Point AI Best Practices
Getting the most out of Point AI comes down to asking good questions and understanding what the AI can and can't do. Here are the key tips from our most experienced users.
Writing Good Prompts
Be Specific About the Portfolio
❌ "How is the portfolio performing?"
✅ "How is the Smith Family Office GBP portfolio performing year to date compared to its benchmark?"
The more specific you are about which portfolio, which time period, and what you want to compare, the better the response.
Specify the Time Period
❌ "What are the top holdings?"
✅ "What are the top 10 holdings by value in the Henderson portfolio as of today?"
State the Purpose
❌ "Summarise the portfolio."
✅ "Summarise the Jones portfolio for a client meeting tomorrow. The client is primarily concerned about their equity exposure and recent performance."
When you tell Point AI the purpose, it can tailor the response appropriately.
Effective Use Cases
✅ Great Uses for Point AI
| Use Case | Example Prompt |
|---|---|
| Pre-meeting preparation | "What should I know before my meeting with the Smith family?" |
| Quick performance check | "How has the Henderson portfolio performed this month?" |
| Exception triage | "Are there any data quality issues I should know about today?" |
| Portfolio comparison | "Compare the performance of all GBP portfolios year to date." |
| Allocation analysis | "Which portfolios have more than 60% equity exposure?" |
| Report drafting | "Draft an executive summary for the Q4 investment committee." |
⚠️ Use with Caution
| Use Case | Why to Be Careful |
|---|---|
| Precise calculations | For exact figures, verify in the analytics module |
| Compliance-critical statements | Always review AI-generated text before using in regulated communications |
| Historical data before go-live | Point AI only has data from when your organisation started using Point |
❌ Not Suitable for Point AI
| Use Case | Better Alternative |
|---|---|
| Making investment decisions | Point AI informs, it does not recommend |
| Predicting future performance | Point AI is backward-looking only |
| Analysing external documents | Point AI only works with data in Point |
Reviewing AI Output
Always review Point AI output before using it:
- Check the data citations — click through to verify the underlying numbers
- Check the valuation date — ensure the AI is using the data you expect
- Review for context — the AI may not know about recent events that aren't yet in the data
- Edit before sending — always personalise client-facing content
Getting Better Results Over Time
Point AI improves as you use it. If a response isn't quite right:
- Click 👍 / 👎 to rate the response
- Use the Feedback option to explain what was wrong
- Try rephrasing your question
Your feedback helps the Point team improve the AI for everyone.