Grisha Pavlotsky, Chief Transformation Officer at Miro.
Think about the last great decision your team made. Not a routine decision, but one where the room was split, each path was riddled with difficult trade-offs and you still landed somewhere smart. Now, try to reconstruct that same process a quarter later. You have the output (decks, docs, call transcripts and spreadsheets), but the thinking is gone. Who pushed back and why? Which alternatives did you seriously consider and reject? That reasoning lived in meetings, threads and whiteboards, and then disappeared.
I’d argue that vanished reasoning is more valuable than the decision itself. I call it the “golden context,” and it’s something AI can help enable that elevates “fast employees” to a fast company.
I run transformation for a living, and this is the pattern I see everywhere. Even if the company is moving faster with AI, it’s typically lagging behind how fast its people are moving with the technology. And that is where the gains get diluted: in the hardest, most valuable part of the work. Teams’ reasoning, debate and decision-making rarely gets captured, so it never compounds.
Every team starts its next hard decision roughly where the last one started: from zero.
Deliverables: The Tip Of A Very Valuable Iceberg
Let me be specific: the golden context is the evolution of the work that produces deliverables. It’s who argued what, which counterargument actually landed, what evidence the team relied on, which alternatives you evaluated, and which one won and why. It’s the “how did we get here,” captured with enough fidelity that someone (or something) could learn from it later.
This matters more than it used to, because in a world of constant iteration, the deliverable is rarely the final thing. It’s usually just the first version. The real question is how fast you get to a better version two, and you can’t do that well if all you kept was the output. To improve version one, you need the story under the hood. Keep only the deliverable, and every new version risks being a fresh guess. Keep the reasoning, and every version gets closer to being right.
Many of the most AI-native companies already understand this, at least in pockets within their organizations. Engineering teams write decision records that capture not just what was chosen but the alternatives that were rejected (and why) so nobody re-litigates a settled question after the person who settled it has moved on. Amazon famously made people write narrative memos instead of slides precisely because long-form forces the reasoning into the open, where a deck lets you hide it.
These aren’t just documentation habits. We are past that now that AI can power so much of our work if we can feed it good context.
Why Context Is Suddenly A Competitive Advantage
The instinct to capture institutional knowledge isn’t new, so it’s fair to ask what is different now. There are two things:
1. The cost of not capturing it has changed.
When individual productivity gains don’t roll up into organizational speed (and mostly they don’t), the uncaptured process becomes the actual bottleneck. The output balloons, the throughput barely moves and the reason is that the decisions and alignment around the work are still being redone by hand every single time.
2. AI finally makes captured reasoning usable.
The archive that used to sit dead in a wiki is now something you can feed to a team or an agent and get leverage back. The model your competitor uses is trained on the same public internet as yours. The way your company actually argues and decides exists nowhere on that internet. Once captured, it becomes a real advantage because you are compounding something no one else even has.
How To Build Speed
You don’t need a new mandate or a two-year program to begin. Start by treating the reasoning behind important decisions as an asset worth preserving, then do three things:
1. Spot where it is evaporating.
Name the two or three places your real decisions actually get made. Obvious places are the weekly leadership call, the design review, the deal strategy thread, etc. Then be honest about what is (and isn’t) being captured in a usable form. A Slack thread may preserve fragments of the discussion, but without the surrounding context, alternatives and rationale, it rarely preserves the reasoning well enough to reuse.
2. Capture the reasoning, not just the decision.
One of our operating principles at Miro is that everything should be machine-legible. You’ll want to capture debates, evidence, alternatives you killed and the reasoning for your final decision. That context will live across Granola transcripts, Slack threads, decks, spreadsheets and meeting notes. It’ll be annoying to stitch it all together at first, but we’re all moving towards a future of ambient context where this becomes much easier.
3. Put it to work.
Captured context that just sits there is exhaust. Feed it forward to the next team so they start from your best thinking instead of from scratch. Then, increasingly feed it to your agents so the organization learns at the speed of its best work rather than the pace of its slowest handoff. That’s where the compounding can start.
And while you’re at it, communicate to your team that this is shared memory to make the next team smarter, not surveillance to score people. Capturing context is an investment in the people who will have to make the next decision.
Conclusion
None of this looks dramatic from the outside. There’s no launch or transformation banner. It starts with the decision to treat the reasoning behind your work as something worth keeping.
The companies that make that decision are the ones I believe will pull ahead, one decision at a time, while everyone else keeps solving the same problems from zero.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
