Building software used to look like building a house. Sequential, layered, each phase feeding the next, following a strict architectural blueprint. The cost was in construction, so the whole methodology – agile, waterfall, everything in between – was designed to reduce the cost of rework and make the build phase as efficient as possible.
The better metaphor now is a symphony orchestra. AI becomes the instruments. The Product Development Lifecycle (PDLC) team are the musicians. Everything runs simultaneously: design, development, testing, documentation, and the pieces must work in harmony or the whole thing falls apart. Unlike construction, we don’t need to fix the plumbing before hanging the drywall. It all should move at once.
We’ve all heard “AI is a 10x multiplier.” The problem with that framing is what it leaves out.
Give a guitar to Slash and you get something extraordinary. Give it to me and you get something that sounds passable. AI compresses that visible gap; I can now produce something that sounds like a seven out of ten. But the judgment behind it is still a three. I may not even know what’s wrong with it.
This is confident incorrectness at high velocity. It’s more dangerous than obvious incompetence, because the quality signal that normally triggers oversight gets suppressed. The output looks good enough to ship until it very much isn’t.
The implication: our most senior people need to be the ones leading adoption, not as a reward, but because they’re the only ones who can actually evaluate the output. Measuring throughput in this environment is a trap. The metric that matters is quality of output and cost of rework.
AI lets a single person play multiple instruments at once. We've seen it, and it's been genuinely valuable for understanding what's possible and where the limits are. Brandon has demonstrated it across a number of projects. I did something similar within the Aileen project, and was able to accelerate delivery in a meaningful way, precisely because I already understood the what, how, why, and when. The tools amplified the judgment I brought in, they didn't replace it. Both cases share something important: senior practitioners wielding the tools on their own terms, learning the instruments while composing the music — expertise is what makes that possible.
A 10/10 guitarist who’s also managing the drums, bass, and keyboard isn’t playing guitar at a 10. Divided attention is divided performance. As our projects grow in scope, stakeholders, and compliance complexity, the soloist trying to cover the entire orchestra stops being impressive and starts being a liability. The question isn’t whether it works for one person; we’ve proven it does. The question is what it looks like when we have more musicians, larger stages, and less room for error.
If you’ve been to a large concert, you’ve experienced the crew’s work without ever seeing them. The sound is balanced, the levels are right, the lights hit on cue — and none of it happens by accident. The crew is often bigger than the band, and they’re not a one-time setup; they’re there for every performance.
As we think about team structure, the temptation will be to frame AI purely as a headcount reduction play. That’s the wrong lens. Some traditional roles will evolve, others will consolidate, but these new crew roles are real, and organizations that don’t staff for them will find their AI investments underperforming. The question isn’t just what we can eliminate. It’s what we need to add.
An orchestra with brilliant musicians and no conductor still falls apart. Not because anyone is playing the wrong notes, but because without a unified interpretation, the music drifts.
The conductor in an AI-assisted organization is whoever can hold a coherent vision of what we’re building and why across simultaneous workstreams all moving at speed. That requires product judgment, architectural fluency, and enough AI literacy to hear when the instruments are drifting out of tune with each other.
This role is newly critical for a specific reason: sequential development naturally surfaces misalignment. Checkpoints catch drift before it compounds. When everything runs simultaneously, misalignment is silent until it’s expensive. The conductor isn’t a nice-to-have — they’re what keeps the ensemble from drifting into noise.
This paradigm shift isn’t unique to us. Every enterprise is going to have to figure out how to evolve, both in how they build and what they build, for a world where AI is the primary instrument of execution. Most of them don’t yet know what that looks like in practice, and the gap between knowing AI matters and knowing how to operationalize it is exactly where we’re positioned.
A new orchestra is forming with a conductor to lead it. We’re not waiting to figure this out. We’re already doing the work, and in doing so we’re developing the judgment, the patterns, and the early service lines that will define how we go to market. The enterprises that move with intention on this will pull ahead. Those that don’t will find themselves rebuilding from a position of disadvantage. We intend to be the people who help them move.
Elevate Intelligence Platform (EIP) evolves in this model from a set of internal observability tools into the hub of a broader customer engagement practice, with three initial spokes:
We’re not just adapting to this shift. If we move with intention, we’re positioned to be the people who help the enterprise world navigate it. The present opportunity is not simply to give customers better instruments, but to conduct ensembles that infuse those instruments with meaningful performance.
Not all development is the same, and the distinction matters for how we staff, scope, and measure work.
Enterprise legacy migration is closer to classical performance. The destination is largely deterministic. Compliance requirements function like time signatures; they are non-negotiable. The conductor’s job is precision and fidelity to spec. AI accelerates execution, but the score is written. The primary risk is the judgment gap surfacing inside high-stakes systems. Confident incorrectness in a migration path or a compliance-critical data model can be deeply costly.
Net-new product development, especially AI-native builds, is jazz. The structure is loose by design, because we’re discovering what the product is through the build. Musicians listen to each other in real time, respond to what’s emerging, and know when to solo and when to lay back. Sheet music written too early becomes a constraint rather than a guide.
But here’s the thing: even work that starts classical eventually has to become jazz. Legacy migrations are deterministic only until the technical debt is cleared. At that point we’re holding a clean, AI-native system and a blank canvas. If we only learn to play classical, we’ll be poorly equipped for what comes next.
The real competitive differentiator isn’t who can execute a defined score faster, because AI levels that playing field quickly. It’s who develops ensemble sensibility: the ability to improvise coherently, to build musicians who can both read written music and respond in the moment. Jazz looks like chaos from the outside. The best jazz musicians have deeply internalized theory, structure, and each other’s tendencies. The improvisation is only possible because of the underlying discipline.
That’s the disposition we need to develop now. Not just fluency with AI tools, but the judgment and ensemble literacy that makes the output sound like music rather than noise.
Everything we’ve been working through–the judgment gap, the crew roles, the classical vs. jazz distinction–our customers are going to have to solve for the exact same things. The organizations that figure this out first won’t just have a competitive advantage internally. They’ll have a repeatable formula.
Palantir Technologies spent twenty years proving a single insight: AI infrastructure only creates value when someone is embedded inside a customer’s organization to make it work in context. Their FDE model and the proprietary platform behind it were the moat. The reason no one replicated it wasn’t the concept; it was the cost. GraphRAG, vector databases, and modern LLM tooling have collapsed that barrier. What required a purpose-built platform and hundreds of engineers can now be assembled by a smaller, senior team. We’re positioned to bring that model to the enterprises Palantir never served.
What makes it defensible over time is how the model compounds. Each engagement surfaces patterns: where judgment gaps are most costly, what crew configurations work at different scales, what enterprises actually need versus what they think they need. Those patterns feed better tools and frameworks, which make the next engagement sharper. Over time, that builds into a continuously refined set of proprietary assets, both the internal PDLC infrastructure we run on and the licensable technologies we deploy for customers to fit their unique needs. Not a methodology deck. A system that gets better with every engagement.
This is where the Innovation Division becomes central to the story.
The compounding advantage only works if there’s a dedicated function responsible for driving it. The Innovation Division is where we run the experiments that define what the ideal crew looks like in practice, build and refine the tools and frameworks that make our model reproducible, and validate both current and emerging service opportunities before they enter the delivery funnel at scale.
Part of that work includes defining what the right crew actually looks like in practice. The roles emerging in the Sound Check section are a starting point, not a finished roster. The Innovation Division is where we pressure-test them, figure out which are essential from day one, which can be phased in, and what combinations work at different engagement scales.
AI Strategy & Enablement, Enterprise Knowledge & Intelligence Platforms, and Enterprise Legacy Modernization represent our earliest and strongest signal on where enterprise demand is forming, and we need to mobilize quickly to deliver on them. Each is an opportunity to validate whether what we’re building is reproducible across enterprise customers at scale, and to stress-test the model before we accelerate. The Innovation Division isn’t built around those three engagements alone, though. It’s built around the continuous process of discovering, pressure-testing, and productizing what comes next. As we deepen our engagement patterns and stay ahead of a rapidly advancing AI landscape, new opportunities will surface. It gives the structure to surface and pursue those opportunities intentionally, rather than discovering them by accident.
The division, in effect, is the ensemble that plays jazz first—developing the tools, building the repertoire, and refining the model, enabling us to bring the full orchestra to our customers.
--
See where your organization stands: take our AI Assessment to identify the gaps between experimentation and operational reality.