Why do firms confuse the two?
Because both are sold as the answer to the same complaint: we have bought AI licences and nothing has changed. Training says the problem is skill. Implementation says the problem is the job. Both can be true, but in most professional services firms the second is the bigger issue.
A fee earner who has been on a good course can write a better prompt. They still open last month's report, copy it, chase the figures and rewrite the commentary, because that is how the report is made. The skill improved; the process did not.
How do they compare side by side?
| AI training | AI implementation | |
|---|---|---|
| What it changes | What individuals know and can do with AI tools. | How one specific job is carried out, from input to sign-off. |
| Cost structure | Per session, per head or per course. Easy to budget. | Per process, fixed or on time. Set by how complex the job is. |
| Time to value | Immediate for individual tasks. Firm-wide change depends on uptake. | Once the process is live, typically weeks rather than days. |
| Risk | Low cost, but a high chance that habits fade within weeks. | Higher commitment. Lower if payment depends on the process going live. |
| Consistency | Varies by person. Keen staff pull ahead, others do not start. | Everyone doing the job uses the same process. |
| Control | Hard to see who uses what, or on which client data. | Review and sign-off are built into the process. |
| What happens after | Skills stay with the people trained and leave when they do. | The process stays with the firm, whoever does the job. |
When is training the better choice?
When the work is varied rather than repeated. Partners drafting one-off advice, researching an unusual point or preparing for a difficult meeting benefit from knowing how to use an AI assistant well. There is no single process to rebuild, so the investment belongs in the person.
Training is also the right call when the risk is behaviour rather than time. If staff are pasting client material into public tools without guidance, the urgent need is a clear policy and a session on safe use, not a rebuilt process. Writing an AI policy covers that. In both of these cases a training provider will serve you better than an implementation partner like us.
When is implementation the better choice?
When you can name the job. If proposals, file notes or compliance evidence take the same senior people the same hours every week, better prompting will shave a little off but will not change the shape of the work. The job needs rebuilding so that it draws on the firm's own records and templates, produces a draft for review and stops depending on someone assembling it by hand.
Implementation is also better when consistency matters to clients or regulators. A process with a named reviewer and a clear trail is easier to stand behind than a dozen people each using AI in their own way.
How should you decide?
- Ask what you are trying to change. If the answer is confidence, safety or general fluency, choose training. If it is hours on a particular job, choose implementation.
- Look at what happened last time. If you have trained before and usage faded, more of the same is unlikely to stick. Getting a team to adopt AI explains why.
- Consider staff turnover. In a firm where people move on often, skills walk out of the door. A rebuilt process does not.
- Put a number on the job. Hours a week, times the people doing it, times your charge-out rate, times 46 working weeks. If that figure is large, it deserves more than a course.
Aldbry does implementation: one process mapped, rebuilt and live in 30 days, with the team trained on the new way of doing that job. We do not sell general AI courses. If your need is fluency across the firm rather than a specific job, a training provider is the better fit. If you want to know which job to start with, the audit takes about three minutes.