Knowledge
Clients pay for judgment. They should not pay for you to rebuild the last proposal from scratch.
Research, proposal assembly, scheduling, status, and file prep still consume the people you hired to think. A generic writing tool helps a paragraph. It does not change the delivery operating system.
How I see it
Professional services sell judgment and waste time on the packet
Miami professional services firms are often founder-led, relationship-heavy, and capacity-constrained. The expert is the product. The operating tax is the packet around the expert: a proposal, a research memo, a workpaper, a client update.
AI is useful when it assembles context from prior work, drafts the first complete packet, and keeps the file current. It is not useful as a promise that the firm will become an AI product company.
The first workflow should be one that repeats weekly: proposal assembly, a class of research, or delivery admin. Measure hours of non-billable assembly and cycle time to a complete draft. Keep humans on advice, negotiation, and anything that creates client exposure.
Do not automate the relationship. Do automate the reconstruction of what the firm already knows.
Common mistakes
What teams usually get wrong.
Training everyone on prompts and leaving the proposal process alone
People will write faster paragraphs and still miss the same sections.
Putting client files into a tool nobody approved
Shadow use creates risk. Name the approved path or the work will stay in email.
Trying to productize the whole firm at once
One packet type is a win. A new operating model is a stall.
A useful diagnostic
Five questions before you fund the work.
Which packet consumes the most non-billable expert time?
That is the first workflow.How much of a proposal or memo is reconstruction versus new judgment?
Reconstruction is the addressable slice.What data must never leave the approved environment?
If this is fuzzy, adoption will freeze or go underground.Will partners reject a packet that arrives the old way?
If not, the old path remains the real path.Which statements create client or legal exposure?
Those stay human.
Economic model
Non-billable assemblyexperts × hours per week on reconstruction × loaded cost × weeks per year
A conservative slice of that number is often enough to justify changing one packet type.
Three credible paths
How far should you go?
Do not force one solution. Choose the path the economics, the risk, and the organization can support.
Improve the work you already have
Keep the process mostly intact and use AI on the bottlenecks that create delay, rework, or follow-up.
The workflow is already sound and a few steps create most of the friction.
Gains are usually incremental. The operating economics do not change much.
Redesign the workflow around AI
Question every handoff, queue, and duplicate step, then rebuild the process around what AI can now do.
The process grew over years and coordination now costs more than the work itself.
Requires process change, clearer ownership, and a willingness to retire old steps.
Put an agent on a high-value outcome
Give an agent responsibility for one valuable result across systems, with humans at the control points that require judgment, authority, or risk acceptance.
The workflow is high-value, variable, multi-step, and worth engineering for production.
Needs stronger architecture, evaluation, controls, and monitoring. This is not a prompt project.
When this is the wrong next step
Do not start with a sector program.
- The firm wants a public AI brand more than a changed packet.
- Partners will not share real work product for examples.
- Client confidentiality rules have not been decided.
- The work is truly one-of-one with no repeatable file.
A useful next step
Bring one expensive workflow from this industry.
We guide the implementation, go deep on the technical path, and stay hands-on through operations — or tell you when a simpler answer is better.
Discuss an AI opportunity