We Let Our Financial Planners Write the Agents — Matthew Ragone, Range
Read the talk
We Let Our Financial Planners Write the Agents
At Range, planners turned their own judgment into reusable agent skills. The method starts with familiar work, captures the questions behind each decision, and keeps both client recommendations and shared skills under human review.
From a talk by Matthew Ragone
At a glance
Ideas worth remembering
An agent can prepare another client’s analysis while a planner is on a call, but a planner reviews every recommendation and controls delivery.
Start with repeated, relatively simple work whose quality the author can judge quickly. Teach it through real cases and explain why each action matters.
A useful skill includes required questions and limits on independent action, alongside the steps for completing the task.
Run five to ten real cases before sharing, correct missing questions and confident mistakes, and assign both a maintenance owner and an independent reviewer.
A faster message does not build the next client’s plan
Range’s first use of AI was familiar: write a client message, paste it into a chatbot, and ask for clearer or friendlier wording. That saved a few minutes, but the planner still performed the work manually. Matthew Ragone, a certified financial planner who leads AI adoption for Range’s planning team, opens with the constraint that made a different workflow necessary: a planner can only be in one place at a time.
During a call with one client, the planner cannot also build another client’s plan. An agent changes that allocation of time. While Ragone is on the call, it drafts a cash flow analysis or an insurance analysis for a different client. After the call, he opens the draft, reviews it fully, makes corrections, and sends the recommendations. The first draft can now progress without his continuous attention.
The handoff matters because the output is financial advice to real clients in a regulated business. “The agent doesn’t send anything.” Every recommendation passes through a planner before a client sees it. The workflow delegates preparation while keeping the decision to deliver advice with the professional reviewing it.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
A skill captures how this firm does one piece of work
Getting to that workflow required more than a polished prompt. Ragone identifies two requirements that solve different problems:
- Work in context. The agent must operate where the work lives, avoiding a separate chat window that requires the planner to copy information back and forth.
- Use the firm’s judgment. General financial-planning knowledge does not specify what Range checks first, what it always flags, or how it phrases a particular recommendation.
A skill is the firm’s way of doing one specific piece of work, written down so an agent can follow it. That definition makes the practitioner’s knowledge the main input: the person must know the sequence, the exceptions, and what a good result looks like. Ragone organizes the build process into five steps—pick, teach, pack, test, and ship—none of which required the planners to write code.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Pick work you can judge, then explain the judgment
The first selection criterion is familiarity. Choose a workflow performed every week whose bad drafts are easy for the practitioner to recognize. Ragone advises starting with relatively simple tasks that still require substantial manual effort, rather than reaching immediately for the most complex work. At Range, cash flow analyses and insurance reviews fit that combination.
Recognizing quality is part of the build process itself. If the author cannot judge an output quickly, testing becomes difficult and confidence in the skill has little foundation. Repetition supplies opportunities to use the skill; expertise supplies a way to assess whether it performed the work properly.
Teaching starts with several actual client cases from plans already delivered. Walk through each case from beginning to end and narrate the reasons behind the actions. The pay-stub example makes the distinction concrete: “Next, I open the pay stub” records an action, but leaves out why that document matters, what the planner is looking for, and what discovery would require stopping to ask a question.
That narration turns an observable sequence into guidance for decisions. A list of documents to open cannot, by itself, tell an agent when the available information is insufficient. Explaining what prompts a question captures some of the judgment that makes the result Range’s work rather than generic advice. The talk gives this teaching method without specifying the financial checks performed on the pay stub.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Pack the questions and stopping points alongside the procedure
Packaging the skill means writing more than a procedure. Two kinds of instruction govern how the agent should proceed:
- Questions it must ask. Specify what it needs to know before starting and what it should confirm with the planner instead of guessing.
- Limits it must respect. Specify actions it must never take independently. At Range, the central example is sending recommendations to a client: the agent drafts, and a planner decides whether the draft needs changes and whether to send it.
These instructions address different failure modes. Required questions interrupt unsupported assumptions before they become part of the analysis. Limits prevent a completed draft from turning automatically into a client communication. The procedure describes how to do the work; the questions and limits describe when human input or a human decision is required.
Ragone compares a skill to a recipe that Claude can follow on command. The useful part of that analogy is the reusable written process. His next step—repeated testing on real cases—also makes clear that writing the recipe does not finish the job.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Test before distribution, then give the skill an owner
Before anyone else uses the skill, run it five to ten times on real cases. The failures to watch for are specific: an edge case the instructions do not cover, a question the agent should have asked, or an answer that sounds confident but is wrong. Each discovery calls for changing the skill, saving it, and running it again. The corrective artifact is the reusable instruction, so the next attempt can benefit from what the author learned.
The cost comparison is distribution. Five to ten trials take effort, but an unfinished skill handed to 15, 20, or 100 colleagues can multiply the cleanup. Those figures illustrate Ragone’s rollout warning; the testing recommendation is a practical starting point, not a demonstrated error-rate guarantee.
Shipping adds two distinct responsibilities:
- Owner. Keep the skill current as internal processes and tax rules change. A skill that was correct in March can be wrong by December even if nobody edits it.
- Reviewer. Have someone other than the author check the skill before the team relies on it. Range applies the same independence it expects when reviewing a financial plan: “You don’t grade your own work.”
Skill review and recommendation review operate at different levels. A separate reviewer checks the shared instructions before colleagues depend on them; a planner still reviews each resulting client recommendation before delivery. Maintaining the process does not remove the need to inspect its outputs.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
The work spreads; responsibility has to spread with it
Six months into the effort, Ragone reports 24 skills across Range, including 17 in planning. Sales, product, and marketing adopted the approach without a mandate after seeing the planners use it. The skills were written and maintained by the practitioners themselves—planners, marketers, and salespeople.
Practitioner authorship made the instructions more specific. The person with the strongest insurance expertise could build the insurance skill; the tax specialist could build the tax skill. Peer review also had an existing home in the organization. Planners already reviewed one another’s plans, so reviewing one another’s skills extended a familiar professional habit to a new artifact.
The ending shifts attention from whether a model can produce a cash flow draft to who makes that draft useful and dependable inside a firm. Someone must write down the firm’s method, test it before release, and own it once 15 other people depend on it. Ragone presents Range’s experience as one working method and acknowledges that there is little field data yet. The reported adoption shows that practitioners can build and maintain these skills; it does not establish a general solution for checking professional judgment at scale.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Resources
Related talks
- Beating RL With Reflection: GEPA and Optimize Anything
Extends the question of how skills improve: execution feedback can guide instruction changes, and human annotations can help train evaluators for tasks requiring judgment.
Read the complete timestamped transcript
- 0:01
Hi everyone. I'm Matt Ragone, and I'm a financial planner at Range. I want to show you how our financial planners ended up writing the agents that we use every day, and none of us, the financial planners for that matter, write code. This is the build guide, how we did it step by step, so you can take it back to your own teams, companies, et cetera. So a quick bit about me because it matters for how you'll hear this talk. I'm not an engineer. I'm a certified financial
- 0:31
planner, and I spend my days with clients talking about cash flow, taxes, insurance, retirement, estate, you name it. I also lead the AI adoption for the planning team at Range, which is an all-in-one wealth management company. Most agent talks you'll hear come from the people who built the platform. This talk comes from someone on the other side, the people you hand the platform to. We work in a regulated business. The output is financial advice to
- 1:01
real clients, and the review step is not optional. So this isn't a vendor pitch, it's a method, and it's what actually worked for us. We started where everyone starts: AI as a chatbot. I write a client message, paste it in and ask it to make it a little clearer or friendlier, and that was useful. It saved me a few minutes here and there, but I was still doing all of the work manually. I was just doing it a little bit faster. The
- 1:31
change that actually mattered came later. Here's what we figured out. A planner's constraint was never typing speed, for example. It's that we can only be in one place at a time. If I'm on a call with a client, I'm not building someone else's plan. Writing faster doesn't fix that problem. What fixes it is work that happens while I'm somewhere else. So here's what a day looks like now. I'm on a call with one client.
- 2:01
At the same time, an agent is drafting a full cash flow analysis or an insurance analysis for a different client. When the call ends, I open that draft. I fully review it, adjust what needs to be adjusted, and send the client their recommendations. I want to be really clear about that last step. The agent doesn't send anything. A planner reviews every single recommendation before a client sees it. What changed is that the
- 2:31
first draft no longer needs me sitting in front of it. Getting there took a lot more than a good prompt, however. Two things had to be true. First, the agent has to work where the work lives, not in a separate chat window I'm copying and pasting it into. Second, it has to know how we do it, not how financial planning works in general. The model already knows that. But the agent needs to know how Range does it, what we look
- 3:01
at first, what we always flag, how we phrase a specific recommendation. That combination is what a skill is. Our firm's way of doing one specific piece of work written down so an agent can follow it. The bulk of this talk is how you build one. Five steps: pick, teach, pack, test, and then ship. None of them require writing code. Every one of them requires
- 3:31
knowing the work really well, which is exactly why the people doing the work should also be the ones building it.
- 3:40
Step one, pick a workflow you repeat. Your initial instinct is to go for the flashiest work, whatever's taking the most amount of time and the most complex. Don't. Pick something you do every week where you already know what good looks like, where you can spot a bad draft at a glance. Additionally, we recommend choosing some of the least complex workflows to start, but some that require the most manual work. For us, that meant things like cash flow
- 4:09
analysis and insurance reviews. Just for us here at Range, that required the most manual work. And if you can't judge the output quickly, you can't test it, and you can't trust it. So look for two things. You do it often, and you know how to judge it. Step two, show it several real cases for previous plans you've delivered. Take several actual client cases and walk through the work start to finish.
- 4:39
And as you do it, narrate your judgment out loud. Not just, "Next, I open the pay stub," why you're opening it, what you're looking for. What would make you stop and ask a question? That narration is the whole game. The steps are easy to write down. The judgment is what makes it your firm's work instead of generic advice.
- 5:03
Step three, pack it into a skill. A skill isn't just a lips-- a list of steps. Two things matter most. The questions it must ask. Before it starts, what does it need to know? What should it confirm with me instead of simply guessing? And the limits it must respect, what it never does on its own. In our world, the big one is that it never sends anything to a client. It drafts, and then a planner decides if it gets sent or needs to be adjusted.
- 5:33
If you only write the steps, you get something that simply just runs on its own. If you write the questions and set the limits, you can get something you can trust. Think of a skill as a recipe. Once Claude knows how to do the skill, it will repeat your process on command. Step number four, run it five or ten times on real cases before anyone else sees it. This is the step everyone wants to skip. Every run teaches you something the skill
- 6:04
didn't know yet, an edge case, a question it should have asked, a spot where it sounded confident and was wrong. You need to fix it, save the skill, and run it again and again and again. Five to 10 runs sounds like a lot, but it's a lot less work than cleaning up after you hand a half-baked skill to 15, 20, 100 people in your team. Step five, ship it with an owner and a reviewer. The owner keeps it
- 6:34
current. Our processes change, tax rules change, and a skill that was right in March can be wrong by December. The reviewer checks it before the team relies on it, and it's someone other than the author. Same as a plan. You don't grade your own work. Once other people depend on a skill, somebody has to be accountable for it. That's the difference between a clever personal shortcut and something a team can build on. So where are
- 7:04
we at currently? Six months in, we have 24 skills across the company, and 17 of them are in planning. And here's the part I didn't expect. It spread to sales, product, marketing, all without a mandate. Nobody sent a memo. People saw what the planners were doing and started building their own agents. Every one of those skills was written and is maintained by planners, marketers, and salespeople, not developers or engineers.
- 7:34
A couple of things changed once subject matter experts started building for their own niche. The skills got a lot more specific. The person who knows insurance, for example, best builds the insurance skill. The person who best knows tax, tax code, tax law should be building the tax one. And a review came naturally. We already review each other's plans, and that's just how planning works. So now we review each other's skills in that same way. It's the same
- 8:04
habit, just a new thing to review. So why does this matter right now? Because the bottleneck has moved. Whether a model can draft a cash flow analysis, that question's already been settled. It can. What's not settled, who writes down how your firm does it? Who tests it before it ships? Who owns it once 15 other people depend on it? If you're being asked to push agents out past the engineering org, you're about to hit these
- 8:34
exact questions. There isn't much field data on this yet, but this is how we've done it here at Range.
- 8:43
So that's it. Thank you so much for listening to this talk. I'm Matt Ragone from Range, and you can reach me fr- at [REDACTED]. Have a great day.