Notes from the front line of AI delivery

What we learn putting a company's own intelligence into production at European mid-caps: the numbers, the failure modes, and the delivery model that gets you past the pilot.

  1. 27
    ROI & Business Case

    What Enterprise AI Implementation Really Costs (and What Drives the Number)

    AI implementation consultant cost is one line of six. What drives enterprise AI development cost at a mid-cap: integration, data, compliance, labour, change and run costs.

  2. 26
    AI Risk & Regulation

    EU data residency for Claude, OpenAI, Gemini and Mistral: what each offers and what it does not solve

    Claude data residency in Europe compared with OpenAI, Gemini, Mistral and Copilot: where prompts, agent logs and indexes sit, checked on 29 October 2026.

  3. 25
    Forward-Deployed Model

    Outcome-based pricing vs day rates: examples, and what the contract has to say

    Outcome-based pricing examples from TCS, McKinsey, HCLTech and Sierra: who carries which risk against day rates, what the contract must specify, and when not to use it.

  4. 24
    Forward-Deployed Model

    How to choose an agentic AI implementation partner: questions to ask before you sign

    Questions to ask AI vendors before you sign a build: ownership, the model swap, day 90, the EU AI Act provider and deployer split, exit, and a test for agent washing.

  5. 23
    AI-Native Enterprise

    AI use case discovery: why asking teams doesn't work (and a prioritisation matrix)

    AI use case discovery fails when you ask teams what they need: they ask for a hire. Show AI on their own process first, then rank use cases owned vs rented.

  6. 22
    Context Layer & Platforms

    Fabric IQ vs Palantir Foundry: which ontology should a European mid-cap's agents run on?

    Fabric IQ vs Palantir Foundry for a European mid-cap: what each ontology models, what leaves with you, the pricing basis, EU residency and GA status in October 2026.

  7. 21
    AI-Native Enterprise

    Nadella's Reverse Information Paradox, read by a European CIO: where your company's know-how should live

    The reverse information paradox: Satya Nadella says you pay for AI twice. What his remedy leaves on one cloud, a six-point exhaust audit and the exit clauses to sign.

  8. 20
    Context Layer & Platforms

    SAP Joule's limits: ECC, Microsoft Copilot, MCP and the handoffs it can't see

    SAP Joule limitations from SAP's own pages: ECC and on-premise access, MCP and A2A, Microsoft Copilot, AI Unit pricing, and the cross-system work Joule cannot see.

  9. 19
    Forward-Deployed Model

    Forward-deployed engineer vs consultant vs system integrator: a decision rule for buyers

    Forward deployed engineer vs consultant, system integrator or vendor services: who leaves a system in production, who carries the risk, and when each is right.

  10. 18
    Forward-Deployed Model

    Microsoft Frontier Company: Whose Intelligence Does It Protect?

    Microsoft Frontier Company launched in July 2026 with $2.5 billion and a promise to protect your intelligence. What that covers, and what you own if you leave.

  11. 17
    AI-Native Enterprise

    Salesforce fixed the CRM. Workday fixed HR. Nobody owns the handoffs.

    Agentic AI and the ERP: agents will not replace the system of record. SAP, Salesforce and ServiceNow agents each see their own system; the value sits in the handoffs.

  12. 16
    Forward-Deployed Model

    What Palantir AIP and Foundry Cost a European Mid-Cap

    Palantir's public prices are a marketplace placeholder and UK public-sector lists. What Foundry and AIP would cost a EUR 250M to 2B company, line by line.

  13. 15
    AI-Native Enterprise

    What Is an AI-Native Enterprise?

    An AI-native company is one whose business model breaks if you remove the AI. The removal test, AI-native vs AI-enabled vs AI-first, and what it takes.

  14. 14
    Forward-Deployed Model

    KPMG, EY and PwC all published AI-hallucinated reports in 2026. What it means for who builds your intelligence layer.

    Three Big Four firms withdrew or were caught publishing research with fabricated citations in 2026. The failure is structural, and it is a buying signal.

  15. 13
    Forward-Deployed Model

    Salesforce's own partners are not seeing Agentforce ROI: what that means before you buy a packaged agent platform

    Agentforce pricing from Salesforce's own pages, the ROI assumptions to check before signing, and why Salesforce's partners report no Agentforce-driven bookings.

  16. 12
    Forward-Deployed Model

    Accenture, Deloitte and EY All Run Forward Deployed Engineering Now: What Actually Differs

    Accenture, Deloitte, EY, IBM, Capgemini, Cognizant and TCS now sell forward deployed engineering. What separates them is which platform funds the pod.

  17. 11
    Forward-Deployed Model

    Accenture's worst day on the market: what it tells a CIO about the vendor they are about to sign

    Accenture shares fell as much as 20% in June 2026 on a soft outlook and AI demand pressure. What it means before you sign a multi-year AI programme.

  18. 10
    AI-Native Enterprise

    Own Your Intelligence: What It Actually Takes to Build, Not Just Fund

    Own your intelligence for a European mid-cap: what Sequoia's guide leaves open, sovereign vs private vs owned AI, and who builds it without an AI bench.

  19. 09
    Forward-Deployed Model

    OpenAI's DeployCo, Anthropic's Ode, and the case for an independent forward-deployed partner

    OpenAI Deployment Company, Ode with Anthropic, Microsoft Frontier Company, AWS and Google compared on model lock-in, exit terms, pricing and Europe. Who owns the result?

  20. 08
    Forward-Deployed Model

    Build vs buy vs embed: choosing an enterprise AI delivery model

    Buy commodity workflows. Build only with a bench you can permanently assign. Embed when the workflow is yours. The rule, including for the context layer itself.

  21. 07
    Pilot to Production

    Why 95% of Enterprise AI Pilots Fail (And the 5% That Don't)

    MIT found 95% of generative AI pilots show no P&L impact. Why AI pilots fail to scale, what the 5% do, and the four gates from pilot to production.

  22. 06
    ROI & Business Case

    The ROI of Enterprise AI Agents: 2026 Benchmarks

    AI agent ROI in 2026: executives expect 171%, self-reported GenAI returns of 152%. What each measures, and three questions before one enters your case.

  23. 05
    Forward-Deployed Model

    What Is a Forward-Deployed Engineer (and Why Enterprise AI Needs One)

    A forward-deployed engineer builds AI into production inside your team. How an FDE engagement runs, who is on it, what you provide and how it is priced.

  24. 04
    Governance & Control

    Keeping Humans in the Loop: Human Oversight for Enterprise AI Agents

    Which enterprise AI platforms offer human oversight for agents, what each one documents, what EU AI Act Article 14 requires, and how to tier approval gates.

  25. 03
    Governance & Control

    Audit Trails for AI Agents: What Regulated Enterprises Actually Need

    What an AI agent audit trail must log under EU AI Act Article 12, field by field, when the high-risk deadlines now apply, and how long the logs must be kept.

  26. 02
    Governance & Control

    AI-Native Operating Model: Rearchitecting vs. Layering AI on Top

    Layering AI on unchanged workflows stalls in pilots. When a European mid-cap should rearchitect instead, and what its target operating model must specify.

  27. 01
    Governance & Control

    Legacy System Modernization With AI Agents: The Mainframe and COBOL Angle

    AI agents vs human developers for COBOL modernization: agents map, translate and test, people own business rules and cut-over. Named results from Toyota, BMW and Itaú.

The future belongs to those who see it before it is obvious, and build it while everyone else is still out there scouting for the best AI tools.