Hook: The Odd Little Prompt That Supercharges Your Workflow
You’ve probably typed it without thinking: “Review the AI features you’ve used.” Then the magic happens—your assistant summarizes what you tried, what worked, and what failed. It’s a tiny phrase with outsized impact, and it’s quietly becoming one of the most useful meta‑prompts for knowledge workers who touch AI every day.
This is a product review meets explainer of the deceptively simple command: review the AI features you’ve used. We’ll unpack what it does across tools, why it matters for teams, and how to make it a weekly ritual that saves hours.
What does “review the AI features you’ve used” actually mean?
At its core, this prompt asks your assistant to compile and reflect on your AI usage: the tools you invoked, the models you selected, the prompts that worked best, and the outputs worth keeping. Think of it as a personal changelog—only smarter, because the assistant can:
- List which AI features were used (summarize, rewrite, analyze, extract, generate code, translate, brainstorm, etc.)
- Map features to tasks (e.g., “Used ‘Summarize PDF’ for Q4 earnings call notes”)
- Evaluate effectiveness (speed gained, quality delivered, issues encountered)
- Recommend next steps (try different model, tune prompt, create reusable template)
It’s the difference between “I used AI a lot this week” and “Here’s the 20% of AI features that delivered 80% of the value.”
Why this prompt matters now
- Cognitive offload: AI sessions are fragmented. You jump between chat, documents, emails, and dashboards. A review consolidates the chaos.
- Performance transparency: If you can’t measure which AI features actually help, you can’t improve your workflow.
- Team knowledge sharing: A repeatable review makes it easy to broadcast what works to teammates.
- Compliance nudge: Reviewing features helps you catch sensitive content, model choices, and data sharing settings.
The Review Format: A Template You Can Reuse
Try asking your assistant:
“Review the AI features you’ve used this week. Organize by task, feature used, model, prompt pattern, time saved, and issues. End with three recommendations to improve.”
You’ll often get a structured response. If not, add headings like:
- Tasks covered (with links or filenames)
- AI features used (summarize, classify, translate, code generate, image captioning)
- Model & settings (model versions, temperature, tokens)
- Prompt patterns (role prompts, chain-of-thought-lite, few-shot examples)
- Metrics (turnaround time, edits required, factuality checks)
- Failures (hallucinations, formatting issues, timeouts)
- Suggested improvements (templates, automation, guardrails)
Real-world examples: Where “review the AI features you’ve used” shines
- Situation: You processed 12 PDFs, scraped two sites, and ran three entity extractions.
- Review outcome: Identifies that “table extraction” underperformed on scanned docs, while “abstract summarization” consistently delivered. Recommends adding OCR preprocessing.
- Situation: You generated briefs, outlines, and first drafts.
- Review outcome: Shows that using “tone transfer” and “SEO outline generator” produced fewer revisions than raw drafting. Suggests standardizing an outline template.
- Data analysis in spreadsheets
- Situation: You tried AI formulas for cleaning, categorizing, and forecasting.
- Review outcome: Notes accuracy improved when providing column headers as schema examples. Recommends a reusable prompt: “Use these headers as ground truth schema.”
- Situation: You summarized 60 tickets and generated response macros.
- Review outcome: Flags that “sentiment analysis” helped triage but misclassified sarcasm. Suggests adding 5 few-shot examples with edge tones.
- Situation: You used AI for code comments and API doc summaries.
- Review outcome: Finds that “diff explanation” saved time, but “auto-refactor” created style regressions. Recommends running a linter preset before acceptance.
Pros and cons of the meta‑prompt
Pros
- Fast clarity: A single line delivers a usable audit.
- Habit-friendly: Weekly cadence compounds benefits.
- Team-ready: Easy to share as a lightweight report.
Cons
- Tool variance: Not every assistant tracks feature usage equally.
- Partial visibility: Private or external tools may be missed.
- Metric fuzziness: “Time saved” can be subjective without benchmarks.
Mitigations
- Ask for citations: “Link to the original files, chats, and timestamps.”
- Calibrate metrics: Set baselines (e.g., average minutes per task pre‑AI).
- Add structure: Provide a template the model must fill.
How to get 10x more from your weekly AI feature review
- Standardize prompts: Convert winning prompts into templates with variables (tone, audience, length).
- Create a "Do Not Use" list: Document features that underperform for your workflows.
- Bias for reuse: Turn strong outputs into snippets, style guides, or checklists.
- Compare models: Ask the assistant to A/B feature performance across models.
- Close the loop: End every review with automations or saved workflows.
Example meta‑prompt chain:
- “Review the AI features you’ve used this week.”
- “Convert the top 3 into reusable templates.”
- “Package them as step‑by‑step workflows with guardrails.”
- “Create a one‑page team summary with links.”
Metrics that matter in an AI features review
- Completion quality: Edit distance to final (number of edits or revisions)
- Time saved: Minutes or % reduction vs. baseline
- Factuality: Number of citations checked or corrections needed
- Reusability: How many times a template is used afterward
- Reliability: Error rate (timeouts, formatting breaks)
Ask explicitly: “Estimate metrics and note confidence levels. If uncertain, suggest how to measure next time.”
Common pitfalls and how to avoid them
- Overgeneralizing from a single success: Require at least 3 examples before promoting a template.
- Hidden context drift: Include constraints in your prompts (audience, region, reading level).
- Hallucinated feature names: Cross‑check with your tool’s feature list; ask for logs.
- Ignoring security: Include a section on data handling (PII, model boundaries, redaction).
A short, repeatable weekly ritual
Try this 15‑minute cadence:
- Monday kick‑off: Define 3 high‑value tasks; pin prompts.
- Daily: Tag sessions by task (or ask the assistant to auto‑tag).
- Friday review: “Review the AI features you’ve used this week.”
- Share: Post the report with 3 improvements and 1 deprecated pattern.
Tooling note: Where this prompt works best
This meta‑prompt shines in AI workspaces that centralize chats, file actions, and model runs—so the assistant has context to analyze. If your environment supports document summarization, code assistance, spreadsheet formulas, and web extraction in one place, the review can capture the full picture.
Worth noting: If you’re using Sider.AI to work across chats, PDFs, webpages, and spreadsheets, asking it to “review the AI features you’ve used” can surface which features (like summarize, extract tables, outline generator, or code explain) actually moved the needle. By the way, Sider.AI’s unified history makes it easier to link specific actions to outcomes, which improves repeatability and team hand‑off. Advanced: Turn reviews into automated playbooks
- Auto‑tagging: Ask your assistant to label sessions by task and feature as you work.
- Playbook extraction: “From this week’s review, generate a playbook with inputs, steps, and quality checks.”
- Drift detection: “Alert me if a feature’s accuracy drops below 90% week‑over‑week.”
- Team roll‑outs: “Create a 10‑minute training with examples and a quiz.”
The bottom line
“Review the AI features you’ve used” is a powerful, low‑friction prompt. It converts scattered experiments into a compounding system: measure, learn, standardize, and scale. Use it weekly, demand structured outputs, and turn insights into templates and playbooks. Your future self—and your team—will thank you.
Next steps
- Today: Run the review on your latest week of work.
- This week: A/B test two features on the same task and compare.
- This month: Build a shared team playbook from your top three workflows.
FAQ
Q1:What does “review the AI features you’ve used” do in practice?
It compiles a structured summary of your recent AI activity—features used, prompts, models, and results—then surfaces what worked, what didn’t, and how to improve. Think of it as a smart retrospective for your AI workflow.
Q2:How often should I review the AI features I’ve used?
Weekly works best for most teams because it captures enough data to spot patterns without becoming a burden. Pair it with a simple template so the review stays consistent over time.
Q3:Which metrics should I track when I review AI features?
Track time saved, edit distance to final output, factual accuracy, error rates, and reuse of templates. Ask the assistant to estimate confidence levels and suggest better measurement if data is thin.
Q4:Can I automate a review of the AI features I’ve used?
Yes. Prompt the assistant to auto‑tag sessions by task and feature, then generate a weekly report with links to source files and chats. You can also trigger alerts for performance drops or failed runs.
Q5:Does a review work across different AI tools and models?
It works best in unified workspaces that track actions across chats, documents, and code. If your tools are fragmented, ask the assistant to include links and logs so you can reconcile results in one report.
