Targe is a multi-tenant SaaS that turns half-formed ideas into graded, gated, audit-ready AI projects. Every tenant gets its own white-labelled subdomain, its own Claude and voice keys, and row-level isolation enforced at the database — not by a WHERE clause.
UIA and DPIA templates live in scattered Word docs. Compliance is a slide someone updates monthly. Sign-offs happen by email. When a regulator asks show me how this AI system was approved, a spreadsheet is not an answer — and the EU AI Act, POPIA s71, GDPR Art. 22, ISO 42001 and King V don't grade on effort.
Every AI initiative in the company walks the same path. No dev without green compliance. No deploy without the sign-offs the playbook requires.
Anyone can log a half-formed idea. A Coach sub-agent guides framing — problem, users, value, risk. A BA chat (voice or text) drills into scope until the brief is promotable.
One click turns the idea plus chat history into a project with a pre-filled solution doc, user stories and required stack. The analyser runs synchronously.
Required tech cross-referenced against your approved catalogue. If Clerk is proposed but Supabase Auth is approved, Targe recommends the substitute — automatically.
Every requirement scored against POPIA, GDPR, EU AI Act, NIST AI RMF, ISO 42001, King V. Auto-re-runs when jurisdictions or evidence change.
A five-phase playbook engine enforces the transitions. Server-side. Clients can't fake green — the gate is the database, not the UI.
Sponsor and steerco views, model + prompt versioning, incident register, DPIA on demand, quarterly workspace review exported to PowerPoint in a click.
Push-to-talk transcript in, streamed audio out, animated avatar. Generates DPIAs, user stories and risk callouts on demand. BYO ElevenLabs key.
Upload the BA doc; Claude decides whether it's AI-related, extracts user stories with acceptance criteria, proposes an approach, lists the stack — under your Claude key.
Every model and prompt change is hashed and dated. Compliance re-runs automatically. "What was the model running when this decision was made?" — answerable to the day.
Quarterly AI-strategy review grounded in NIST AI RMF, covering AI Policy, Roles & Responsibilities matrix and 12-month Roadmap. One-click export to a board-ready .pptx.
Severity 1 → 4 lifecycle, server-enforced state machine, auto-fires alerts for sev 1/2. Closes SOC 2 CC7.3 evidence — no more Slack-thread-as-incident-record.
UIA scorer, Claude 14-item enforcement checklist, vendor due-diligence RAG scoring, agent prompt library — persisted per project.
Bring your own Claude and voice keys, register your first initiative, and see the compliance grade before you commit to anything.
Row-level isolation from day one. You get {workspace}.targe.okodi.ai, an admin seat, and space for your team.
Your keys, envelope-encrypted at rest. Your model traffic never mixes with anyone else's. You keep the invoice.
Toggle POPIA, GDPR, EU AI Act, NIST AI RMF, ISO 42001, King V. Grade a real initiative against the strictest control set that applies.
Every decision written down, append-only, tamper-evident. Export a board-ready PowerPoint for your next steerco.
No credit card required. Upgrade when you're ready.
Bring an AI project you're weighing up — internal, live or paused. We'll walk it through the flow end-to-end in under an hour.
Book a walkthroughTurn on ZA, EU, US, UK or Global from Settings. A highest-standard resolver picks the strictest applicable control set — EU's 72-hour breach window wins over POPIA's "reasonable time", and so on.
s8 lawful basis · s71 automated decisions and right-to-recourse · Cybercrimes Act alignment.
Art. 6 lawful basis · Art. 22 right to explanation · Art. 35 DPIA · Chapter V cross-border transfers.
Annex III high-risk triage · Annex IV technical documentation · model and prompt versioning.
Map · Measure · Manage · Govern — the workspace compliance review is grounded in this framework.
§7.4 AI management responsibilities · ethical leadership · board-level visibility · documented decisions.
AI Policy · Roles & Responsibilities matrix · 12-month Roadmap · SOC 2 CC7.3 incident response evidence.
Real multi-tenancy, real key isolation, real audit — the parts that matter when a CISO reads the security questionnaire.
Postgres RLS on every tenant-owned table with FORCE ROW LEVEL SECURITY. 42 pgTAP assertions prove tenant A can't read tenant B — in CI, every commit.
Per-tenant Claude and ElevenLabs keys, AES-256-GCM encrypted, master key from KMS. Decrypted in memory for a single run; buffer zero-filled in finally.
RLS denies UPDATE and DELETE. Every gate transition, evaluation and key rotation is recorded and tamper-evident — readable only to the tenant that owns it.
Each tenant gets {tenant}.targe.okodi.ai, plus an optional custom domain. Brand colours and logo drive the whole portal via CSS variables.
The questions we hear most from executives and risk owners weighing up an AI governance platform.
AI governance is the set of policies, controls and evidence that decide which AI initiatives a company runs, how they are approved, how they are monitored in production, and how the record survives an audit. It sits between AI strategy and day-to-day AI use — the layer that turns "we should govern this" into a system of record.
If your teams already use AI — even informally, in copilots, chatbots or spreadsheets — you already need governance. The question is whether it is written down or whether the person answering "who approved that?" is guessing. Any company subject to POPIA, GDPR, the EU AI Act, ISO 42001 or a board risk register benefits from formal AI governance.
Not as a standalone law today, but it is required to comply with several existing laws that touch AI. POPIA (South Africa) governs automated decisions and personal data. GDPR applies if you process EU data. The EU AI Act applies to any high-risk AI system used with EU users. King V requires the board to oversee technology risk — which now includes AI. Targe grades against all of these automatically.
The NIST AI Risk Management Framework is the US National Institute of Standards and Technology's voluntary framework for governing AI — organised around four functions: Govern, Map, Measure and Manage. It is the most widely adopted AI governance framework in the world. Yes — Targe's workspace compliance review and playbook engine are structured around the four functions.
A policy document is a snapshot in time. Targe is the system that enforces the policy at every idea, every model change, every gate and every deploy — and writes the evidence down as it happens. When a regulator asks for proof, you point them to the append-only audit log, not a Word file.
Yes. Targe grades every registered AI initiative against POPIA, GDPR, EU AI Act, ISO 42001, NIST AI RMF, King V and Michalsons — automatically and continuously. Turn on the frameworks that apply to you and every project inherits the strictest applicable control set.
Every company is racing to use AI. Far fewer are governing it. Here's the gap Targe closes.
The short list every CIO and risk owner should put to a proposing team — in under fifteen minutes.
The difference between a calm regulator meeting and an emergency board call.
OKODI