Deal Sourcing & Market Research · Playbook 02

Build a client investment strategy agent.

Most agents brief an AI tool in one sentence and get back generic advice. Set it up as a dedicated Project, let it interview you or your client before it generates anything, and the output changes shape entirely.

Why this matters for client advising.

Most agents who ask an AI tool for an investment strategy for a client paste three sentences and accept whatever comes back. That output is generic by design. The model has no idea what capital the client actually has to deploy, what timeline they're underwriting against, or what risk they'll live with through a cycle.

Set the tool up as a dedicated Project and let it interview the client before generating anything, and the output changes. You get a strategy tied to specific capital, goals, and risk profile, something you can hand over and defend, not a paragraph anyone could have written.

The strategy that comes back is only ever as good as the inputs you put in.

Three clicks. One conversation.

Create the Project

Open Claude, go to Projects, click New Project. Name it after the client or the strategy exercise. Add a one-line description of what it's building.

Feed it the client's profile

Paste a short template covering age, residency, capital available today, monthly capacity, primary goal, and risk appetite. Vague inputs produce vague strategies, so push for specifics.

Let it interview

The tool will ask clarifying questions. Treat the interview as the actual work. Don't rush it, and don't let it stop early; ask if there are three more questions worth asking.

Generate the strategy

Once the tool has read the context back correctly, ask for a personalised strategy document: capital deployment plan, strategy archetypes that fit, sample target deal profiles, and a 12-month roadmap.

The client profile template

I am advising a client who is [age]-year-old [nationality/residency].
They have [capital available today], plus a monthly investing capacity of [amount].
Their primary goal is [objective, e.g. long-term appreciation, cash flow, hybrid].
They are [risk appetite, e.g. comfortable with leverage / risk-averse].

What it still gets wrong.

Context decays over a long chat

After a few dozen messages the model starts losing early details. Start a fresh chat inside the same Project rather than continuing an old one.

ii.

It invents market specifics

Ask for 'the best area to invest in' and it writes something confident without live data. Treat any named area as a starting point to verify, not a recommendation.

iii.

Skip the interview, get a generic plan

If you ask for the strategy before the interview runs, you get advice it would give anyone. The interview is where the personalisation happens.

iv.

One strategy at a time

This produces a single coherent strategy. It won't run three competing strategies side by side or benchmark against a model portfolio in one pass.