A vetted bench of senior AI architects — security, engineering, analytical, design. Real engagements. Sponsored path to the Claude Certified Architect Foundations (CCAF) credential.
Every wealth-creation moment in history has produced a class of practitioners who actually built it. Kimberley had its diggers and its geologists. Johannesburg had its deep-level mining engineers. Silicon Valley, in the late 1990s, had its web engineers and product builders.
The AI age needs the same — and the craft has not yet been named. We call it architecting: the practitioner who decides what to build, how to structure it, what to make it from, and what it should not become.
The TBL Talent Network is the bench we are building around the South African Claude practice — vetted senior architects, working on real customer engagements, on a sponsored path to a credential the company that builds the model recognises.
The same person rarely does more than one of these well. A practice that cannot field all four cannot deliver complex AI engagements.
Design AI deployments that survive a regulator's scrutiny. Prompt injection, training-data leakage, agent privilege escalation, supply-chain attacks. The practitioners SA's regulated industries will trust to sign off on production AI.
Take an AI system from working demo to production deployment that survives a Black Friday traffic spike, a SARB audit, and an external pen test. Streaming, retries, observability, evaluation harnesses, cost governance.
Turn institutional knowledge — operational manuals, regulatory decision history, MFMA reports, clinical pathways — into machine-readable structure that an agent can reason over without hallucinating. Part librarian, part ontologist, part domain expert.
Decide how a human and an agent share a workflow — when to ask, when to act, where the trust seams are. Interaction design taken seriously in a context where the system is no longer deterministic.
Show evidence, not just a CV.
A repo, a deployed agent, published prompts, a written architecture decision, a body of teaching. If you have shipped something with Claude, lead with it.
Write clearly.
Most of what we deliver is judgement, expressed in writing. If you can write a clean technical memo, the rest is easier.
Have taste.
Knowing when not to build an agent is more valuable than knowing how to build one.
Care about contextual intelligence.
TBL's moat is using AI inside the regulatory, infrastructural, and behavioural realities of African markets. Local depth is a feature, not an inconvenience.
Frameworks before problems. Leading with methodology before you have understood what is actually being asked.
AI as a generic horizontal. Treating it as a tool that fits everything, rather than something to match carefully to context.
Certainty before commitment. Needing every variable resolved before you'll engage with the work.
Impatience with operational reality. Public-sector pace, informal-economy ambiguity, regulated-industry due diligence. These are not bugs.
This is not a judgement about who is a good engineer — it is a shape question, and being honest about it up front is fairer to everyone.
Anthropic — the company that builds the Claude family of models — has begun shaping a structured certification path for AI architects working with frontier models in production. The first technical milestone is Claude Certified Architect Foundations (CCAF), earned at the organisational level by teams who complete the Claude Partner Network learning path through Anthropic Academy.
TBL is on the Claude Partner Network certification path with Anthropic. We are not yet a partner — that designation belongs to a milestone Anthropic has not yet conferred — but we are sponsoring our team's enrolment in the Anthropic Academy CPN learning path, and designing the customer engagements that produce the artefacts CCAF asks for.
For practitioners who join the network and step up to live engagements with us, TBL sponsors the path. The credential is earned, not granted — but we pay the cost of getting there, and we put you in front of the work that makes it real.
The Honest Framing
A serious credential, recognised by the company that builds the model. A pipeline to paying projects. A network of senior practitioners around you while you build. That is what is on offer.
We read every application in full. We will not always have an immediate engagement match — but we keep serious responses on file, come back when one matches, and treat the relationship with the discipline a senior career deserves.
A confirmation is on its way to your inbox. We read every application — usually within a few working days.
Read the article that started this →