Skip to content
Cognitecta

Training / Leadership

Leadership · 1 day

AI for Technical Leaders

A decision-making course for CTOs, engineering managers, architects and senior technical product leaders. How to choose architectures, vendors and investments that will still make sense in production.

1 day · Leadership · Leadership · CTOs, engineering managers, architects and technical leaders

Talk to us

Overview

Technical leaders are being asked to approve AI work with incomplete information: vendor claims, promising prototypes, and pressure to have an AI strategy. The cost of a poor decision is not a weak demo. It is an architecture that cannot be operated, a team that cannot be hired for, or a use case that never had a production path.

This course is for people who have to make those decisions. It covers current capabilities and limits, then spends the day on the choices that actually matter: model selection, hosted versus self-hosted, open versus commercial, build versus buy, RAG versus fine-tuning, agents versus workflows, and the production concerns of security, privacy, latency, cost, evaluation and observability.

The teaching is architectural and commercial, not a coding class. You work through proposed projects, score them, and design a first version of an AI roadmap that is honest about skills, data, risk and operating cost.

By the end of the day, you can interrogate a proposal, choose a shape that fits your organisation, and explain why a prototype is — or is not — ready to leave the lab.

Audience

CTOs, heads of engineering, engineering managers, architects, technical product leaders and senior decision-makers who own architecture, budget or delivery for AI work.

Prerequisites

You should be comfortable with software delivery and organisational constraints. Direct machine-learning experience is not required. The course assumes responsibility for decisions, not a need to write model code.

Duration

1 day

Often run as a private session for a leadership team. Can be extended with an architecture workshop on a live problem.

Learning outcomes

  1. 01

    Assess current AI capabilities and limits in terms that survive an engineering review.

  2. 02

    Compare hosted and self-hosted models, and open versus commercial options, against control, cost, latency and data boundaries.

  3. 03

    Choose between APIs, RAG, agents, fine-tuning and conventional software for a given problem.

  4. 04

    Evaluate build versus buy, including the hidden cost of integrating a vendor’s abstraction.

  5. 05

    Identify the security, privacy, governance and evaluation work a production system will require.

  6. 06

    Estimate the skills, platform and operating model needed to move from prototype to production.

  7. 07

    Produce a short, defensible AI roadmap rather than a list of interesting experiments.

Outline

  1. 01

    A usable picture of current AI

    • What current models do reliably, and what they still cannot be trusted to do.
    • The difference between a compelling prototype and a system with owners, interfaces and failure modes.
    • Where latency, cost, context and evaluation become architectural facts rather than later concerns.
    • Exercise: interrogate a fluent vendor or internal claim and extract the actual capability.
  2. 02

    Architecture choices that matter

    • Model selection: quality, context, tool use, hosting, data handling and switching cost.
    • Hosted APIs versus self-hosted inference; open-source versus commercial models.
    • Build versus buy, including platforms that hide the wrong details.
    • RAG, fine-tuning and agents as different answers to different problems.
    • Exercise: choose an architecture for two contrasting proposals and defend the trade-offs.
  3. 03

    Production, risk and operating cost

    • Security, privacy, identity, data boundaries and model-provider risk.
    • Governance: what must be logged, versioned, approved and refused.
    • Latency, scaling, observability, evaluation and cost controls.
    • Team skills: what engineers, platform and product actually need to own.
    • Exercise: list the production work missing from a successful prototype.
  4. 04

    Use cases, platforms and the path out of the lab

    • Identifying worthwhile use cases: value, feasibility, data, risk and reversibility.
    • AI platform strategy: what should be shared, and what should stay in product teams.
    • Moving from prototype to production without freezing a demo architecture.
    • Exercise: evaluate a set of proposed AI projects and draft a 90-day leadership roadmap.

Practical work

You work in small groups on real or representative proposals. You score use cases, choose architectures, identify missing production work, and produce a short roadmap. The emphasis is on decision quality, not on producing a presentation.

Takeaways

  • Decision frameworks for model selection, build versus buy, and RAG versus agents versus fine-tuning
  • A production-readiness checklist covering security, evaluation, cost and operations
  • A scored view of proposed use cases
  • A first-pass AI roadmap for the organisation or product area

Delivery

Cognitecta delivers private corporate training, on-site or as remote live training. Courses can be run as published, or adapted to your organisation’s stack, domain and experience level.

Instructor

Nicholas Johnson, AI architect and software engineer. He has a degree in Artificial Intelligence and around twenty years of professional technology training, including hundreds of courses for engineering teams and large organisations. About.

Related courses

  1. Half day or 1 day · Leadership

    AI for Business Leaders

    A concise course for senior leaders and directors: where AI creates actual value, where it does not, and how to judge investment, risk and organisational readiness.

Talk to us