Discovery Process
Discovery creates enough clarity to decide whether we should build, test a critical assumption first, pursue a different approach, or stop.
Every engagement begins with discovery. Through focused conversations with the people closest to the work, we understand the relevant workflows, users, challenges, data, constraints, and desired business outcomes.
Our Forward Deployed Engineering approach keeps this hands-on. Engineers work directly with your team so assumptions can be tested early and technical feasibility can be assessed alongside the real operating context.
For organisations getting started with AI, we identify the key Jobs to Be Done, map the workflows around them, and determine where AI could improve efficiency, quality, customer experience, decision-making, product capability, or revenue.
Where AI already exists, we assess the current architecture, prompts, models, retrieval systems, workflows, data, integrations, infrastructure, and operating costs. We establish evaluations where needed so improvements can be measured objectively.
Data sensitivity, security requirements, regulatory obligations, deployment restrictions, latency expectations, operational ownership, acceptable error rates, and ROI are established before architecture decisions are made.
When discovery is not enough
Some opportunities contain a critical assumption that must be tested before either side can confidently commit to implementation. In those cases, we move to a focused Proof of Concept.
The POC validates the core hypothesis: can the proposed approach deliver the required outcome at the level of quality and reliability the use case demands? It does not recreate the full production system; it resolves the uncertainty that matters most.
When a paid pilot is required
Fine-tuning, complex data pipelines, model experimentation, agentic workflows, substantial integrations, specialised infrastructure, or production-scale performance testing may require deeper validation under realistic conditions.
A paid pilot creates the room to reduce both technical and commercial risk before full implementation. Where agreed, its cost can be credited towards the larger project. We increase investment only as confidence increases.