How to get ahead of the AI shockwave

The past few years have seen a deluge of agentic development proclaiming to upend everything from protein folding to self-driving cars, leading to pretty much everyone demanding their organisation somehow exploit this new magic by shoehorning it into everything from outreach engines to running shoes.

In reality, 95% of these efforts fall by the wayside — busy-work disguised as game-changing in an effort to meet impossible expectations driven by hype rather than need.

But — and there is a but — that 5%. That 5% is where genuine, industry-changing value is found. Think chat agents, material engineering, medical science, big data.

It's that last part where the interests of Red-Shift and Sayvr meet: building a completely new ecosystem designed to link diet, goal setting, cooking skill, inventory management and the unavoidable time-sink of shopping into a single, holistic experience tailored to the individual.

Building an integrated and harmonised AI data layer

Sayvr's mission was nothing short of gargantuan. Take all of these elements, map them, figure out how to seamlessly unify them, and build it all out at once to meet near-impossible user expectation first time. Clear direction and definitive clarity were essential.

Luckily, we're quite good at this here at Red-Shift, and began by walking the team through our Delivery Governance Framework.

The first cycle was based on Daniel Bond's (Head of Engineering) proposed architecture, which allowed us to get several layers deep — from the "What" to the "How" and all the way down to the "When". Which is when we realised we'd overshoot our timeline by a factor of 4. Sub-optimal.

A whiteboarding session was needed. Cue the founders, Sam and Alex, frantically ideating against a whiteboard, explaining the long-term vision and bringing that down to a tangible near-term objective for us to refocus on.

You don't need to boil the ocean to make tea

This led us to completion of the second cycle of the process, which output a plan some 6× smaller than the first. Narrowed, refined, focused — now we're talking.

The result is a shared understanding of exactly what we are doing, how we are going to do it, when it's getting done and who's doing it. At a high level:

• Pull the idea out of the founders' minds and write it down. Send it back to them as a formal document — which naturally requires iteration.
• Write down the how — a similar exercise from an engineering lens, abstracting the high-level concept into Lego blocks of explicitly-defined deliverables, validated by the Head of Engineering.
• Move onto the when. Size those Lego blocks as week-sized multiples — one week, two weeks etc. You now have a schedule.
• Send it back to ensure a shared understanding of what the road ahead looks like in detail. Leadership can interrogate the proposal rather than be told what it is.

Show — don't tell. And voilà: you now have a legitimate delivery schedule. Now you can start to build.

Breakdown. Define. Align. Measure.

Taking a completely novel idea and refining it down to the task level — code can be written against it — should be the default practice. In reality, it rarely is. CEOs dictate, Product Managers fail to challenge, Engineers interpret, and at the end you end up with some horrific mutation, begging to be put out of its misery.

The secret ingredient is leadership. This exercise can only survive with CEOs and founders who understand their organisations in depth and have the courage to be challenged. This level of clarity is only possible when leadership not only respects the necessity of friction, but revels in it.

The result is genuinely groundbreaking progress — the kind that wins funding, upends industry norms and establishes brand-new markets. Who said software development can't be fun?

Founder's perspective — Samuel Day