Should we train our team on AI first, or buy the AI tools first?
Start with tools first when your team is already curious, the work is mostly writing and research, and you can pick one approved tool quickly. Start with training first when people are skeptical, when client data rules are unclear, or when you have already bought licenses that nobody uses. For most firms the best order is a small step of each: choose one tool and set the rules, then train people on their real work within the first few weeks, before the licenses go stale.
The deciding question is whether your bottleneck is access or know-how.
Who this is for
Owners, managing partners, COOs and operations leads who know their firm needs to use AI and are deciding where the first dollars and hours should go.
The short version
- Buying tools first is fast and cheap to start, but access alone rarely changes how work gets done.
- Training first builds skill and confidence, but training without a tool people can use the next morning fades quickly.
- The real risk in either order is a long gap between the two. Licenses without training sit idle, and training without tools turns into theory.
What each option means
Tools first means buying licenses for an AI assistant such as ChatGPT, Claude or Microsoft Copilot, turning them on for the team and letting people explore. Training, if it happens, comes later. If you are still choosing a tool, see ChatGPT vs Claude vs Copilot.
Training first means teaching people how to use AI on their own tasks, what is safe to share and where it helps, before or while the firm decides which tools to buy and roll out widely.
How the two usually compare
These are typical patterns, not rules. Results depend on your team, your work and how each step is run.
| Factor | Tools first | Training first |
|---|---|---|
| Speed to start | Fast, licenses can be live in days | Slower, needs planning and scheduling |
| Upfront cost | Low per seat, grows with headcount | A training fee, plus time away from client work |
| Early usage | A few enthusiasts, many idle seats | Broader, because people know what to try |
| Data and client risk | Higher if rules come later | Lower, rules are taught from the start |
| Tool choice | Made before you know your use cases | Informed by what people actually need |
| Best when | Curious team, simple writing and research work | Skeptical team, sensitive data, idle licenses |
When tools first is the better choice
- Your team is already asking for AI, and some people are using personal accounts. Giving them an approved tool is safer than waiting.
- Most of the work is drafting, summarizing and research, where a general assistant helps right away.
- The firm is small enough that you can watch who uses it and help them directly.
- You need to learn quickly which tasks benefit, and a few weeks of real use will tell you more than a planning exercise.
Even here, write a short set of rules before the licenses go out. Our guide to writing an AI acceptable use policy covers what to include.
When training first is the better choice
- People are skeptical or worried about their jobs, and a tool appearing overnight will raise more questions than it answers. See employees afraid AI will replace them.
- Your work involves client documents, NDAs or regulated data, and people need to know what is safe before they start. See keeping client data safe with AI.
- You already bought licenses and usage is low. More seats will not fix that. Our page on why licenses are not a strategy explains why.
- You are not sure which tool fits your work, and you want real use cases to guide the purchase. Our guide to finding workflows to automate with AI helps with that.
The honest trade-offs
Idle licenses are the common failure of tools first. Many firms turn on seats and find that a small group uses them heavily while most people try once and stop. The spend continues either way.
Training without tools fades fast. If people learn a skill on Tuesday and cannot use it on real work until next quarter, most of it is gone. Training works best when people go straight back to their desks with a tool they are allowed to use.
Tools first can lock in the wrong choice. Buying a firm-wide plan before you know your use cases can leave you with a tool that fits some teams poorly. Our comparison of one AI tool vs many AI tools covers that decision.
Training first can turn into a long planning phase. A training program that waits for a full strategy may never start. You do not need every answer to begin.
Neither one replaces engineering. Some work, such as repeated multi-step tasks across your systems, needs agents or automation rather than more prompts. See training vs consulting vs engineering.
A sequence that often works
- Pick one approved tool for the whole team, or a small pilot group, and write one page of rules.
- Train within the first two to four weeks, using your team’s real documents and tasks.
- Watch usage and time saved on specific tasks for a month. See how to measure AI training results.
- Expand seats and add tools only where people are using them well.
- Build agents or automations for the workflows that keep coming up.
If you want to test this on one team before the whole firm, see company-wide AI rollout vs one-team pilot.
Questions to ask before choosing
- Is anyone already using AI on personal accounts?
- How sensitive is the data your team works with?
- Do you already own licenses, and how many are used each week?
- Can people use what they learn on real work the next day?
- Who will answer questions in the weeks after launch?
How Ruby helps
Ruby Advisory’s recommended path for most firms is to do both close together: help you choose a tool and set the rules, then train your team on their own work within a few weeks, so the licenses get used. When usage shows a workflow worth automating, we build it. We help with:
- AI strategy: choosing tools, writing the rules and deciding where to start.
- AI training: live sessions and cohort programs built around your documents and workflows.
- AI engineering: agents and automations for the work your team repeats most.
To talk through the right order for your firm, book a call.
