Strategy

Determine how AI and data science is relevant to your business, and plan a path forwards with the help of a seasoned independent expert. A lightweight way to get started.

Deep dive into your organisation, it's products, customers and datasets. Identify and rank opportunities according to your organisational goals, answering questions such as:

• Where are the most valuable AI and data science opportunities for my business? Which should be prioritised and which should be parked?
• What will the likely return on the investment be, and what are the stages to get there? What are the recognised best practices and processes?
• What off-the-shelf services and providers are available? What are their strengths, risks and limitations? Are they a fit for your opportunities? How should you select between them?
• Will bespoke modelling or software development be required? Or integration with existing systems? Which technology choices offer the best fit?
• What skills will be needed? Will you require AI, machine learning or statistical models? Data engineering? Software development? UX/UI Designer? Product management?
• Should you train, hire or outsource? Which of these roles can be combined? When should you recruit and at what level?
• Are there human or business change elements involved in realising your opportunities - how should these be achieved?

We combine deep technical knowledge with years of experience prioritising, testing and productionising real world products. These sessions enable you to benefit from our experience up front when planning and prioritising your approach.

Through a combination of workshops, interviews and rigorous lab work - we will develop a comprehensive assessment of your opportunity and a path forward.

To understand modelling effort and feasibility, we will:

• Identify the desired business outcomes and translate these to technical requirements and recommended modelling approaches.
• Recommend appropriate model performance metrics and testing strategies, and identify the minimum viable performance required.
• Identify any additional modelling constraints, such as the creation of relatable confidences, error bounds or automated model explanation.
• Identify potential risks (e.g. the impact of model errors, or historic bias in training data) plus mitigation strategies where possible (e.g. human-in-the-loop processes or model un-biasing techniques).
• Perform a detailed data review (may be based on samples of representative or anonymised data) to understand quality and fitness of the data.

Where applicable:

• Run preliminary models to identify baseline performance.
• Test identified business hypotheses against the data. e.g. the predictive power of particular features.
• Perform external data discovery, to assess availability of free or paid datasets and pre-trained models.
• Assess current state-of-the-art performance in literature.

To understand the downstream effort required to realise benefits (productionise the solution) we will:

• Identify the mechanisms for benefit realisation. e.g. sales of newly created product, increased uptake of existing products.
• Identify high-level solutions. e.g. creation of predictive APIs, development of cloud hosted client-facing applications, or interactive internal dashboards.
• Identify requirements for new data pipelines for input, feedback loops and model retraining. (e.g. real-time vs batch).
• Identify associated human and business change requirements (e.g. uptake of new AI driven processes) and recommended rollout approach.

Whether or not you choose to deliver your opportunity with us, following our assessment you will have a clear understanding of your opportunity, the relative effort and complexity, and a plan to proceed.

Making the right strategic and planning decisions and investments around AI and ML projects is something best done early, particularly when it comes to knowing and planning for what is possible and collecting the right data from day one.

Having access to a seasoned chief data scientist with over a decade of real world experience, to help guide and advise on AI, ML and analytics projects, can make the difference. Save months of effort, and prevent a multitude of common mistakes and pitfalls - to get the most from the resources you have today.

An experienced data scientist and machine learning engineer can solve by experience and intuition what would take less experienced months to achieve by trial and error.

• What data to collect early on, and when and where to introduce AI.
• What external datasets and transfer learning methods can be used to realise opportunities early or with little data.
• When it is better to use pre-trained services (i.e. from cloud providers) rather than building in house.
• When it is worthwhile to 'go deep' on a problem versus when simple is most effective. i.e. 80/20.
• Early no-nonsense view of which opportunities are possible / not possible.
• Which opportunities should be prioritised and when.
• How to communicate about AI both internally and externally. Evangelise Data Science and AI. Explaining models and model outcomes, when is model explainability more important that model accuracy?