AI meeting notes for Saudi teams — in Arabic and English, confirmed by a human

Saudi business runs in Arabic by default — government tenders, board minutes, regulatory submissions, the محضر اجتماع that has to be on file. But a real Riyadh meeting is not pure Arabic. It is Arabic with English business terms dropped in mid-sentence — the code-switch that happens dozens of times in one call.

The incumbents optimize for one language per recording. Otter, Fireflies, and Read.ai expect Arabic or English, then lose the lines where the meeting switched. You get a transcript with holes exactly where the decisions were made.

Knowcap keeps language per utterance. The Arabic stays Arabic, the English stays English, and the switched line in between is captured whole. When the meeting ends, you export the decision record — the محضر اجتماع — in Arabic or in English, whichever the file needs.

Most tools treat AI output as ground truth. They transcribe, summarize, and — in several cases — auto-push action items into your stack with no human in the loop. When the AI mishears “don’t ship Friday” as “ship Friday,” that becomes a real ticket, a real follow-up, a real mistake nobody approved.

Knowcap puts a named person between the AI and the action. A claim card is a proposal, not a fact, until someone confirms it. Reject the ones the AI got wrong; confirm the ones it got right. Only confirmed claims reach the agents.

The result is an audit trail instead of a black box. When a client later says “that is not what we agreed,” you do not have a paraphrased summary — you have the exact recording, the timestamp, the speaker quote, and the name of the person who confirmed it. The confirmation is the product.

PDPL is Saudi Arabia’s real Personal Data Protection Law, regulated by SDAIA. Knowcap’s claim here is narrow and defensible.

First, data residency: Knowcap can run in EU/MENA regions, so meeting recordings and the personal data inside them need not leave the region. Second, a named-human audit trail: every extracted claim is confirmed or rejected by a named person, and that confirmation — with the exact timestamp and speaker quote — is kept as the permanent record. Together these support PDPL accountability and record-keeping, the kind audit and regulated firms care about. For what audit firms must keep on file, see PDPL and AI meeting records.

We do not claim Knowcap is “PDPL certified” or “SDAIA approved” — no such certification exists, and anyone telling you they have one is guessing. The defensible line is simply this: data-residency options plus a named-human audit trail help your team meet PDPL accountability and record-keeping requirements — versus tools that auto-push raw AI output with no human gate and no traceable record of who approved what.

Knowcap is built for Saudi teams whose meetings are bilingual and whose mistakes are expensive. Saudi SMEs riding Vision 2030 expansion that need real meeting documentation, not a transcript with holes in it. Odoo implementation partners turning a client call into an Odoo SH ticket and a GitHub PR before the meeting ends — Knowcap is built by an Odoo partner. Agencies capturing scope and approvals with receipts. And audit and professional-services firms, who need every agent action tied to a human-confirmed claim with a timestamp and speaker quote — because when a regulator or client asks who decided something and when, “the AI summarized it” is not an answer.

The interface is in Arabic where it counts, the capture handles the code-switching your team actually does, and the محضر exports in the language the file requires.