Agents that pre-audit your affidavit against the DTI Codes, draft the preferential-procurement sections of the response, and watch for ownership or spend changes that would drop your level — all inside your own network.
Based on current ownership register, EE report, and Q2 procurement extract. Skills development is the main drag.
Generic ScorecardNot a portal you log into once a year. Agents that sit next to your files and do the repetitive checking your compliance officer currently does by hand.
Share registers, EE reports, training ledgers, ESD agreements, tender briefs. Drop them in a watched folder, SharePoint library, or hand them over for a one-off run. Nothing is uploaded to a foreign cloud.
Ownership look-through (trusts, companies, vesting), management control, skills spend classification, ESD contribution tests, preferential procurement claims — scored against the active Gazette version, not a stale template.
Risk-ranked issues, missing evidence, and a remediation sequence. Or a first-draft tender response with BEE and preferential sections already filled from your certificate and past wins.
What actually happens between “we need a certificate” and a SANAS-accredited result — and where agents cut the grind.
Share register, EE reports, WSP/ATR, procurement extract, ESD agreements, SED proofs.
Ownership, management, skills, ESD (preferential + ED/SD), socio-economic development.
Priority elements need 40% of available points or the overall level is discounted.
Accredited agency audits the measurement period. Certificate valid 12 months.
Tenders, preferential claims, supplier status — until the next measurement cycle.
Voting rights, economic interest, net value. 40% sub-minimum required. Look-through on trusts and intermediate companies is where most packs fail.
Spend on learning programmes as % of leviable amount, learnerships, absorption. Classification of black training providers is a frequent shortfall.
Preferential procurement from empowering suppliers, supplier development, enterprise development. TMPS calculations and beneficiary eligibility matter.
Agentic SA is a Johannesburg company. The agents run on your premises. The scorecards are South African. The visual language draws from Ndebele geometry, Khoi and San rock-art traditions, and the animals and plants that mark this country — springbok, protea, blue crane, eland.
Not stock “Africa” decoration. Motifs and photographs rooted in South African art, landscape, and national symbols — used as structure for a modern compliance product.
National animal. Agility, confidence, endurance — the leap that clears the fence.
National flower. Beauty, resilience, change — flowering potential.
National bird. Grace and vigilance over the grasslands.
Sacred in many San and Khoi traditions. Power, rain, healing — and the oldest continuous art on the continent.
Example output from a QSE on the Generic Scorecard. Real clients see the same structure — their numbers, their documents.
Hermes is the orchestrator. Specialist agents handle verification prep, tender response, monitoring, local content, and governance — each with a clear job, clear inputs, and a human sign-off gate.
Routes work, loads the correct rule pack (Generic or sector), versions every run, and never bypasses compliance owner approval. Built for on-premise and private-cloud deployment so scorecard data stays inside your boundary. Async sub-tasks (ownership vs skills vs ESD) run in parallel so a full pre-audit does not block the session.
Pre-audits the full scorecard pack against the Codes: Ownership (incl. Net Value), Management Control, Skills (leviable amount), ESD, SED. Flags priority sub-minimum risk and discounting.
Maps RFP clauses to mandatory, functionality, and preference phases. Drafts preferential and B-BBEE claim language consistent with the live certificate. Flags designated local-content and CIDB gates.
Watches ownership changes, material spend shifts, EE headcount, and certificate expiry. Alerts before the next measurement period closes — not after verification finds the gap.
One view of estimated level, open remediation items, verification and tender deadlines, and a full log of every Hermes run. Export packs for the board or as tender annexures.
Hermes can spawn these for deeper work without changing the product surface your team uses day to day.
Net Value, flow-through, modified flow-through, trusts/BBOS/ESOP Annex 100 checks, designated-group flags. Surfaces fronting-risk indicators for human review — never certifies legitimacy.
Leviable-amount reconciliation, category spend mix, preferential procurement TMPS, supplier certificate currency, SD/ED beneficiary eligibility.
Detects dtic-designated products and thresholds in RFPs; lists expected returnables (SBD 6.2, Annex C–E). Does not calculate or verify local content %.
Maps YES headcount and absorption evidence against Gazette targets when a level-up is claimed. Flags missing 12-month completion proof.
Loads Construction, ICT, Financial Services, Legal, or specialised (Statement 004) rules at run start. Records pack version on the audit log.
Indexes uploaded packs, redaction notes, and prior findings so the next pre-audit starts from what already failed — not a blank slate.
Hermes coordinates. Specialists analyse. Humans approve. No agent issues a B-BBEE certificate, designs ownership structures, or replaces SANAS verification or legal advice.
Agents help you assemble and check this pack before the SANAS visit — not after findings land.
Figures from early client work and structured interviews with SA compliance officers in manufacturing, construction, and professional services.
If any of these sound familiar, the agents were designed around your week — not a generic digital transformation brief.
Twenty-plus hours a week on affidavits, ownership packs, and last-minute fixes before the SANAS visit. Measured on pass rate and overtime.
Lost a material tender on BEE points. Paying consultants R1,500+/hour for work that should be systematic. Needs predictability, not heroics.
Preferential claims written under deadline pressure. Inconsistent language across bids. Needs first drafts that match the certificate and past wins.
Anonymised snapshots from manufacturing and construction teams in Gauteng and KZN.
Skills shortfall was the recurring verification finding. Agent flagged the spend gap three weeks before submission; training budget reallocated to Level 1–2 black providers. Zero findings on ownership and skills at verification.
Tender volume high; preferential sections written under pressure. Tender agent now produces first drafts aligned to the current certificate. BD team reports fewer rewrites and clearer claims.
Honest comparison. Consultants still matter for strategy. Agencies still issue certificates. Agents handle the volume and the grind.
| Capability | Consultants | Generic SaaS | Agentic SA |
|---|---|---|---|
| Cost per document review | R1,200–2,000+ | R800–1,500 | R300–600 equiv. |
| Turnaround | 3–5 days | 24–48 hrs | < 2 hours |
| Handles ownership look-through & nuance | ✓ | Limited | ✓ |
| Always on latest Gazette / Codes | Reactive | Manual update | ✓ Self-updating packs |
| Data stays on your premises | Often shared | Cloud | ✓ On-prem / private |
| Scales with tender volume | Linear cost | ✓ | ✓ Near-zero marginal |
Typical path for a Starter or Growth client. Enterprise is scoped separately.
15–20 min call to quantify pain. Optional free readiness check on one anonymised submission so you see the output on your numbers.
Map document sources, scorecard type (Generic or sector), risk areas. Rule packs tuned. Runtime installed on your side of the firewall.
Full BEE check or tender draft with your data. Feedback loop. Adjustments to classification rules if needed.
Your compliance team runs day-to-day. We stay on for Gazette pack updates, support, and monthly check-in on Growth+.
Sectors that live or die by BEE and preferential points
Aligned with the frameworks that decide your score and your contract
We used to lose two full days every quarter locking the affidavit. Last cycle the agent flagged the skills shortfall three weeks early. We fixed the spend, went into verification with zero findings on ownership and skills, and the compliance overtime dropped off the roster.
Preferential sections used to be the bottleneck on every bid. First drafts from the tender agent match our Level and pull language from wins we already know work. Legal still reviews — but we’re not starting from a blank page at 11pm.
What compliance teams run through before the SANAS booking — the same ground the Verification Agent is designed to cover early.
Patterns that show up repeatedly in verification reports and lost tenders — not theory.
Scoring ownership from the register percentage only. Agencies recalculate Net Value with acquisition debt and the time-based graduation factor. Priority sub-minimum fails here more than anywhere else.
Claiming training spend against management-accounts payroll instead of the SARS leviable amount. The agency uses EMP201. The gap almost always reduces points.
Preferential procurement claims on suppliers whose B-BBEE certificates expired mid-measurement period. Track issue and expiry dates, not just “Level 1 on file.”
Black ownership through trusts or intermediaries without Annex 100(B/C/D) compliance, beneficiary awareness, or trustee minutes. Participation can be limited or disallowed.
Preferential sections that claim a higher level than the live certificate. Under deadline pressure this is common — and easy for evaluators to mark down or query.
Designated products require a minimum local manufacturing threshold. Missing SBD 6.2 / Annex C–E, or missing the threshold, is usually a disqualification — not a preference deduction.
Where a gazetted sector code applies, it prevails over the Generic Scorecard. Rule packs on Growth and Enterprise are selected at onboarding.
| Sector / case | What changes | Agent implication |
|---|---|---|
| Construction | Own scorecard; CIDB grade often an eligibility gate; subcontracting / CPG rules on larger works | Sector pack + Tender Agent notes CIDB class/grade |
| ICT | ICT Sector Code weightings and targets | Sector pack on Growth+ |
| Financial services | FSC scorecard; ownership and ESD nuances | Sector pack where applicable |
| Legal (LSC) | Legal Sector Code gazetted 2024 | Map at discovery; do not force Generic |
| Specialised entities | No measurable ownership (e.g. some NPCs / SOEs) → Statement 004 specialised scorecard | Different element set; no fake ownership claims |
| EME / QSE affidavits | Turnover thresholds; 51% / 100% black-owned automatic levels via sworn affidavit | Agent still useful for tender claims and monitoring; verification path differs |
Sector charter status changes over time. Confirm the active Gazette for your industry. Draft Generic Code amendments (2026 public comment) are monitored and shipped as versioned packs only when finalised.
Separate from B-BBEE preference points. When a product is designated by the dtic, the threshold is usually a pass/fail gate.
Does the RFP name a designated product (cables, furniture, steel pylons, textiles, valves, etc.) and a minimum local content %?
Expect SBD 6.2 and Annexures C, D, E. Calculation uses SATS 1286 and the SARB rate on the advert date.
Bids below the stipulated minimum are typically unacceptable and must be disqualified — not merely scored lower on preference.
Flags designated-sector clauses and usual forms. Does not calculate or verify local content % — that remains SABS / dtic / your manufacturing evidence.
Short definitions so your team and ours stay aligned. Based on the Amended Codes of Good Practice and Preferential Procurement Regulations in force — draft 2026 Code amendments are noted separately.
Default measurement for measured entities above the QSE threshold. Five elements: Ownership (25), Management Control (~15–19), Skills Development (20), Enterprise & Supplier Development (40), Socio-Economic Development (5). Sector codes may re-weight.
Ownership, Skills Development, and ESD. Miss the 40% sub-minimum on any priority element and your overall B-BBEE level is discounted by one — even if total points look fine.
Tracing black ownership through trusts, companies, and intermediate vehicles — not only the name on the share certificate. Flow-through and modified flow-through rules apply.
Ownership priority indicator. Measures economic value to black participants after acquisition debt, using a time-based graduation factor. Sub-minimum is 40% of the Net Value points.
Total Measured Procurement Spend. The base used to calculate preferential procurement under ESD. Intra-group spend is typically excluded from recognition.
Independent audit by a SANAS-accredited verification agency under R47-03. Certificate is typically valid for 12 months from issue. Only accredited agencies issue certificates.
Size bands by annual turnover (indicative): EME under ~R10m (affidavit); QSE ~R10m–R50m; Generic above. EMEs that are 100% black-owned can claim Level 1 by affidavit.
PPPFA preference systems. Contracts ≤ ~R50m: 80 price + 20 preference. Above: 90 + 10. Preference often maps to B-BBEE level; organs of state may also set specific goals.
If a Generic entity fails any priority-element sub-minimum, the issued level is reduced by one. QSEs must hit Ownership plus either Skills or ESD.
Status that affects how buyers score preferential procurement from you. Linked to local production, jobs, and other criteria under the Codes — checked at verification.
Industry-specific scorecards (construction, ICT, financial services, tourism, agriculture, etc.). Where a gazetted sector code applies, it prevails over the Generic Scorecard.
Organs of state may allocate preference points to goals such as black ownership, women, youth, disability, or local content — not only overall B-BBEE level. Read each RFP carefully.
A transaction or arrangement that undermines the B-BBEE Act — window-dressing, benefit diversion, or opportunistic intermediaries. A criminal offence. Agents estimate scores; they do not legitimise sham structures.
Youth Employment Service (Gazette 41866). Qualifying entities can gain one or two recognition levels (and bonus points) by creating 12-month youth jobs and meeting absorption targets — subject to priority-element preconditions.
The SARS skills-levy base used as the denominator for Skills Development spend. Agencies reconcile to EMP201 — management-accounts payroll often fails at verification.
dtic-designated products must meet a minimum local manufacturing threshold (SATS 1286). Failure to meet the threshold is typically a disqualification, not a scoring deduction. SBD 6.2 + Annex C/D/E are common returnables.
Construction Industry Development Board registration grade and class of works. Often a pass/fail eligibility filter on public construction tenders — separate from B-BBEE preference points.
| B-BBEE level | Score (Generic, indicative) | 80/20 preference pts | 90/10 preference pts |
|---|---|---|---|
| Level 1 | ≥ 100 | 20 | 10 |
| Level 2 | ≥ 95 | 18 | 9 |
| Level 3 | ≥ 90 | 14 | 7 |
| Level 4 | ≥ 80 | 12 | 6 |
| Level 5 | ≥ 75 | 8 | 4 |
| Level 6 | ≥ 70 | 6 | 3 |
| Level 7 | ≥ 55 | 4 | 2 |
| Level 8 | ≥ 40 | 2 | 1 |
| Non-compliant | < 40 | 0 | 0 |
Indicative thresholds under the Amended Codes. Sector codes and draft 2026 amendments may differ. Preference points under PPPFA 2022; organs of state may use specific goals instead of pure level mapping. Always confirm against the active Gazette and the tender document.
Generic global tools do not know a SANAS agency, a Gazette notice, or a look-through ownership chain. These agents were written for the Codes you are measured on.
Rule packs track the Generic Scorecard and sector codes — Ownership, Skills, ESD priority sub-minimums included.
On-premise runtime. POPIA-aligned by design. Share registers and EE reports never leave your network by default.
Built around Q3–Q4 booking pressure, certificate expiry, and the 12-month validity cycle SANAS agencies work to.
Tender drafts that match your current level and pull from past wins — PPPFA and preferential clauses first-class.
Geometric forms drawn from Ndebele mural language and national symbols — built for a modern South African compliance platform.
Most teams are productive inside two weeks — not a six-month transformation programme.
Scorecard type, document sources, who owns alerts. Optional free pre-audit on one pack.
Runtime on your VM or private cloud. Rule packs loaded. Connectivity checks pass.
Live BEE run on a real (or redacted) pack. Tender draft on a past RFP if you use that agent.
Your team runs day-to-day. Gazette packs update in place. Support on SAST hours.
Send one anonymised (or redacted) submission. We’ll return a structured findings list so you can judge the agent on your real numbers — not a marketing deck.
Request the free pre-audit
