A frame we use with owners who are weighing whether to bring AI capability in-house or partner for it. There’s a right answer for each business — but it’s not the same right answer for every business.
The setup
A common conversation: the owner of a $25M service business is deciding whether to hire a “Director of AI” or work with a partner on retainer. The Director-of-AI candidates are asking for $180-250K base. The partner retainer runs $30-120K/year. The math seems obvious — until you look at it carefully.
Here’s the version of the math that survives scrutiny.
What you’re actually buying
The comparison isn’t apples-to-apples. The internal hire is one person; the partner is a team.
An internal hire gives you:
- One person, full-time, dedicated to your business
- Deep institutional knowledge (eventually)
- Direct supervision and culture fit
- A long-term capability — if you can retain them
A partner gives you:
- A team — typically 2-4 people across senior strategy, build engineering, and support
- Pattern recognition across many similar businesses
- Faster time to first production output
- Optionality — scale up and down without HR consequences
These are not the same thing. The pricing comparison isn’t “one apple vs. another apple.” It’s one apple vs. a basket of fruit.
The honest cost stack
Internal hire (Director of AI, mid-market):
- Base salary: $180-220K
- Loaded cost (taxes, benefits, equipment): add 30-40%
- Fully loaded, year 1: $235-310K
- Year 1 output: ramp 4-6 months before significant production work; 1-2 workflows shipped by year-end at best
- Year 2 output: 2-4 workflows shipped; cumulative capability emerging
- Hiring risk: real. Thin talent market. 6-9 months to recruit if your network isn’t deep.
- Retention risk: real. Senior AI talent has options.
Partner (operator tier, ~$5,000/mo + Sprints):
This is the AI Office model — a senior partner on retainer, not a single hire.
- Year 1 retainer: $60K
- Year 1 Value Sprints (typically 1-2): $40-100K
- Total year 1: $100-160K
- Year 1 output: roadmap by week 2; first workflow in production by day 45-60; 3-5 workflows shipped by year-end
- Capability building inside your team: an explicit goal of the engagement
- Hiring risk: zero (they already hired the talent)
- Lock-in risk: low (month-to-month, no minimum term)
For most mid-market businesses ($5M-$50M revenue), the partner produces more output, faster, at lower total cost — for the first 18-24 months. After that, the math shifts.
When the partner is the right call
Choose a partner if:
- You’re in the first 18-24 months of serious AI work. The partner ramps faster and produces more in this window.
- You don’t have a deep network of senior AI talent. Recruiting takes 6-9 months and the pool is thin.
- You’d rather build internal capability alongside the work than wait for a hire to ramp.
- You want flexibility to scale up or down based on what the work calls for.
- You’d benefit from pattern recognition across many businesses. The partner has seen what works at dozens of companies like yours; your internal hire has seen it at one or two.
This describes the majority of mid-market service businesses today.
When the internal hire is the right call
Choose to hire if:
- You’re past 18-24 months of serious AI work and the team is operating with a working portfolio.
- The AI work is mission-critical to your competitive position in a way that warrants single-throat-to-choke ownership.
- You have a specific person in mind who’s qualified and culturally aligned (rare, but real).
- Your AI workload has stabilized at a level a single full-time person can handle — typically when you have a mature workflow library and the work is more about operating than building.
- Your business is large enough ($75M+ revenue) that the loaded cost is small relative to the leverage the role produces.
The pattern we see most often at successful businesses: they partner for 12-24 months, then hire an AI Operations Lead internally ($95-130K, often promoted from operations), then continue a leaner partnership focused on senior strategy and specialized builds.
The hybrid model that works best
The most successful mid-market businesses don’t choose. They run a hybrid:
- One internal AI Operations Lead ($90-130K, often promoted from operations) — owns the workflow library, runs the cadence, manages adoption.
- One internal Prompt Librarian (part-time, a current employee) — owns the prompt and template library.
- A continued, lighter partnership — provides senior strategy, specialized build capacity, and outside perspective.
- Value Sprints when bigger builds emerge — engaging the partner’s deeper engineering capacity on demand.
This is where most of our 18-month-plus clients end up. It’s leaner than the year-1 engagement. The internal team owns more; the partner provides specific capabilities the internal team can’t justify hiring for full-time.
Total annual spend on this hybrid: $130-200K cash plus the internal $90-130K loaded. Comparable to what one senior internal hire would have cost — with materially more capability.
The mistake to avoid
The most expensive version of this decision is making it too early. Timing and foundation are exactly where AI initiatives die: the share of companies abandoning most of their AI initiatives jumped to 42% in 2025, up from 17% the year before, according to S&P Global Market Intelligence — and most of that waste traces to building before the foundation is there.
An illustrative, composite scenario (not a specific client): a $30M services firm hires a Director of AI at roughly $215K base ($280K loaded) before they have a single AI workflow in production. The Director spends months trying to figure out what to build first, ships one workflow that doesn’t land, and leaves inside the first year. The total cost of a failed hire like this — well over $200K plus the lost momentum — is the trap to avoid.
In a scenario like this, the same firm could have spent a fraction of that with a partner over the same period, shipped several workflows, and built the foundation that would have made an internal hire effective later. The hire wasn’t wrong. The timing was.
What we’d recommend right now
If you’re at a $5-50M service business and you’re early in AI:
- First 12 months: partner. Period. Speed matters, and ramp time on internal hires is too long.
- Months 12-18: evaluate adding an internal AI Operations Lead. The signal: your workflow library has grown to a point where day-to-day ownership matters more than build velocity.
- After month 18: hybrid. Internal lead plus a lighter partnership.
If you’re at a $50M+ business with AI work that’s already material, the hybrid is the steady state. The partner contribution shifts toward senior strategy and specialized work; the internal team grows.
A 30-minute conversation
If you’re weighing this decision right now, that’s exactly the right question for a 30-minute conversation. We’ll tell you straight what we’d recommend — including whether we think you should hire instead of working with us. We turn down work that isn’t the right fit. This is one of those judgment calls we’re happy to make honestly.