Tools
Perplexity Deep Research Tool Debuts, Challenging OpenAI and Google
Perplexity AI’s new freemium Deep Research product is shaking up AI, offering lightning-fast, expert-level insights across many industries.
Published
3 months agoon
By
AIinAsia
TL;DR – What You Need to Know in 30 Seconds
- Freemium Launch: Perplexity AI debuts “Deep Research” with a free tier, challenging pricey AI tools.
- Lightning-Fast: Gathers and summarizes data from dozens of sources in minutes, mimicking a human researcher.
- Strong Performance: Slightly trails OpenAI in some metrics but still outperforms many other AI models.
- Market Disruption: Major funding fuels a direct challenge to Google and OpenAI, despite ongoing legal issues.
- User-Driven Refinement: Encourages feedback to continuously improve accuracy, speed, and reliability.
Introducing the Perplexity Deep Research Tool
Well, folks, the AI research arms race just got a bit more interesting—and a lot more affordable. Perplexity AI has officially unveiled its new “Deep Research” tool, and it’s not just an incremental update. This is a full-on assault on giants like OpenAI and Google, all wrapped up in a lovely freemium bow.
Launched on 15 February 2025, Deep Research is designed to cut your time spent trawling the internet down to mere minutes. Whether you’re investigating market trends in finance, refining marketing campaigns, or just planning your dream holiday, Perplexity aims to do the legwork for you—dozens of searches, hundreds of sources, and a neat summary at the end. And best of all, it’s free for casual use, with a paid tier if you really want to ramp up your daily queries.
Key Features and Capabilities
One of the big selling points Perplexity emphasises is its iterative search and reasoning process, which is meant to mirror the thought process of a human researcher. Here’s the gist of what you get:
- Expert-Level Analysis
Tackles finance, marketing, technology, health, product research, travel planning, and more. Essentially, it’s like having a mini think tank at your disposal. - Automated Deep-Dive
Performs dozens of individual searches, sifts through hundreds of online sources, and draws it all into a comprehensive report. So no more 30-tab chaos in your browser. - Lightning-Fast Turnaround
Delivers results in 2–4 minutes (under three minutes in most cases). Competitors like OpenAI’s Deep Research take 5–30 minutes for complex queries. - Shareable Outputs
Once you get your report, you can export it as a PDF or transform it into a Perplexity Page for easy sharing with colleagues or friends. - Citations for Transparency
Deep Research includes references and citations to its sources, so you can cross-check, trust, or verify any points. Great for academic or professional work where you need that extra layer of credibility. - Iterative Refinement
The system learns as it goes. It reads a chunk, decides what else it needs to look for, and continues to refine its approach until it’s satisfied it has a well-rounded view.
Pricing and Accessibility
According to Perplexity’s own website, the freemium model is turning heads. Here’s the breakdown:
- Free Tier:
Limited to around 5 Deep Research queries per day. Perfect if you just need the occasional deep-dive or want to give it a whirl before upgrading. - Pro Subscribers ($20/month):
Access to 500 daily queries, a huge jump in usage allowance. That’s significantly undercutting some major players, like OpenAI’s $200/month plan.
In comparison to enterprise-level AI research tools (which can run up to $75,000 per month), Perplexity is practically handing out advanced AI research on a silver platter, at least from a cost perspective. The company’s strategy here seems to be a combination of “democratise AI research” and “force the big boys to rethink their pricing.”
Performance Benchmarks
Alright, so how does Deep Research stack up under the hood? Let’s get to the numbers:
- Humanity’s Last Exam:
- Perplexity’s Deep Research: 21.1%
- OpenAI’s Deep Research: 26.6%
- Google’s Gemini Thinking: 6.2%
- Grok-2: 3.8%
- GPT-4o: 3.3%
- While Perplexity lags a bit behind OpenAI’s top-tier offering, it beats other well-known AI models by a fair margin.
- SimpleQA Benchmark:
- Perplexity’s Deep Research: 93.9% accuracy
So, yes, it’s not absolutely top of the tree on some metrics, but it’s definitely competitive—and far ahead of many alternatives.
Market Impact and Competition
This move puts Perplexity squarely on a collision course with OpenAI and Google, who’ve been jockeying for position in the advanced AI research space. It’s worth noting:
- Company Growth:
Perplexity was founded in 2022 by ex-OpenAI researcher Aravind Srinivas. Rapid expansions, plus a $500 million funding round last December, have given it a staggering $9 billion valuation. Investors include heavyweights like Jeff Bezos and Nvidia. - Legal Hurdles:
Perplexity is currently wrestling lawsuits from some media organisations over alleged unauthorised use of copyrighted articles. They’re seeking to mollify publishers with collaboration deals and revenue-sharing agreements, signing the likes of Time and Fortune. - Competition:
Google is well-established, and OpenAI has name-brand recognition plus GPT’s massive user base, but Perplexity’s “high-value, low-price” approach could disrupt the market if it continues to deliver on speed and research depth.
Challenges and Future Outlook
Despite the buzz, Perplexity’s Deep Research isn’t without its pitfalls:
- Accuracy vs Speed
While 21.1% on Humanity’s Last Exam is solid, it’s still below OpenAI’s 26.6%. Perplexity’s big claim is that it outdoes many rivals on speed. But for critical, expert-level tasks, some users may still lean towards the highest possible accuracy. - Ethical Concerns
Like all AI research tools, there are worries about diminishing human critical thinking and potential reliance on “fast” answers over “deeply validated” insights. That’s an industry-wide conversation that won’t end soon. - Ongoing Lawsuits
The legal back-and-forth with major media outlets is no small matter. To remain fully legit (and maintain public trust), Perplexity will likely have to sign more licensing deals or refine how it sources content. - User Trust and Adoption
Breaking into a space dominated by OpenAI and Google is no walk in the park. Even if their product is fantastic, Perplexity needs to keep scaling its user base while handling the stress test that a wave of new users can bring.
Despite these challenges, Perplexity is forging ahead with expansions to iOS, Android, and Mac platforms, as well as continued refinement of its models. If they strike the right balance between accessibility, accuracy, and cost, there’s a good chance they can secure a sizeable share of the AI research pie.
Why Speed and Cost Matter in AI Research
The AI research sphere is going through a bit of a metamorphosis: big enterprise solutions can cost tens of thousands per month, yet more and more small businesses, freelancers, and academics also want access to advanced AI. By offering a freemium tier that performs at a near-competitive level with top-tier solutions, Perplexity is effectively lowering the barrier to entry for advanced research.
And speed? If you can get a comprehensive, properly cited report in 3 minutes rather than 30, that’s a massive productivity win. For time-critical fields like finance, health, or real-time marketing campaigns, it can be the difference between making the right call or missing an opportunity.
Potential Applications Across Industries
Perplexity is keen to emphasize how Deep Research can handle multiple verticals:
- Finance: Collating market data, generating forecasts, and providing real-time financial analysis.
- Marketing: Performing competitor analysis, consumer behavior insights, and strategic planning.
- Technology: Deep dives into emerging tech trends, algorithmic benchmarks, and scoping out R&D projects.
- Health: Acting as a personal consultant for health and wellness research (with obvious caveats that it’s not a medical professional!).
- Travel Planning: From recommending itineraries to budgeting and flight/hotel comparisons.
- Product Research: Assessing product features, user sentiment, and market viability.
Essentially, if you need to wade through lots of data quickly, Deep Research might be your new best mate.
Accuracy, Trust, and User Feedback
Ensuring Accuracy
How does Perplexity aim to keep the nonsense and hallucinations to a minimum? They rely on a variety of strategies:
- Searching Hundreds of Sources
It’s all about cross-referencing. If 90% of sources say “X,” the final answer probably leans that way. - Trust-Based Ranking
A PageRank-esque system that looks at source credibility, giving more weight to, say, reputable news outlets over random forums. - Ongoing User Feedback
Perhaps the most crucial: they allow users to flag dodgy info or provide improvements, using that data to retrain and refine the model.
Feedback Loop
Perplexity encourages a user-driven improvement cycle. Types of feedback they actively seek:
- Accuracy Assessment: Point out any wrong or outdated info.
- Source Quality: Let them know if a chosen source is questionable or irrelevant.
- Comprehensive Coverage: Tell them if the final report missed a critical subtopic.
- User Experience: Interface or design tweaks that could smooth out the workflow.
- Domain-Specific Nuances: For fields like finance or health, domain experts can highlight deeper complexities to refine the AI’s output.
This iterative approach helps Perplexity calibrate its models over time, building a more reliable system that better meets user expectations.
Final Thoughts
So there you have it—Perplexity’s Deep Research is here, and it’s looking to shake up the AI research market by delivering swift, thorough, and fairly accurate results without the punishing subscription fees of some competitors. While it may not quite surpass OpenAI’s top-tier solution in raw accuracy, it’s coming close enough for most everyday use cases—and, in some respects, it’s leaving everyone else in the dust.
The company’s growth, ambitious partnerships, and willingness to face legal and ethical questions head-on show that Perplexity is more than a flash in the pan. If you’re in finance, marketing, or tech, or simply a curious researcher wanting a user-friendly, budget-friendly AI tool, you might want to take Deep Research for a spin. It’s free, after all, so why not?
One thing’s for sure: with market leaders being nudged by smaller but nimble players like Perplexity, the AI research landscape will only get more interesting—and more competitive. Watch this space.
What do YOU think?
Will Perplexity’s freemium ‘Deep Research’ tool be the breakthrough that finally topples AI giants like OpenAI and Google, or is this just a temporary shake-up in an ever-evolving battlefield? Let us know in the comments below.
You may also like:
- Perplexity Assistant: The New AI Contender Taking on ChatGPT and Gemini
- Perplexity Introduces Pages: The Newest AI Article Generator
- Can Singtel’s Free Access to Perplexity Pro Redefine AI Search?
- Or try out Perplexity AI for free by tapping here.
Author
Discover more from AIinASIA
Subscribe to get the latest posts sent to your email.
You may like
-
Upgrade Your ChatGPT Game With These 5 Prompts Tips
-
If AI Kills the Open Web, What’s Next?
-
Build Your Own Custom GPT in Under 30 Minutes – Step-by-Step Beginner’s Guide
-
Adrian’s Arena: Stop Collecting AI Tools and Start Building a Stack
-
GPT-5 Is Less About Revolution, More About Refinement
-
Apple’s China AI pivot puts Washington on edge
Tools
Upgrade Your ChatGPT Game With These 5 Prompts Tips
Most people ask ChatGPT the wrong way. These 5 prompt upgrades will train the AI to think sharper and deliver smarter answers every time.
Published
1 day agoon
May 29, 2025By
AIinAsia
What if the problem isn’t ChatGPT — but how you’re talking to it? As businesses across Asia scramble to integrate AI into daily workflows, far too many professionals are getting half-baked answers and wasting time refining prompts. Precision equals clarity. And clarity starts with knowing what to ask. Read on for 5 elite ChatGPT prompt tips.
TL;DR — What You Need To Know
- ChatGPT often gives vague or flawed answers because the prompt lacks structure
- Adding 5 targeted follow-up requests can dramatically improve output quality
- Smart prompting forces the AI to self-check, ask better questions, and aim higher
- These techniques turn ChatGPT into a sharper thinking partner for professionals
1. Make ChatGPT Score Its Own Work
Like a student handing in homework and marking their own essay, ChatGPT performs better when it’s made to judge itself. Give it a red pen:
“I want you to assess your response against this checklist. Rate your answer on a scale of 1-10 for each of these criteria: accuracy, completeness, relevance, clarity, and practical usefulness…”
By forcing the AI to reflect on its own output, you raise the baseline for what counts as “done.” You get more precise responses — and fewer excuses to settle for less.
2. Demand Reasoning, Not Just Answers
Don’t let ChatGPT blag its way through your questions like a first-year intern. Make it show its working — maths teacher-style.
“For each main point in your response, explain your reasoning process…”
This single move helps you spot weak assumptions and gives you greater confidence in the result. Especially important when decisions — or reputations — are on the line.
3. Make It Ask the Right Questions First
If your prompt sounds like a half-baked WhatsApp message at 2am, don’t expect brilliance. Before ChatGPT replies, let it play detective.
“Before giving me any answer, point out exactly what I’ve left out of my request…”
It’s the kind of intelligent friction that turns vague ideas into actionable prompts. Think of it as ChatGPT playing devil’s advocate before it turns into your co-pilot.
4. Find Your Blind Spots Early
Even the best minds occasionally forget the obvious. Enter ChatGPT, your AI-powered Socrates, asking the awkward questions others won’t.
“Review my request and tell me what angles I’m completely missing…”
This elevates the conversation. ChatGPT stops being a yes-man and starts acting like the curious challenger every business leader needs.
5. Push It to Think Like an Expert
You wouldn’t ask a junior exec to run your quarterly strategy — so don’t let ChatGPT deliver B-grade insights. Demand elite thinking.
“Respond to my question as if you were in the top 1% of experts in my field…”
This unlocks deeper insights, smarter trade-offs, and far more useful recommendations — especially when your questions relate to strategy, branding, or customer psychology.
Get Better Answers From ChatGPT Every Time
The best ChatGPT users aren’t the ones with the fanciest tools. They’re the ones who ask better questions. Use these five prompt upgrades as your new default. Over time, the difference isn’t just better responses — it’s better thinking.
Share YOUR ChatGPT prompt tips!
Do your ChatGPT answers leave you with digital drivel? Share your own tips and tricks in the comments below!
You may also like:
- 7 Effective AI Prompt Strategies to Elevate Your Results Instantly
- Game-Changing Google Gemini Tips for Tech-Savvy Asians
- Or tap here to try these ChatGPT prompt tips now
Author
Discover more from AIinASIA
Subscribe to get the latest posts sent to your email.
Learning
Build Your Own Custom GPT in Under 30 Minutes – Step-by-Step Beginner’s Guide
Create your own GPT without writing code. This beginner-friendly guide shows you how to build, customise, and publish a ChatGPT assistant using OpenAI’s builder — complete with personality, knowledge, and tools.
Published
4 days agoon
May 26, 2025
A friendly guide to building your personalised ChatGPT assistant (custom GPT creation) in under 30 minutes
The Quick Essentials
Before we dive in, here’s what you need to know:
- You’ll need a ChatGPT Plus or Enterprise account (the paid version)
- Creating a custom GPT involves naming it, giving it a personality, uploading knowledge files, and enabling tools
- The entire process happens at chat.openai.com/gpts
- No coding required — just plain English instructions
- You can keep your GPT private, share it via link, or publish it in the GPT Store
Step 1: Getting Access
First things first, you need a ChatGPT Plus subscription:
- Head over to chat.openai.com
- Click “Upgrade to Plus” if you haven’t subscribed yet
- Confirm your plan and payment (USD $20/month at the time of writing)
Once you’re all subscribed:
- Navigate to: chat.openai.com/gpts
- Click the black “Explore GPTs” button in the left menu
- At the top right, click “Create”

Brilliant! You’re now in the GPT Builder interface where all the magic happens.
Step 2: Starting in “Create” Mode
You’ll see a chat interface asking: “What would you like to make?”
This is where you describe your GPT in plain language. For example:
“I’d like to create a friendly GPT that helps junior marketers in the UK write social media copy. It should use casual British English and understand cultural references from across the UK.”

The system will respond with follow-up questions like:
- What tone should it use?
- Should it browse the internet or run code?
- Will you upload any knowledge files?
Answer naturally and conversationally. The system builds a draft GPT based on your answers.

It may even recommend a name and an image:

Once the chat version feels roughly right, click “Configure” in the top bar to manually fine-tune everything.

Step 3: Configuring Your GPT
This is the control panel where you can edit every setting. Follow these steps:
3.1 — Name Your GPT
- In the Name field: give it a clear name like “Brit Copy Buddy”
- In Description: write what it does — “Helps junior UK marketers write scroll-stopping social copy in casual British English.”
3.2 — Write the Instructions
Scroll down to the Instructions box and type your behaviour settings. Think of this like a personality manual. If you followed the steps, then it may well already be filled in for you. This is where the magic happens, so make sure it truly reflects your purpose. Because it is so important, we created a separate guide for this which you can read here. For example:
This GPT is a friendly and culturally aware assistant designed to help junior marketers in the UK write engaging social media copy. It communicates in a casual, approachable tone using British English, including regionally familiar slang and references from across England, Scotland, Wales, and Northern Ireland. It provides clear, supportive guidance to help users improve their writing, offering creative suggestions while keeping brand voice and audience in mind. It can reference UK cultural events, holidays, humour, and idioms to make content feel local and relevant. It avoids Americanisms and ensures that grammar, spelling, and phrasing are aligned with UK standards. The assistant will ask for context when needed (e.g., target audience or platform), and will aim to keep things breezy, witty, and scroll-stopping.
3.3 — Add Conversation Starters
Under Conversation Starters, add 3–4 useful prompts users might click on:
- “Help me write a tweet for a UK skincare brand launch”
- “Can you make this Instagram caption sound more British?”
- “Draft some TikTok captions about a new meal deal”
This helps users jump straight in without typing from scratch.

Step 4: Adding Custom Knowledge (Optional but Recommended)
If you want the GPT to reference your own documents — like brand guidelines or FAQs — follow these steps:
- Scroll to the Knowledge section
- Drag and drop your files (accepted formats: .pdf, .txt, .csv, .md, .json)
- Upload limit is around 20 files at present
Example: Upload a “Tone of Voice Guide.pdf” and your GPT will use it to match your brand style.

Important note: Your GPT can reference but not quote files word-for-word. It learns the content conceptually rather than memorising exact phrases.
We have created a step-by-step guide for the best way to structure your Knowledge files here.
Step 5: Enabling Tools
Now choose which abilities your GPT should have.
Scroll to Capabilities, and toggle the following:
- ✅ Web Browsing — useful for real-time info like news or trends
- ✅ Code Interpreter — for handling files, calculations, data plots
- ✅ Image Generation — if you want it to create pictures (e.g., Instagram ideas)
- ✅ File Uploads — lets users feed the GPT spreadsheets or PDFs
Select only what’s genuinely useful — too many tools can make your GPT’s responses confusing.

Step 6: Testing Your GPT Thoroughly
Before publishing, have a proper chat with your GPT.
Ask both straightforward and unusual questions to test:
- Does it maintain the right tone?
- Does it understand your uploaded files?
- Does it use tools properly (like making charts or browsing)?
- Does it clarify things when uncertain or guess incorrectly?
If anything feels off, go back to the Instructions and tweak your wording. Even changing one line can make a significant difference.
Step 7: Create New Action
This function allows you to turn your GPT into a powerful API-aware assistant that can fetch data, trigger services, or complete tasks — all from inside the chat.
This is an optional extra when creating a Custom GPT and is complicated enough to need its own guide, which you can read here.
If this is your first attempt at creating a Custom GPT, we suggest skipping this step for now and moving on to Step 8.
Step 8: Switch Off Training
You should always consider anything sensitive you share with any AI chatbot. However, its always a good idea to switch off the request to improve the AI models.

Step 9: Publishing Your GPT
At the bottom right, click the “Publish” button.
You’ll be asked to choose:
- Private — only visible to you
- Unlisted — only people with your link can access it
- Public — listed on the GPT Store for anyone to use
Give it a thumbnail image (upload one or use the auto-generated option), choose a category (e.g., Marketing, Productivity), and confirm.

Congratulations! Your GPT is now live and you can view it.


Final Tips for Success
- 🧹 Keep it focused — One GPT = one clear purpose
- 🪪 Use your brand voice — match the tone your users expect
- 🔁 Iterate regularly — update your files and instructions as you learn what works
- 💬 Share wisely — use private/unlisted first before going public
Happy GPT creating! With these steps, you’ll be up and running with your custom assistant in no time.
You may also find useful:
- How to Upload Knowledge into Your Custom GPT
- How to Use the “Create an Action” Feature in Custom GPTs
- Or try this playbook out now at ChatGPT by tapping here.
Author
-
Adrian is an AI, marketing, and technology strategist based in Asia, with over 25 years of experience in the region. Originally from the UK, he has worked with some of the world’s largest tech companies and successfully built and sold several tech businesses. Currently, Adrian leads commercial strategy and negotiations at one of ASEAN’s largest AI companies. Driven by a passion to empower startups and small businesses, he dedicates his spare time to helping them boost performance and efficiency by embracing AI tools. His expertise spans growth and strategy, sales and marketing, go-to-market strategy, AI integration, startup mentoring, and investments. View all posts
Discover more from AIinASIA
Subscribe to get the latest posts sent to your email.
Learning
How to Upload Knowledge into Your Custom GPT
Want your Custom GPT to actually know your stuff? Learn how to upload your own documents into ChatGPT’s Knowledge section, with step-by-step instructions, supported file formats, and key watchouts.
Published
1 week agoon
May 20, 2025By
AIinAsia
Your step-by-step guide to making a smarter GPT with your own documents.
Why Upload Your Own Knowledge?
Even the best GPT won’t really sound like you — or know your stuff — unless you teach it. That’s where knowledge uploads come in.
By uploading files (PDFs, DOCs, TXT, etc.), your GPT can:
- Answer questions based on your business material
- Speak in your tone and style
- Cut down on repetitive manual responses
- Act as a trained assistant, contract reviewer, customer explainer, or internal helpdesk
Think of it as giving your GPT “homework.” You hand it your documents, and it quietly studies them behind the scenes so it can sound smart in front of your users.
Why Upload Your Own Knowledge?
Even the best GPT won’t truly sound like you — or know your stuff — unless you teach it. That’s where knowledge uploads come in.
By uploading files (PDFs, DOCs, TXT, etc.), your GPT can:
- Answer questions based on your business material
- Speak in your tone and style
- Cut down on repetitive manual responses
- Act as a trained assistant, contract reviewer, customer explainer, or internal helpdesk
Think of it as giving your GPT “homework.” You hand it your documents, and it quietly studies them behind the scenes so it can sound smart in front of your users.
Step-by-Step: How to Upload Documents to a Custom GPT
Prerequisite: You’ve already created your Custom GPT (via https://chat.openai.com/gpts). You’re now ready to add your own knowledge base by uploading documents.
Step 1: Go to the GPT Builder
Go to https://chat.openai.com/gpts. Click on your Custom GPT and select “Edit GPT.”
Step 2: Find the “Knowledge” Section
In the left-hand menu, select “Knowledge” and click “Upload files.”
Step 3: Add Your Files
Drag and drop or browse to upload your documents. Supported formats include .pdf, .docx, .txt, .md, .csv. You can upload up to 20 files with a combined size of 512 MB. You can ask ChatGPT to help you assess and convert documents into these formats if you need to with this prompt:
I want to use this document inside a Custom GPT as part of its Knowledge section. Please assess the content and do the following:
Identify if this content is suitable to be uploaded directly (e.g. clear, clean, complete), or if it needs to be rewritten, summarised, or broken into smaller chunks.
If the formatting is poor (e.g. tables, layout issues, scanned PDF style), convert it into clean, text-based markdown or plain text format that preserves all meaning and structure.
Remove any unnecessary elements such as headers/footers, page numbers, duplicated content, or visual formatting that won’t translate well into plain text.
Structure the output into a clean, well-labelled text file that can be uploaded into the Knowledge section of a Custom GPT (i.e. .txt or .md format). Use clear section titles and bullet points where appropriate.
Keep all the important content, but make sure it’s optimised for retrieval by a GPT model. That means using simple, clear language and logical structure.
Name the output file appropriately (e.g. “2025_PricingOverview.txt” or “Legal_Terms_Guide.md”).
Please begin by assessing the suitability of the input and then output a clean, upload-ready version.
[Optional Tip (if you’re uploading a file):
Start with:]
“Please assess the uploaded file using the instructions below…” and paste the prompt afterward.
Step 4: Check the File List
You’ll see a list of your uploaded files. Use the trash icon to remove any if needed. You can update this list at any time.
Step 5: Save and Publish
Click “Save” or “Publish” to apply your changes. Your GPT can now access your uploaded documents to answer relevant prompts.
What Kind of Files Work Best?
Ideal Files:
- Cleanly written PDFs (guides, SOPs, FAQs)
- Contracts and legal templates
- Onboarding documents, pricing sheets
- Internal wikis (exported to .txt or .md)
Avoid These:
- Scanned documents with images
- Slides with only images or no speaker notes
- Encrypted or locked PDFs
- Files full of links without explanations
Tip: For web pages, copy-paste the content into a clean .txt or .md file.
How the GPT Uses This Info
Your GPT will search the uploaded documents in real time when a relevant prompt is given. It doesn’t memorise the content — it retrieves from it. It performs best when the material is clearly written and structured.
Watch Outs
- No File Structuring = Confused GPT
If you upload a single giant PDF with 50 topics and poor formatting, the GPT will struggle. Break it into smaller, well-labelled files. - Bad Formatting = Bad Responses
If your file has unusual fonts, broken tables, or visual layouts (especially common in PDFs), the GPT may misread it. Clean formats like .txt, .docx, or markdown work best. - No Source Citations
By default, GPT won’t say where the information came from. If this matters, add an instruction like: “Always mention which document you’re referencing.” - File Limit
You can only upload 20 files per GPT. Curate carefully and consider trimming or combining related documents.
Curating the “Core Knowledge” for Best Results
Ask yourself:
- What do I want this GPT to do? Only upload documents relevant to those tasks.
- Will someone else use this? Include glossaries or context if needed.
- Is this content clear and self-contained? If not, simplify or split into manageable chunks.
Example Use Cases:
LegalGPT: Upload contracts, clause trackers, fallback templates
SalesGPT: Upload pitch decks, product specs, objection-handling guides
HRGPT: Upload company policies, onboarding FAQs
Bonus Tip: Pair With System Instructions
After uploading, adjust your GPT’s instructions to reflect how it should use that knowledge. Example: “You are a helpful assistant trained specifically on SQREEM’s legal contracts and internal SOPs. Always answer using information from the uploaded documents. If unsure, say ‘I’m not certain — please check with legal.’” You can also use the “Prompt Starter” section to load reusable queries.
Updating Your Knowledge Files Later
Return to “Edit GPT > Knowledge” anytime to remove outdated files or upload new ones. Save to apply changes. Your GPT will instantly use the latest content.
Version Control and Multiple GPTs
You can create multiple GPTs with different document sets, or use file naming conventions to stay organised. Examples:
01_PricingOverview_Q1-2025.txt
02_TOS_Updated_April2025.docx
03_FAQ_InternalOnly.md
What To Do Next
Now that your GPT has your content, test it by asking:
“Summarise our latest pricing model”
“What’s our refund policy?”
“Write a client email using our onboarding flow”
“Check clause 7.3 in the uploaded SOW template”
You May Also Find Useful:
Author
Discover more from AIinASIA
Subscribe to get the latest posts sent to your email.

Upgrade Your ChatGPT Game With These 5 Prompts Tips

If AI Kills the Open Web, What’s Next?

Build Your Own Custom GPT in Under 30 Minutes – Step-by-Step Beginner’s Guide
Trending
-
Life2 weeks ago
7 Mind-Blowing New ChatGPT Use Cases in 2025
-
Learning1 week ago
How to Use the “Create an Action” Feature in Custom GPTs
-
Business2 weeks ago
AI Just Killed 8 Jobs… But Created 15 New Ones Paying £100k+
-
Tools3 weeks ago
Edit AI Images on the Go with Gemini’s New Update
-
Learning4 days ago
Build Your Own Custom GPT in Under 30 Minutes – Step-by-Step Beginner’s Guide
-
Learning1 week ago
How to Upload Knowledge into Your Custom GPT
-
Business1 week ago
Adrian’s Arena: Stop Collecting AI Tools and Start Building a Stack
-
Life2 weeks ago
Adrian’s Arena: Will AI Get You Fired? 9 Mistakes That Could Cost You Everything