- A senior data scientist hire takes 2-4 months to close and ramps for another month after that.
- Most early AI questions are strategy questions, not modeling questions — consulting fits better.
- In-house makes sense once you have 3+ ongoing AI use-cases needing daily iteration.
- The two aren't mutually exclusive — consulting first, then hire once the roadmap is proven.
Most early AI problems are strategy problems
In the first few months of exploring AI, the real bottleneck usually isn't modeling skill — it's deciding what to build at all. That's a strategy and product question, which a short consulting engagement answers faster and cheaper than a full-time hire whose job description assumes the roadmap is already settled.
When in-house actually wins
Once a company has three or more live AI use-cases that need daily iteration — retraining, monitoring, prompt tuning — an in-house hire starts to make sense, because the ongoing maintenance cost outweighs a consultant's day rate. Before that point, the volume of work rarely justifies a full-time salary.
The two aren't rivals
The strongest pattern we see: a consulting engagement defines the roadmap and ships the first use-case, and the hiring decision gets made afterward with real data instead of a guess. That sequencing avoids hiring for a role whose scope is still undefined.
What to actually compare
Compare total cost over six months, not the headline day-rate versus salary number — include recruiting cost, ramp time, and the cost of a wrong first hire. Most founders find the numbers are closer than they assumed, and the deciding factor becomes speed to a working roadmap, not price.