Introducing the TIF Framework - A simple prompting framework for the busy professional
Over the last 6 months working with Asset Managers, I’ve seen many people (including myself) get what I call “slop” from AI—blurry, off-target answers that leave you more confused than when you started. After lots of trial and error, I discovered a straightforward way to steer AI toward more relevant, higher-quality responses, even in specialized fields like asset management. That’s why I developed the TIF (Topic, Intent, Format) framework: it’s lean and user-friendly, so you don’t need a deep background in AI to start seeing immediate improvements. In fact, by simply following TIF, you can often boost the clarity and usefulness of AI-generated content by 3× to 5×—without extra jargon or guesswork getting in the way.
How LLMs (Large Language Models) Work
Modern AI tools, often called Large Language Models, operate through a method known as “next-word prediction.” When you type in a prompt, the model predicts the most likely sequence of words that fit together, based on patterns it has learned from vast amounts of text data. While this means the AI can generate incredibly human-like responses, it also means that precise input—your prompt—directly influences the quality of the output.
Think of it like asking for directions from a GPS that relies on your initial address. If you provide a vague address (e.g., “somewhere in New York”), you’ll get a vague route. But if you specify the exact street name and house number, you’ll get much more accurate guidance.
Why Prompting an LLM is Different from Google or Talking to a Person
Google vs. AI Models
When you type a query into Google, it hunts for existing webpages that match your keywords. An LLM, on the other hand, uses predictive text generation to craft entirely new content or context-driven answers. A simple keyword search isn’t enough—you need detailed instructions to guide the model toward the right kind of response.
People vs. AI Models
Humans can “fill in the blanks” with context or common sense. AI requires explicit direction. If you omit details about the style or length of your answer, the model may guess—but that guess might not match your expectations. Getting the right output from an LLM is a new skill set, akin to giving detailed, unambiguous instructions to a capable but literal assistant.
The TIF Framework: A Quick Overview
- Topic: Provide the core information or question (e.g., text to analyze, a rough draft to revise, or a detailed question about an asset management issue).
- Intent: Specify what you want the AI to do with that information (summarize, rewrite, compare, craft a response, etc.).
- Format: Indicate the style, length, or structure of the final output (e.g., bullet points, paragraphs, formal tone, or word limit).
Below are seven examples that demonstrate how you might use the TIF (Topic, Intent, Format) framework in asset management scenarios, including four that are posed as direct questions to an AI chatbot with access to your firm’s grounded data. Each example shows how clarifying the Topic, Intent, and Format leads to more targeted, high-quality responses.
Example 1: RFP Response
Scenario: You’re responding to an RFP that focuses on your firm’s track record and risk management process for large-cap growth equity strategies.
Topic (Question to the Chatbot):
“Describe our risk management approach?”
Intent:
Draft a compelling, concise RFP response focusing on these key points.
Format:
Use a formal, professional tone in two paragraphs.
Example 2: DDQ (Due Diligence Questionnaire)
Scenario: You’re asked in a DDQ about your firm’s ESG integration practices, and you need to provide a succinct yet informative answer.
Topic (Question to the Chatbot):
“Explain our ESG integration and stewardship approach?”
Intent:
Craft a concise, clear response that highlights our ESG policies and track record.
Format:
One short paragraph (3–4 sentences) in a formal tone.
Example 3: Quarterly Commentary Summary
Scenario: You have a 15-page quarterly commentary covering market trends, portfolio changes, and future outlook. You want to boil it down into key insights.
Topic (Question to the Chatbot):
“What are the main points of our quarterly commentary?”
Intent:
Highlight the major performance drivers, market trends, and key takeaways for our leadership.
Format:
Provide 5 bullet points, each with one to two sentences, in a professional style.
Example 4: Policy Comparison
Scenario: You’ve updated an internal trading policy, and you need to show what’s changed from the old version.
Topic (Question to the Chatbot):
“Describe our new internal trading policies?”
Intent:
Clearly outline the policy changes and the reasoning behind them.
Format:
Present a concise list or table using short explanations.
Example 5: Email Rough Draft Clean-Up
Scenario: You have a rough draft of an email to a client that needs to be polished for clarity and professionalism.
Topic:
Dear Client,
We wanted to let you know we’re changing up how we present your portfolio statements. Starting next quarter, we’ll be using a new template that might look a little different, but it’s better overall for showing performance details. Let us know if you have any concerns. Thanks, and looking forward to hearing from you!
Intent:
Clean up this draft to be more concise and professional, while maintaining a friendly tone.
Format:
Rewrite it in 3-4 sentences with a clear introduction and conclusion.
Example 6: Soften Language in an Internal Teams Message
Scenario: You need a friendlier way to request a delayed report from a coworker.
Topic:
Please send me that report ASAP. It was due yesterday and we’re behind schedule. Let’s get it done now.
Intent:
Rewrite this message to sound friendly yet firm when asking about the delayed report.
Format:
A short, casual internal Teams message (1-2 sentences) with a cooperative tone.
Example 7: Condense Dense Text into Bullet Points
Scenario: You have a long internal memo full of dense information about compliance guidelines, HR policies, and remote work changes.
Topic:
Our internal memo covers new compliance guidelines, updated HR policies, and a shift in remote work practices for the upcoming year.
Intent:
Condense the critical points into an easily digestible format.
Format:
Provide 4-5 bullet points summarizing the essential updates, each in one sentence.
How to Use These Examples
-
Copy and Adapt
- Take any example as a template and swap in your own details, questions, or documents.
-
Stick to TIF
- Clearly define the Topic (in these first four examples, it’s an actual question to the chatbot), specify what you Intent to do with it, and outline the final Format required.
-
Refine & Iterate
- If you don’t get the response you need initially, add more details or constraints to your prompts. The more specific you are, the better your AI output will be.
By applying the TIF framework in each situation, you’ll give your AI-powered tools the context they need to deliver relevant, accurate, and well-structured answers—saving you time and ensuring you produce consistent, high-quality work in the asset management space.