Skip to main content

Why Your AI Vendor Is Lying About Time to Value

"Production in 4 weeks" usually means a demo in 4 weeks. How to decode vendor timelines and set realistic expectations for enterprise AI delivery.
28 June 2024·3 min read
Isaac Rolfe
Isaac Rolfe
Managing Director
"We can have you live in four weeks." Every AI vendor says it. Almost none of them mean what you think they mean.
Here's the translation guide:
"Live in 4 weeks" usually means: a demo environment, running on sample data, accessed by 3 people on the vendor's infrastructure, with no integration to your systems, no governance framework, and no handling of your edge cases. It's a proof of concept dressed up as production.
"Production-ready" usually means: the vendor's platform is production-ready. Your data, integrations, governance, and user training are not. The gap between "vendor-ready" and "enterprise-ready" is typically 8-12 weeks of work that the vendor conveniently doesn't include in their timeline.
"Easy integration via API" usually means: they have an API. Connecting that API to your legacy claims platform, your CRM from 2015, your document management system, and your compliance reporting tool is your problem. Integration is 60-70% of enterprise AI work, and the vendor's timeline includes 0% of it.
"Works with your data" usually means: works with clean, structured, well-formatted data. Your scanned PDFs with handwriting, your inconsistent spreadsheets, and your email attachments named "final_v3_ACTUAL_final.docx"? That's a data readiness conversation they'd rather not have yet.
"Minimal training required" usually means: the interface is simple. The organisational change management (new workflows, new decision processes, new governance, new roles) is not simple and is entirely your responsibility.
How to decode any vendor timeline:
  1. Ask: "What's included in that timeline?" Get specifics: data integration, system connections, governance, user training, edge case handling.
  2. Ask: "What's our responsibility in that timeline?" The honest answer is usually: data preparation, system access, user training, governance, change management, and IT support.
  3. Ask: "Can you show us a production deployment at an enterprise with similar complexity to ours?" If they can't, their timeline is aspirational, not empirical.
  4. Double their timeline. If they say 4 weeks, plan for 8-12. If they say 8 weeks, plan for 16-20. Then be pleasantly surprised if they're faster.
The best AI vendors are honest about timelines. They'll tell you: "Our system deploys in 2 weeks. Your data preparation, integration, governance, and training will take 8-10 more. Let's plan the full picture." That honesty is a better signal than any demo.