06.1
“We need better models, or better tools.”
You need deployment discipline. The models improved dramatically over the same years the failure rate stayed where it was.
01
The gap is never the model.
Most AI never generates value. Not because the models don’t work — they got dramatically better — but because nobody owns getting them into production. AI that never ships never pays.
Get your AI readiness score →2 minutes.
>80%
RAND · 2024
~30%
Gartner · 2024
95%
MIT NANDA · 2025
02
This is not a rumour. It is measured.
>80%
of AI projects fail — twice the failure rate of non-AI IT projects.
~30%
of GenAI projects were forecast to be abandoned after proof of concept by the end of 2025.
95%
of enterprise GenAI pilots show no P&L impact, self-reported.
The five in a hundred that report P&L impact.
02b
RAND · 2024
Gartner · 2024
MIT NANDA · 2025
03
The demo and production are different sports.
Clean samples. Dirty data from systems nobody has cleaned in a decade. Nothing to integrate. Legacy ERP and the internal systems around it. No permissions. Row-level permissions and your IAM. No review. Security review, vendor review, change control. No SLA. Uptime, on-call and someone to page at 03:00.
The demo
Production
03.1Clean samples.
Dirty data from systems nobody has cleaned in a decade.
03.2Nothing to integrate.
Legacy ERP and the internal systems around it.
03.3No permissions.
Row-level permissions and your IAM.
03.4No review.
Security review, vendor review, change control.
03.5No SLA.
Uptime, on-call and someone to page at 03:00.
Five constraints the pilot dodged. Every one of them is an engineering problem inside your environment, and none of them can be solved from outside it.
Nobody owns the last mile. That is the whole story.
04 — Why this exists
I watched this happen from the inside, project after project.
I spent close to ten years at Microsoft as a Forward Deployed Engineering Manager — the engineer who goes into the customer’s environment when the model already works and the value still has not arrived.
Across more than 100 projects the pattern did not change. The modelling was rarely the hard part. The hard part was the last mile: access, permissions, the legacy system nobody wanted to touch, and the question of who owned the outcome after the demo.
The projects that shipped had one thing in common. Someone was inside the building, accountable for production, from the first week. That is not a methodology I invented. It is the one that held across 100+ projects and over $500M of business impact.
— Miguel Fierro, founder of Datacean
05 — The method
Embed where the problem lives. Ship in 30. Production in 90.
Embed where the problem lives.Day 0306090First thin slice live in 30 days.Production in 90.
- Embed where the problem lives.The work happens inside your environment — your repos, your cloud, your identity provider. Never in a vendor sandbox.
- First thin slice live in 30 days.Real users, real data. Working software, never a status deck.
- Production in 90.Permissions, audit and security review done, then cutover with runbooks and handover.
- Measured on a business metric.Revenue in, cost out, or risk down. Shipping is the toll gate, not the finish line.
- Your team runs it.Enablement is how the engagement ends, by design.
05b
The market already converged on this model.
- Palantir1,300+ bootcamps in 2024 — five days, the customer’s own data, a working use case by Friday. Two decades of forward deployed delivery, now its core sales motion.
- OpenAIA dedicated forward deployed engineering unit. They hold the best models on earth and still concluded they need engineers physically inside customer environments to reach production.
- C3.ai · Scale AIThe same delivery model — engineers deployed into the customer’s environment, paired with the customer’s own team, across heavy industry, defense and finance.
Note
Cited as evidence the model works — these are not Datacean clients.
06
Three objections, answered straight.
06.2
“Our people can’t do this.”
They can be trained to. IKEA automated 8,500 customer-service roles, reskilled those people as interior design advisors, and generated around $1.4B in new revenue with zero layoffs tied to the automation. AI took the repetitive calls; humans moved up the value chain.
8,500 roles automated
Reskilled — interior design advisors
~$1.4B new revenue0 layoffs
Source link pending — figures approved, citation to be attached before ship
06.3
“We have no room for a year-long programme.”
Neither do we. Working software in 30 days, production in 90, and a decision point at every gate.
07
Two ways to close the gap.
We ship it with you.
AFor enterprises whose AI is stuck between pilot and production
Senior engineers embedded in your environment, accountable for a production outcome on your stack.
Five constraints from 03, closed. Same five rules, no breaks.
We teach you to ship it.
BFor leaders training a team, and engineers building the career
The same craft, taught hands-on: training for technology teams, and a track for engineers who want to become forward deployed.
08
Not sure which fits?
Get your AI readiness score in two minutes. You get a straight diagnosis of what is actually blocking you, and the one next step that matches it.
Get your AI readiness score →2 minutes. Your score on the spot, and the plan by email.
09 — Who is behind this
[FOUNDER: photograph]
The person who ships the work is the person who teaches it.
Miguel Fierro spent ~10 years at Microsoft working on AI as a Forward Deployed Engineering Manager — participating in over 100 projects deploying AI workloads inside customer environments and generating over $500M of business impact. He is the creator of Recommenders, the most popular open-source recommendation library on GitHub. He has made over 200 interviews for AI profiles. He holds a PhD in Robotics (UC3M with King’s College London, best doctoral thesis award) and executive education from MIT Sloan. He delivers personally.
- Forward Deployed Engineering Manager, Microsoft
- 100+ projects deploying AI inside customer environments
- $500M+ of business impact
- Creator of Recommenders
- 200+ interviews for AI profiles
- PhD in Robotics, UC3M with King’s College London, best doctoral thesis award
- MIT Sloan executive education