Find the useful work
AI for professional services: turn expertise into a wider service
AI can change how a service firm delivers its expertise. It may let clients get useful help between major projects, or let the firm serve work it cannot take on today. In a large business, the same approach can help specialist teams support more colleagues. The starting question is what you want people to be able to do.
Choose a service people need, then design how your experts, your users and the system will deliver it together.
Find the work your current model cannot serve well
Your clients may still need help after a strategy project ends. They may be able to carry out the work themselves but need your methods and judgement along the way.
An ongoing service could help them turn that strategy into a plan, check progress and ask for expert help at the right moments. AI can do some of the preparation while your team keeps the decisions that need its judgement.
First, check whether clients want this help. Ask which job they need to do often and how they do it now.
Inside a large business, look for the same pattern. A small team of experts may serve many regions or teams. The aim could be to help those teams use what the experts know without waiting for every answer.
What we have already built
Our delivered examples come from marketing and PR agencies. They show parts of this approach in use. The new service ideas below are made-up examples, not past client work.
| Job | What the system does | What people still do | Evidence and limit |
|---|---|---|---|
| Find relevant websites for outreach | Researches pages and scores their fit for a PR agency. | Choose prospects and handle outreach and placement decisions. | The fifteen-minute run covers research and scoring. It does not include winning links. |
| Handle journalist requests | Sorts incoming requests, matches them to clients and prepares drafts. | Check the match, judge the response and approve messages. | People approved every outgoing message at launch. The old routine used 11.25 team hours a week. |
| Bring client reporting together | Joins data from twelve platforms in a shared place for questions and analysis. | Check the meaning of the figures and decide what action to take. | Agency staff and clients use it. This does not establish the return for another firm. |
Design the first service and what could follow
Consider an invented planning service. A client sends a request. The system checks the client’s goals and current agreement, applies the firm’s methods and prepares a plan. An expert checks it before release.
The first build covers one type of plan. It also creates useful foundations: clear methods, up-to-date client records, access rules and a record of approved decisions.
A later service could use those plans for progress reviews. It would need new information about results and its own checks. Reusing part of the first build can help. It does not remove that work.
For a large business, a similar service could help local teams prepare plans using central guidance. A wider rollout must account for different rules, systems and approval roles.
Decide how people and software share the work
The system needs to know which information to use and what it is allowed to do. People need to know which decisions are theirs.
Name who writes or prepares, who checks, who approves and who carries out the work. Some steps may use fixed software rules. Others may use AI or need an expert.
You may use the same method for many clients, but keep their private records separate. Agree who can use each source, who updates it and who handles unclear requests.
You do not need to write down everything your experts know at once. Start with the knowledge needed for the agreed workflow and improve it through reviewed use.
Use these questions to choose where to begin
Use what you know already. This sheet helps you compare ideas. You do not need to fill it in before speaking to us.
A completed opportunity example
Worked example
An opportunity for ongoing client planning
Invented proposal. No demand, sales or test results are claimed.
Need
Strategy clients need help making their next plan between major projects.
Current limit:
The firm cannot offer frequent expert sessions to every client.
Goal:
A useful ongoing service that clients can act on themselves.
First build
One type of plan, prepared from approved methods and client goals.
An expert checks the advice. The client approves and carries out the work.
The system cannot spend money or change the client's systems.
First checks
Will clients buy and use it?
Can the firm deliver it well after expert review and support costs?
Does it replace existing work, and what does that do to the return?
Foundation
Current client records, written methods, access rules and approved plans.
Possible next use:
Progress reviews using those plans. Extra data and tests needed.
Next step
Show a sample to suitable clients and test the hardest delivery question.
Use the evidence to decide whether to fund the full build.
Where LeverageAI fits
We turn business knowledge into systems that get work done. Our workshop helps service firms and large organisations choose a first workflow that supports their wider goal.
We lead the work: map the process, get the data ready, build the system and teach your team how to use it. Your team explains the work, agrees access and makes the business choices.
We promise one agreed workflow, built and working in 90 days or less from the start of the build project. Later services or wider rollout need their own scope and evidence.
Further reading
Written by Said Lopez, founder of LeverageAI and author of Lean AI. Our guides draw on the method in the book and our work with clients.