Why Assessment First
Every AI project I’ve seen fail had one thing in common: they skipped the assessment phase. Someone saw a demo, got excited, and jumped straight to building. Six months and six figures later, they had a system nobody used because it solved the wrong problem.
That’s why every engagement I take starts with a structured two-week assessment. It’s not optional, and it’s not a formality – it’s the single highest-ROI activity in the entire project lifecycle.
What Happens in Two Weeks
Week 1: Discovery
The first week is about understanding the landscape:
- Stakeholder interviews: Who are the actual users? What are their real pain points? (Not what management thinks they are.)
- Codebase audit: What systems exist? What’s the data quality? Where are the integration points?
- Process mapping: How does information flow through the organization today?
- Quick wins identification: What can we improve with minimal effort?
Week 2: Strategy
The second week turns discovery into action:
- Opportunity scoring: Each identified opportunity gets scored on impact, feasibility, and effort
- Architecture sketches: High-level technical designs for the top 3 opportunities
- Risk assessment: What could go wrong? What are the dependencies?
- Roadmap delivery: A prioritized plan with clear phases and milestones
The Deliverable
At the end of two weeks, you get a strategic roadmap document that includes:
- Current state analysis – where you are
- Opportunity matrix – what’s possible and what’s practical
- Recommended pilot project – the highest-value, lowest-risk starting point
- Technical architecture – how it would work
- Resource and timeline estimates – what it takes to build
This document becomes the foundation for every decision that follows. It prevents scope creep, aligns stakeholders, and gives everyone a shared understanding of what success looks like.
Why Two Weeks Is the Right Duration
Shorter assessments lack depth. You end up with surface-level recommendations that miss the real opportunities buried in legacy systems and established workflows.
Longer assessments lose momentum. The goal is to move fast enough to maintain organizational energy while being thorough enough to get it right.
Two weeks is the sweet spot I’ve found across dozens of engagements.
The Real Value
The assessment often pays for itself before a single line of code is written. I’ve had clients discover they were about to build a custom AI system when a well-configured off-the-shelf tool would do 80% of the job. Others found that their “AI problem” was actually a data quality problem that needed fixing first.
The best investment in any AI initiative is understanding what to build before you build it.