Your records weren't
built for Copilot. Yet.
Cleaned up, classified, governed, and safe for the AI tools your team is starting to use. So when you turn on Copilot, it makes you more productive — not more exposed.
Book Free Assessment →Every AI rollout finds the same skeletons.
Three of them, usually within the first week.
-
i.
Redundant, obsolete, trivial records.
ROT, in the trade. Years of duplicates, draft versions, “Final_v7_REAL.docx,” and content nobody's opened since 2018. AI tools read all of it. Every wrong answer Copilot gives, ROT helped write.
-
ii.
Oversharing.
SharePoint sites with “Everyone except external users” turned on. OneDrive folders shared with the whole organisation. Five years of permissions debt. AI doesn't care about org charts. It surfaces whatever it can read.
-
iii.
Sensitive data, hiding in plain sight.
SA ID numbers in old spreadsheets. FICA documents in shared folders. Salary information in board pack drafts. Medical aid claims in customer service archives. The moment AI can search across the estate, all of it becomes findable.
Four things. In a sensible order.
You don't skip step one because step four sounds more interesting. We've seen what happens when you do.
-
i.
AI records readiness assessment.
We scan your records estate (with your permission, obviously) and tell you exactly what shape it's in. ROT volume, oversharing exposure, sensitive data hotspots, classification readiness. You get a report you can take to the board.
-
ii.
ROT cleanup & defensible deletion.
Years of redundant records, gone. Defensibly. With audit trail. Without disrupting the things people actually use.
-
iii.
Oversharing & permissions remediation.
Every SharePoint site, every OneDrive folder, every Teams workspace. Permissions tightened to need-to-know. Documented so it stays tight.
-
iv.
Sensitivity labelling & classification.
Records intelligently classified — public, internal, confidential, restricted. South African sensitive information types built in: SA ID, FICA, tax numbers, medical aid, banking, biometrics. Custom classifiers for your specific record types (board packs, contracts, claims, applications).
Copilot governance, done properly.
Once the records estate is ready, we wrap governance around the AI deployment itself. So the tools work for the business — not against the policies.
-
i.
Copilot pilot governance framework.
Pilot scope, eligibility, monitoring, and the policies that have to be in place before you let the first user in.
-
ii.
AI acceptable use policy and user awareness.
Plain-language rules your team will actually read, with the training that goes with them.
-
iii.
Shadow AI discovery.
Because employees are already using something you don't know about. We find it, then we decide what to do about it together.
-
iv.
AI governance committee secretariat.
We run the committee — agenda, minutes, risk register, the unglamorous work that makes governance real.
-
v.
Continuous AI risk monitoring.
Quarterly reviews, regulator-tracking, and the “what changed” briefing your CIO actually wants.
Bonus: you finally understand what's in your archives.
Cleaning records for AI also means seeing records for the first time. We surface what's duplicated across the estate, what's expired and shouldn't still be there, what's sensitive and isn't labelled, what's business-critical and isn't backed up, and what's quietly costing you money.
Useful for AI. Also just useful.
Records discovery & eDiscovery.
Litigation, regulator investigations, high-volume DSAR or PAIA requests. We collect, process, review, and produce. Legal hold management, forensic collection, predictive coding, privilege review. Records discovery at scale, defensibility under audit.
Most AI consultants don't know records management. Most records firms don't know AI.
We do both. Because the AI part of “AI-Ready Records” is downstream. The records part is upstream. And the firms that can speak fluently to both — the CIO who's deploying Copilot and the Information Officer who's worried about POPIA — are rare. That's the gap we sit in.
Thirty minutes. We'll tell you where your records stand on the modernisation curve, what's likely to break when AI hits, and what would actually move the needle.
— T.