Four specialised workers trained on the Codes of Good Practice, PPPFA, and the verification practices SANAS agencies actually apply — running inside your network.
Not chatbots. Structured workers that ingest your documents, apply the current Codes, and return findings or drafts your team can act on.
Pre-audits your draft affidavit and supporting packs against the active Codes — including priority-element sub-minimums (Ownership, Skills, ESD) so you are not discounted at verification.
Reads the RFP, maps mandatory and preferential clauses under PPPFA, and produces a first draft with BEE claims consistent with your current certificate and language from past wins.
Watches ownership registers, procurement extracts, and EE reports. Alerts when a change would move your estimated level — with plain-language note and next step.
One view of estimated level, open remediation items, verification and tender deadlines, and a full log of every agent run. Export packs for the board or as tender annexures.
Aligned to the Generic Scorecard structure used by SANAS-accredited agencies. Sector codes may re-weight elements; the agent loads the pack that matches your measured entity.
Voting rights (25%+1 target), economic interest, Net Value with time-based graduation, designated groups. Look-through on trusts and intermediate companies. 40% of Net Value points = sub-minimum.
Board and executive participation by race and gender against Statement 200 targets. Senior, middle, and junior management; employees with disabilities.
Spend vs leviable amount (typically 6% target), learnerships, absorption bonus. Classification of black training providers. 40% of element weighting = sub-minimum.
Preferential procurement against TMPS, supplier development, enterprise development. Beneficiary eligibility and empowering-supplier status. Sub-minimum on each category.
Typically 1% of NPAT toward income-generating initiatives for black beneficiaries — timing within the measurement period and evidence of beneficiary status checked.
Miss any priority-element sub-minimum and the discounting principle reduces your overall level by one. Draft 2026 Code amendments under public comment may adjust weightings — rule packs will track the gazetted final text.
Agents run where your documents already live.
Runtime installed behind your firewall. File-system and optional SharePoint / Drive watchers. Nothing uploaded to a foreign cloud by default.
Same agents in a VPC or private Azure/AWS account you control. Suitable for groups with central IT standards.
REST hooks for ERP / CRM sync when volume and integration demand it. Scoped during enterprise onboarding.
One orchestration pattern for every pre-audit, tender draft, and monitor alert — versioned, parallel where safe, and always gated by a human.
Evidence pack or RFP is hashed, classified (scorecard vs tender vs monitor event), and bound to a measurement period or tender ID.
Generic, sector (Construction, ICT, FSC, Legal…), or specialised Statement 004. Pack version is written to the run log.
Ownership, Skills/ESD, Tender clause, Local Content Scout, and YES tracker run as async sub-agents where the job needs them.
Hermes merges findings, ranks remediation, and holds output until the compliance owner approves. Nothing ships to a tender or board pack unattended.
| Agent | Owns | Does not |
|---|---|---|
| Hermes | Routing, pack version, parallelisation, audit log, approval gate | Element scoring itself |
| Verification | Scorecard pre-audit, priority sub-minimums, findings | Issue certificates |
| Tender Response | Clause map, preferential draft, local-content / CIDB flags | Submit bids or sign forms |
| Continuous Monitor | Drift alerts, cert expiry, ownership/spend triggers | Auto-change share registers |
| Governance | Scoreboard, deadlines, board/tender export packs | Replace board judgement |
| Ownership Analyst | Net Value, trusts, flow-through, fronting indicators | Design or approve ownership deals |
| Local Content Scout | Designation detection, SBD 6.2 returnable checklist | SABS verification or % calculation |
Enough specificity for a compliance officer or IT lead to judge fit. Not marketing fluff.
Designed for share registers, EE reports, and tender files — not for multi-tenant training on your data.
Runtime installs on your VM or private cloud. Processing stays inside your perimeter unless you enable an optional connector.
Every agent run is logged: input hash, rule-pack version, timestamp, operator. Exportable for internal audit.
Agents propose findings and drafts. Final affidavit and tender submission remain human-approved.
Gazette-aligned packs are versioned. You know which Code edition produced a given report.
Email support on Starter; priority queue and monthly check-in on Growth. Enterprise gets a named contact.
Every findings report records the Code pack version. Updates ship without a full reinstall.
You host the agents. We support the software; your IT controls the VM or private cloud SLA.
Install guide, scorecard mapping checklist, and a short “first verification season” playbook included.
Redact where needed. The clearer the pack, the more useful the findings list.
Send one redacted submission. We return a structured findings list — no deck, no obligation.
Request free pre-audit
