What Compliance Teams Say
About Working with Teguh

CHALLENGE

A domestic commercial bank's compliance team was spending an estimated 60% of analyst hours on first-read document processing — reviewing policy updates, AMLA circulars, and case file summaries before any analytical work could begin. The head of compliance wanted to know whether any of this reading load was appropriate for AI assistance without creating regulatory risk.

WHAT TEGUH DID

Teguh conducted a Compliance Workflow Review: reading a sample of the policy library and structured case files under a signed data access agreement, then meeting with the compliance head and one senior analyst. A written brief identified three document categories where AI reading assistance was appropriate and two where it was not, with reasoning for each. A data-classification map was delivered alongside the brief.

OUTCOME

The brief gave the compliance head a structured basis for her internal proposal to technology governance. The data-classification map was incorporated directly into the bank's subsequent vendor evaluation framework. The review was completed within the agreed scope at RM 1,330 and took approximately three weeks from data access agreement to delivery.

"We finally had a document we could share with legal and with the technology team that spoke the same language as our compliance framework."

CHALLENGE

Following an internal decision to explore AI reading assistance, a bank's compliance function needed a supervised pilot that would produce RMiT-compatible documentation and train their analysts to work with AI outputs critically. Previous vendor proposals had not addressed the documentation requirement in sufficient detail.

WHAT TEGUH DID

Teguh deployed a read-only AI assist in a closed environment within the bank's infrastructure over a six-week period. The AI produced summary memos for policy documents and surfaced cross-references in case files for analyst review. Training was delivered to two analysts covering output interpretation, known limitations, and the escalation protocol Teguh wrote for their function. Documentation was maintained throughout in alignment with BNM RMiT requirements.

OUTCOME

At the close of the six-week supervised period, analyst first-read document volume had reduced by approximately 35% for the policy categories covered by the pilot. The RMiT-aligned documentation was accepted by the bank's technology risk function without revision. Both analysts reported increased confidence in identifying when to escalate AI output for manual verification.

"The pilot ran quietly. That was actually what we needed — no drama, no surprises, and documentation that held up when internal audit reviewed it."

CHALLENGE

Having deployed an AI reading assist following a successful pilot, a compliance team found that they had no systematic process for monitoring the quality and boundary adherence of AI outputs over time, nor a standard format for reporting AI governance to their compliance committee each quarter.

WHAT TEGUH DID

Teguh began a Quarterly Compliance Stewardship engagement. Each quarter, Teguh samples AI outputs from the preceding period, audits boundary adherence against the original data-classification map, checks regulatory alignment against current BNM RMiT and AMLA guidance, and produces a written brief for the compliance committee. In Q4 2024, a yearly board summary was also produced for inclusion in the annual board review.

OUTCOME

The compliance committee now receives a quarterly brief that addresses AI governance directly alongside their other standing compliance matters. The Q4 2024 board summary was presented without requiring further preparation from the compliance team and was noted by the board chair as a useful addition to the annual review pack. Stewardship continues quarterly.

"The quarterly brief takes one item off my agenda that I used to spend half a day preparing. The committee trusts it because it comes from outside, and I trust it because Teguh knows our boundaries better than most of our own staff."