Proof
Everything here is checkable. Case studies name what changed and by how much, name the client where we are allowed to, and flag the figures still awaiting confirmation.
Case studies
Across automotive, industrials, finance, energy, government, consumer technology, and sports. The problems repeat; only the regulator changes.
Automotive
$1.2BUnlocking $1.2B in new business
Delivery dates were missed often enough that customer confidence had become a commercial problem rather than an engineering one. Defects were surfacing in late-stage testing, contractual penalties were being paid annually, and the organization was competing for a major new contract it was not, on its record, credible to win.
A major automotive engineering organizationAutomotive
1,800 → 14From chaos to clarity
Eighteen hundred custom Jira projects had accumulated over years, each configured by whoever needed it at the time. No two behaved the same way, no change could be made without an unknown blast radius, and assembling a portfolio report took six weeks of manual work — by which point it described a company that no longer existed.
An electric vehicle manufacturerAutomotive
$20MPowering profits with digital precision
Digital parts content ran on a legacy system that was inefficient, error-prone, and — critically — could not produce an audit trail. That made compliance expensive, scaling impossible, and supplier billing effectively unverifiable.
A global OEMAutomotive
7×A factory of software excellence
Software quality depended on manual testing performed late, by people who were already the constraint on everything else. There was no engineering services function, no shared toolchain, and no way to coordinate release trains across teams that were nominally agile and practically sequential.
A global automotive companyAutomotive
430Unifying 430 engineers across 8 sites
The acquisition of 430 embedded development and hardware engineers from BlackBerry, spread across eight North American sites, delivered talent but not capability. Processes were inconsistent site to site, integration was more complex than modeled, and the acquired teams were not aligned to the strategic objectives that had justified buying them.
A major OEMConsumer Technology
100+How a recurring client problem became empwr.ai
Across engagements with automotive OEMs, Tier 1s, industrial manufacturers, consumer goods companies and silicon providers, we kept finding the same failure. The knowledge needed to run a large program existed, but only in fragments — in meetings nobody captured, in tools that did not talk to each other, and in the heads of whoever happened to be in the room.
Envorso — internal innovation, since spun outOriginal research
four pieces of original work, kept public and kept current — two recall datasets, the Atlassian ROI research, and the AI capability ladder.
Case studies show what happened when a client hired us. The research is the part you can use without hiring anyone — including the parts of it that cut against the argument we would rather make.
All four piecesSeries
Three parts on why organizations buy transformation and don’t get it — the willingness to change, the plans that fail anyway, and the roadmap that was obsolete before it was approved.
Part 1 of 3
Your biggest threat isn’t AI, it’s your comfort zone. The work itself has to change — this is about whether the organization is willing to.
ReadPart 2 of 3
Continues the series on why organizations buy transformation and don’t get it — and what has to be true for the next attempt to land.
ReadPart 3 of 3
Even a one-year AI plan is too long a horizon to bet on — the tooling you anchor to in month one is replaced by month twelve. Build capability, not a plan.
ReadPublic thoughts
Podcasts, talks and interviews — our operators making the argument on someone else’s platform, in front of an audience under no obligation to agree.
Article
Security Point Break · 2026-07
Why the Claude incident reads as an operational containment failure rather than a model-design flaw: the agent was executing its task aggressively while mistaken about where the simulation ended. The controls, not the model, are the gap.
Video
Yahoo Finance · 2026-05
Why Tesla’s camera-first robotaxi stack beats Waymo’s lidar on unit economics: cameras as weak eyes plus heavy AI processing put a $30,000 vehicle against a $150,000 one, and cost per mile decides who scales.
Article
Enterprise Management 360 · 2026-04
Energy storage is cushioning Tesla but is not yet large enough to offset both the credit cliff and automotive margin erosion — the trajectory is encouraging, the magnitude is not.
No deck, no obligation. We listen, we tell you whether we have seen this before, and we say what we think it would take. If a Jump Start is the right next step we will say so — and if it is not, we will say that too.