Why we’re building AISOD 3A

FOUNDER’S MANIFESTO · JOEL TIAGO · VERSION 1.0

People should have a practical way to turn their ideas into something real. Africa should help build the intelligence that shapes its future.

By Joel Tiago, founder and CEO of AISOD · @joeltiagoofficial. Based on my statement dated 30 September 2026; edited for this archive on 24 October 2026. This is a founder’s perspective and development agenda, not a peer-reviewed research finding.

From intention to execution

People have ideas, ambitions and problems they want to solve. They may want to build an application, grow a business, learn a skill, find opportunities or create more time for their families. The obstacle is often the distance between knowing what they want and knowing how to begin.

That distance matters to me. An unrealised idea can mean a business never started, a useful service never built or a person whose ability never finds an outlet. AISOD’s purpose is to help people cross that distance with tools they can understand and direct.

We are building AISOD 3A as an AI automation harness: the layer that organises models, tools and tasks around a human goal. A useful system must do more than produce an answer. It should help plan the work, gather evidence, produce an artefact and evaluate whether the result meets the request. The human must still be able to inspect the work, change direction and decide what happens next.

Four commitments

Own the harness

We want the ability to shape how AI works: how a goal becomes a plan, which tools may act, how outputs are checked and where human approval is required. AISOD 3A builds on Hermes Agent, with AISOD’s orchestration and project direction layered on that foundation. Building local capability includes crediting and understanding the technology we build upon.

Own the data

Our ambition is to build and govern useful African datasets with clear provenance, appropriate permissions and human quality review. Data stewardship means respecting the people and institutions behind the material. A user’s conversation should not automatically become training data merely because it passed through a system.

Choose the model

African organisations should be able to choose models for their task, budget, language and operating conditions. External models can be useful tools. Sovereignty grows when an organisation understands those dependencies and has credible alternatives, including locally operated and adapted open-weight models where they meet the need.

Keep humans responsible

Automation should expand human agency. It should not require people to surrender judgment. Important decisions need accountable people, and actions affecting others need clear authority. The goal is a system that helps us act thoughtfully, not one that obeys without limits or hides its mistakes.

What is available, and what comes next

AISOD 3A Harness: available, according to my current confirmation as its founder. Readers can visit 3a.aisod.tech and inspect the public MIT-licensed repository. The repository’s older staged-release language has not yet been updated to reflect this availability. This article does not report an independent functional test of every integration.

AISOD Chat: available internally, with a public release planned for December 2026. Internal availability is distinct from a public launch. The slide deck presents offline operation as a use case; this archive has not independently benchmarked offline performance across devices or languages.

AISOD African Corpus: more than five million words, according to my current report, with an official release planned for January 2027. The corpus portal is an authentication entry point, not a public dataset download. The reported word count is not an independent audit or a measure of language quality, rights clearance or model performance. Public access terms and licensing should be checked when the release is announced.

These release dates are plans as of 24 October 2026. They are not claims that AISOD Chat or the corpus have already been publicly released.

The future I want for AI in Namibia

Namibia can participate as a builder of AI systems, not only as a buyer of finished services. My proposal is to start with work that people already understand: helping a small business prepare a quotation, organising records, supporting a teacher’s preparation or finding information in a locally relevant collection.

Each use case should earn trust through evidence. Does it produce a correct result? How much human review does it need? Does it work with the language, device and connectivity available to the user? What does it cost to operate, and who can repair it? Those questions should guide deployment more than a dramatic demonstration.

For local-language AI, a large corpus is a starting point. We need competent speakers involved in curation and evaluation, documentation of sources and permissions, and tests that reflect real tasks. Training smaller models and exploring offline use are paths to investigate; their value must be demonstrated in the conditions where people will use them.

I want schools and universities to teach people to use AI while making their own thinking visible. A student should be able to explain a decision, show what the tool contributed and demonstrate understanding. Businesses and public institutions need the same discipline: know where automation helps, record the decisions and keep responsibility clear.

Why the human ambition matters

We do not know what people will build when useful tools become easier to access and control. That uncertainty is a reason to create room for them. I want more people to express an idea, develop it and take part in shaping their communities.

For me, sovereignty is the practical freedom to make informed choices about our technology. Africa’s future should include African builders, African languages and people with the means to solve the problems they know best. That is why we are building AISOD 3A.

Read my Coastal AI Summit account and the practical masterclass lessons · Explore the human side of AI

Sources, status and disclosure

A01: Joel Tiago’s supplied statement dated 30 September 2026 and direct clarifications on 24 October 2026 establish the founder’s intentions, availability report, corpus count and planned release dates. These are first-party statements.

A02: AISOD 3A Harness repository, inspected 24 October 2026: MIT licence, Hermes foundation and four-role architecture. Its older rollout wording conflicts with the founder’s newer availability update; that distinction is preserved here.

A03: Supplied masterclass slides 12–21 introduce the sovereignty argument, harness architecture, corpus pipeline and offline vision. Slide claims about corpus coverage and model performance are not independently validated by this article.

Joel Tiago leads AISOD and has a direct commercial and professional interest in its products. Codex assisted source review, editing and website production. This page does not claim independent human peer review. Source record · Version and correction history.