Why AI Projects Take 2-3× Longer Than Planned

The demo is 20% of the work; the eval and data iceberg is the other 80%, and it's invisible in every plan. How to estimate AI projects honestly.

Mert Mutlu·Founder & CEO, Aiporate··6 min read·Share on XLinkedIn

Key takeaways

  • The demo is 20% of an AI project; evals, data work and edge-case handling are the 80% that never appears in the plan.
  • Fast demos actively mislead: a two-day prototype resets stakeholder expectations for a six-month system.
  • The last-mile is nonlinear: going from 80% to 95% quality costs more than going from zero to 80.
  • Data work is always underestimated, assume your data is worse than you think, because it is.
  • Honest plans budget eval-building as a first-class workstream and quality as iterations, not a milestone date.

AI projects take two to three times longer than planned because teams estimate the demo, which is the visible 20%, and discover the iceberg, evals, data cleanup, edge cases, guardrails, in production month three. This isn't pessimism and it isn't incompetence; it's a systematic estimation error that repeats because the demo arrives so fast it recalibrates everyone's expectations in the wrong direction. The fix is not padding. It's planning the iceberg explicitly.

The iceberg, itemized

  • Eval construction: collecting real cases, defining 'good', grading outputs, weeks, and it gates everything else.
  • Data reality: the 'clean' data has duplicates, gaps, format drift and permissions questions nobody owned.
  • The edge-case long tail: the demo handled ten happy paths; production has ten thousand, each cheap alone, brutal in aggregate.
  • Guardrails and failure handling: what happens on wrong output is a product decision, then an engineering effort.
  • Integration and permissions: wiring into real systems, auth, logging, cost controls, invisible in the demo, mandatory for launch.
  • Nondeterminism tax: every fix must be re-verified against the whole eval set, because improvements regress other cases.

Why smart teams still miss it

  • The demo's speed anchors everyone: if 80% took two days, 100% feels like a week away. The curve bends the other way.
  • Traditional software intuition assumes quality rises linearly with effort; model quality plateaus and fights back.
  • Nobody budgets for evals because nobody was ever asked to define 'good enough' numerically before kickoff.

How to plan honestly

  1. 1Build the eval set first and put it in the plan as a named workstream with an owner.
  2. 2Multiply your demo-based instinct by 2.5, that's not padding, it's the observed base rate.
  3. 3Plan quality as iteration cycles ('six eval-improve loops') rather than a date on which quality occurs.
  4. 4Ship at the honest bar with guardrails and a human fallback, then improve in production, waiting for 99% is how projects die.
  5. 5Report progress in eval scores, not percent-complete, it's the only number that doesn't lie.

Frequently asked questions

Why does the last 20% of an AI project take so long?

Because model quality improves nonlinearly: each gain requires finding failure modes, fixing them without regressing others, and re-verifying against the full eval set. The long tail of edge cases is where most of the calendar goes.

How should we estimate an AI project timeline?

Estimate the demo, then multiply by 2.5. Better: plan the iceberg explicitly, eval construction, data cleanup, guardrails, integration, as named workstreams, and express quality targets as eval scores with iteration cycles budgeted to reach them.

Should we wait until the AI is near-perfect to launch?

No. Ship at an honest quality bar with guardrails, confidence thresholds and human fallback, then improve against production data. Real usage improves the system faster than lab iteration, and waiting for near-perfect is the most common way these projects die.

MM

Founder & CEO, Aiporate

Mert founded Aiporate to close the gap between AI adoption and AI-native capability. He writes on how organizations should reorganize around AI, and on what it actually takes to hire, vet and ship AI talent.

Need the team to make this real?

Describe your need in plain English, get the exact hire, forward-deployed talent or a fractional leader, vetted and matched in 72 hours.

Scope your need →

Keep reading

The Weekly Brief

Intelligence for building AI-native organizations.

One email a week: the sharpest thinking on AI hiring, infrastructure, teams and strategy, for the people building the future of work.

Join operators, founders and CTOs. No spam, unsubscribe anytime.