returns identically on every retrieval.
Q03
“Provide the audit record and evidence chain for this finding.”
Without a stored record, there is no audit trail. Without an audit trail, there is no compliance.
Cannot Comply
Session-based. When the session ends, nothing is stored. No record exists.
Versioned Artifact
Every output is stored, versioned, and retrievable. Always.
0 / 3
General AI · Failed
3 / 3
ProfytAI · Passed
Verdict of the Test
General AI was designed to answer questions. ProfytAI was designed to produce evidence that withstands examination. These are not the same system.
Why General AI Fails Under ExaminationFig. 03 · Three Architectural Failures
One Examination.Three Architectural Failures.
These are properties of how language models work. No model, no prompt, and no configuration removes them.
Exhibit
Fig. 03a
Outputs Diverged · No Canonical Record
Failure 01
Non-Reproducibility
Same query. Different output. Every time.
Under Examination
The examiner asks you to reproduce your evidence.
What Happens
The output will differ. A record that cannot be reproduced on request is not a record.
Fixable by Configuring the AI Differently? No.
Language models sample from probability distributions. Even at temperature zero, outputs vary across sessions and model versions. There is no mechanism that guarantees identical output.
Exhibit
Fig. 03b
Obligation Reworded · Legal Precision Lost
Failure 02
Paraphrasing
The regulator's words, rephrased by a model.
Under Examination
The examiner asks for the exact words of the obligation your control satisfies.
What Happens
The AI output does not contain them. A paraphrase of a statute is not that statute.
Fixable by Configuring the AI Differently? No.
Language models generate text. Generation means producing new words. The original regulatory text is lost in the process, even with retrieval-augmented generation.
Exhibit
Fig. 03c
Evidence Chain · Broken at Retrieval
Failure 03
Inferred Citation
A reference with nowhere to go.
Under Examination
You cite the obligation. The examiner asks for the source text your system retrieved.
What Happens
You cannot produce it. The citation was generated, not retrieved.
Fixable by Configuring the AI Differently? No.
The AI infers what the citation should be based on patterns in its training data. It does not retrieve a verified source record. The reference is a prediction, not a retrieval.
Not a Model Problem
A better model will not fix this. These failures come from how every language model works. Evidence needs a different architecture.
The ProfytAI StandardFig. 04 · Built to Be Defended
Examination-Grade Compliance.Built to Be Defended.
CAP 01
Verbatim
Source Text Preserved
Every obligation stored exactly as written in the regulation. No paraphrasing. No summarization. The regulator's exact words, retrievable on demand.
CAP 02
Reproducible
Same Output Every Time
Any compliance artifact can be reproduced identically at any point in time. The output is the same output that was produced at ingestion.
CAP 03
Traceable
Every Citation Verified
Every obligation links directly to its source clause. Every control maps to the obligation it satisfies. Every evidence item is attached and retrievable.
The Four Guarantees
Guarantee 01
Every obligation is traceable to its exact source clause
Guarantee 02
Every output is reproducible on demand
Guarantee 03
Every artifact is stored, versioned, and retrievable
Guarantee 04
Every citation is independently verifiable
These properties apply to content and workflows delivered through the ProfytAI platform for covered regulatory material. They do not replace your policies or sign-offs.
A language model produces answers. An evidence system produces proof. The difference is not quality. It is architecture.
What ProfytAI Is
ProfytAI is not a compliance assistant. It is a compliance evidence system, the layer between your institution and your regulator that makes examination a process of retrieval, not reconstruction.
How the Engine Uses AI
AI Proposes.The Source Decides.
Our compiler uses AI to propose each record. Verification against the source text decides what is kept, and the Integrity Ledger stores it.
Can a Better Prompt or a Newer Model Fix This?
No. Language models sample from probability distributions, so output can vary across sessions and model versions, even at temperature zero. Generation produces new words, so the regulator's text can be lost even with retrieval-augmented generation. A generated citation is a prediction of where the rule should be, not a retrieval of a stored record. Changing models does not change these properties.
See the Evidence for Yourself.
We will walk through a live compliance artifact. Obligation ID, verbatim source text, evidence chain, and audit record.