Kind: captions
Language: en
You're probably clicking around Grock's
interface wondering what half these
buttons actually do. Maybe even
frustrated that you're not getting the
answers you expected. Well, I spent
weeks testing every single feature in
Grock's interface. From the mysterious
deep search button to those confusing
model modes, and here's what surprised
me. Most people are using maybe 20% of
what Grock can actually do. And that
changes today. Welcome back to
bitbiased.ai, AI, where we do the
research so you don't have to. Join our
community of AI enthusiasts. Click the
newsletter link in the description for
weekly analysis delivered straight to
your inbox. So, in this video, I'm
walking you through every visible
feature in the Gro interface, explaining
exactly what each one does, when to use
it, and how to get dramatically better
answers. By the end, you'll know how to
match the right tool to your task.
Whether that's lightning fast responses
or deep research that pulls from dozens
of sources in seconds. First up, let's
start where every Grock interaction
begins. That search bar at the top, the
search bar, where it all begins. Now,
the search bar might look simple, almost
boring even. But here's where things get
interesting.
This isn't just a text box. It's a smart
prompt field that understands context in
ways that will surprise you. Think of it
as your direct line to one of the most
advanced AI systems available right now.
Developed by XAI to push the boundaries
of what conversational AI can do. Here's
how it actually works.
You click into that bar and type
whatever's on your mind. Could be a
factual question like, "What are neural
networks?" or something more creative
like, "Write me a short story about time
travel.
The beauty is in what happens next
because Grock doesn't just answer and
forget.
It remembers your conversation, building
on what you've already discussed without
making you repeat yourself every single
time.
Let me give you a real example of why
this matters. Say you start with that
neural networks question.
Grock gives you an explanation. Then you
follow up with, great, now explain it
with an analogy. And here's the magic.
You didn't have to say explain neural
networks with an analogy.
Grock knew what it meant because it's
tracking your conversation thread.
This multi-turn memory means you can
have an actual dialogue, not just a
series of disconnected questions. Now,
when should you use the search bar? The
answer is pretty much always.
Whether you need a quick fact, detailed
coding help, creative writing
assistance, or just want to chat about
philosophy at 2 in the morning, this is
your starting point. And if you're on
mobile, some versions even let you speak
your queries instead of typing. Perfect
for when you're driving or your hands
are full. But here's what most people
don't realize. How you phrase your
question in this search bar can
completely change the quality of answer
you get. We'll talk more about that when
we get to prompting techniques, but for
now, just know this simple box is more
powerful than it looks. The real
benefit, speed and accessibility.
The search bar is always there, always
ready, connected to real-time data from
X, formerly Twitter, and the broader
web. So, when you ask Grock something,
you're not just getting canned responses
from old training data.
You're tapping into current information,
often with that signature wit and humor
Grock is known for. It makes the whole
experience feel less like querying a
machine and more like chatting with a
really knowledgeable friend who happens
to have the entire internet at their
fingertips. Deep search. When you need
more than surface level answers, all
right, next to that search bar, you'll
spot something called deep search. And
trust me, this is where Grock starts to
feel less like a chatbot and more like
your personal research assistant. Most
AI tools will give you a quick answer
and call it a day. But what if you need
more? What if you want the full story
backed by multiple sources synthesized
into something you can actually use?
That's exactly what Deep Search does.
When you click this button, you're
telling Grock to stop playing it safe
with quick responses and instead go
allin on your question.
It scour the internet, pulls from dozens
of credible sources, analyzes current
discussions on X, and then presents you
with a comprehensive report. We're
talking the kind of research that would
take you an hour of Googling and tab
switching delivered in about a minute.
Here's a scenario where this becomes
absolutely game-changing.
Let's say you're researching the latest
developments in renewable energy
storage. Without deep search, you might
get a decent overview based on Grock's
training data. But with deep search
activated,
Grock goes out and finds recent news
articles, research papers, expert
commentary, trending discussions,
everything relevant, and synthesizes it
into one coherent answer.
One user reported getting a detailed
report citing 22 different sources in
just over 60 seconds.
Try doing that manually and see how long
it takes you. But here's what I love
most about deep search. Transparency.
Grock doesn't just dump information on
you and expect you to trust it blindly.
It shows you where the information came
from. You get citations, references, the
whole nine yards. This means you can
verify facts, dig deeper into specific
sources if you want, and actually trust
what you're reading. It's like having a
research assistant who not only does the
work, but shows their receipts.
Now, when should you use deep search?
Think of it this way. If your question
starts with phrases like research,
summarize, what are the latest, or
compare different perspectives on, hit
that deep search button. It's perfect
for current events, broad topics that
need multiple viewpoints, or anytime
you're working on something that
requires depth. On the flip side, if
you're asking, "What's 2 plus 2?" or
"Who wrote 1984?"
Save yourself the time. Regular mode is
plenty fast and accurate for
straightforward queries. To activate it,
just type your question, click the deep
search button before you send it, and
let Grock work its magic.
Some interfaces even have preset
research prompts that automatically
enable this feature.
You'll see a little progress indicator
as Grock pulls information from across
the web and then boom, detailed answer
with sources. It's like the difference
between asking someone for quick
directions versus asking them to plan
your entire road trip with alternate
routes, gas stations, and scenic stops.
Both are useful, but one goes way
deeper. create image. Turning your words
into visuals.
Right next to deep search, there's
another feature that's honestly just fun
to play with. The create image button.
[clears throat] And while it's fun, it's
also genuinely useful if you know when
to deploy it. This feature taps into
Grock's image generation capabilities,
letting you describe something in words
and watch it come to life visually.
Here's how it works.
When you click create image, Grock
switches into what they call imagine
mode.
Instead of responding with text, it
fires up its image generation engine,
code named Aurora for the base version
with an enhanced imagine model for
premium users and creates a picture
based on your prompt.
Want to see a futuristic city skyline at
sunset? Type that, hit create image, and
within seconds, you're looking at a
digital rendering that can be
surprisingly photorealistic.
But wait, before you think this is just
a toy for making random pictures, let me
show you the practical side.
Say you're a teacher trying to explain
the solar system. You could describe it
in words, or you could ask Grock to
create an image of the solar system with
labeled planets.
Boom. Instant visual aid.
Content creators use this for concept
art, social media graphics, design
inspiration. Marketers might generate
logo concepts. Generate a logo for a
cafe named Lunar Coffee featuring a moon
and coffee cup. Even if the first result
isn't perfect, it gives you something to
iterate on, to refine, to spark ideas.
The beauty of having this integrated
into the same chat interface means you
don't need to bounce between tools.
You're already working with Grock for
textbased answers and suddenly you need
a visual to illustrate your point. Just
switch modes. Want a fantasy landscape
with floating islands? A diagram
comparing electric car battery ranges, a
visual metaphor of a neural network as a
futuristic city? All of this is possible
with a well-crafted prompt. And speaking
of prompts, this is where specificity
really pays off. The more detail you
include in your image prompt, the better
the result.
Don't just say, "Draw a landscape." Say,
draw a fantasy landscape with floating
islands connected by rope bridges,
waterfalls cascading into clouds below,
and a sunset casting orange and purple
hues across the sky.
See the difference? You're painting a
picture with words, and Grock is
bringing that picture to life. Color,
style, perspective, mood. The more you
specify, the closer you get to your
vision. When should you use create
image? Anytime words aren't enough,
visualization exercises, design
mock-ups, educational diagrams, creative
storytelling, brainstorming sessions
where you need to see your ideas. And if
the first image isn't quite right, no
problem. Tweak your prompt and try
again. It's an iterative process, just
like refining a text answer. The feature
is there to enhance your workflow, to
add a visual dimension to your AI
assisted work. Don't sleep on it.
Model selector. Choosing your
intelligence level. Now we're getting to
one of the most powerful parts of the
Gro interface, the model selector. This
is where you actually control what's
happening under the hood. Choosing the
AI model that'll handle your query. And
here's what most people miss. Different
tasks need different levels of AI
horsepower. Sometimes you want fast,
sometimes you want smart, sometimes you
want both. The model selector gives you
that control. Let me break down your
options. Auto, fast, expert, Grock 4,
fast, and heavy. Each one represents a
different trade-off between speed and
depth. Think of it like gears in a car.
You wouldn't drive on the highway in
first gear, and you wouldn't climb a
steep hill in fifth. Same principle
here. Match the mode to your task for
the best results.
Let's start with auto mode. If you're
not sure which model to pick, this is
your safe bet. Grock automatically
decides what's appropriate for your
question, balancing speed and
sophistication on the fly.
Simple factual question.
Auto might use a faster model complex
reasoning task.
It'll switch to something more advanced.
The benefit here is convenience. You
don't have to think about it. and Grock
handles the decision-m most casual users
probably leave it on auto and never
think twice. And honestly, that's fine.
But then there's fast mode.
This is where you prioritize speed above
all else. Fast mode uses lighter models,
typically Gro 3 or Gro 3 Mini, to get
you answers almost instantly. Perfect
for straightforward questions like,
"What's the capital of France?" or
define polymorphism in programming.
You're on your phone with a slow
connection. Fast mode saves data and
time.
The answer might be slightly less
nuanced than what you'd get in expert
mode, but for simple queries, you won't
notice the difference. Speed is the name
of the game here. Now, when you want
quality over speed, you switch to expert
mode. This engages Grock 4, a more
advanced model that provides detailed,
nuanced responses.
Think of it as telling Grock to take its
time and be thorough. Ask about the
significance of the Higs Boson. An
expert mode will give you a carefully
structured explanation that breaks down
the concept thoughtfully.
It's ideal for complex questions,
research tasks, creative projects, or
anything where the quality of the answer
actually matters. The trade-off, it
might take a few extra seconds, but the
depth you get is usually worth the wait.
Then there's Gro 4 fast, and this is
where things get interesting. It's
essentially a lighter, quicker version
of Gro 4. You get GPT4 level
intelligence, but with faster responses
and slightly less compute intensity. One
source noted it uses about 40% fewer
reasoning tokens than full Gro 4, which
translates to speed without sacrificing
too much quality. This is the sweet spot
for many tasks. Detailed enough to
handle moderately complex questions,
fast enough that you're not sitting
around waiting. If you have access to
this mode, it's great for things like
compare electric cars and hydrogen fuel
cell cars in terms of efficiency, cost,
and infrastructure.
where fast mode wouldn't give you enough
depth, but you also don't want to wait
for heavy mode. And speaking of heavy
mode, this is the beast. Gro 4. Heavy is
the most powerful model in the
ecosystem. Described by some users as a
computational powerhouse. It's a
multi-agent enhanced version with
expanded memory. We're talking up to
128,000 tokens or more. What does that
mean in practice? You can feed it entire
documents, large code bases, lengthy
conversations, and it'll remember and
analyze all of it. Heavy mode is for
when nothing else will do. Complex math
proofs, analyzing legal contracts,
debugging massive code files, designing
detailed strategic plans. It uses
advanced reasoning strategies, sometimes
showing you step by step how it arrived
at an answer, which is incredibly
valuable for learning or verification.
But here's the catch. Heavy mode is
premium.
You need a super grock subscription to
access it and it's slower because it's
doing serious computational heavy
lifting.
Think of heavy as the mode you save for
your toughest challenges when you need
the smartest possible AI assistant and
you're willing to wait a bit for that
next level insight. One user described
trying to get Gro 4 heavy to tackle
problems that few other models could
handle and it delivered.
That's the power you're dealing with.
So, how do you choose?
For everyday questions, stick with auto
or fast. For important work that needs
quality, go expert or Grock 4 fast.
For the absolute toughest tasks,
research papers, complex coding,
strategic analysis, breakout heavy mode
if you have access. The model selector
is all about control. Don't just leave
it on default and forget about it. Think
about what you're asking. Choose
accordingly and watch your results
improve dramatically.
Custom instructions making Grock truly
yours. Now, let's get into something
that can dramatically improve your
experience, but that a lot of people
overlook. Custom instructions.
This is typically found in a settings
menu or gear icon, and it's essentially
your chance to teach Grock about you and
how you like to work. Here's the
concept. Every time you start a new
conversation, Grock doesn't know
anything about you by default. But with
custom instructions, you can give Grock
a bit of permanent context that it'll
remember for all future chats. It's like
onboarding a new assistant with exactly
what they need to know to work with you
effectively. When you open the custom
instructions section, you'll see prompts
like, "What would you like Grock to know
about you?" and "How would you like
Grock to respond?"
These are your opportunity to set the
ground rules in the first field. Tell
Grock about yourself.
Are you a beginner programmer learning
Python for data science? Say so. Now
Grock knows to keep explanations simple.
Use beginner-friendly terminology. Maybe
focus examples on data science
applications.
Are you a marketing professional who
needs business focused insights? Mention
that. You can even describe ongoing
projects or specific interests. In the
second field, you set the tone and
style. Maybe you want concise, formal
answers. Tell Grock that. Maybe you
prefer conversational responses with
humor and step-by-step solutions. Also
fine. You can specify formatting
preferences,
use bullet points when listing items, or
always site sources when possible, or
avoid technical jargon unless necessary.
Essentially, these instructions become
your personalized rule book, so you
don't have to repeat the same requests
in every single conversation. The
benefits here are huge. First, it saves
time. No more typing, explain like I'm a
beginner or keep it concise every time
you ask a question. You set it once,
Grock remembers, and every answer is
automatically tailored to your
preferences. Second, it improves
consistency. If you're using Grock for a
specific project or role, custom
instructions ensure the responses stay
aligned with your needs.
Third, for professionals, writers,
analysts, content creators, it
guarantees a consistent style.
Imagine a copywriter telling Grock to
always write in AP style and avoid
cliches.
Now, every output is already in the
right format, ready to use. And here's
something for the AI enthusiasts
watching. Custom instructions are also a
playground. You can experiment with how
different instructions affect outputs.
Try telling Grock to be more creative or
stricter with logic or to always
challenge your assumptions.
See how the responses change. It's a
fascinating way to understand how these
models respond to meta instructions and
it helps you refine exactly what you
want from your AI assistant.
When should you use custom instructions?
Honestly, right now when you first start
with Grock, spend 5 minutes setting
these up. You'll thank yourself later.
Update them whenever your needs change.
New project, different focus, refined
style.
If you notice Grock's answers aren't
quite hitting the mark, that's your cue
to revisit and tweak your instructions.
It's a living document that should
evolve with how you use the tool. In the
same menu, you might also find a model
section where you can set defaults like
always starting chats in expert mode or
enabling deep search for certain types
of queries. The models menu might also
show which models you have access to
with premium ones locked if you're not
subscribed.
Bottom line, custom instructions
transform Grock from a generic chatbot
into your customized AI assistant.
It's personalization at its best. And
honestly, it's one of those features
that once you start using it properly,
you'll wonder how you ever managed
without it. Don't skip this.
A few minutes of setup can make every
subsequent interaction smoother, faster,
and more relevant.
Prompting techniques, getting the best
answers. All right, we've covered all
the interface features, the buttons,
modes, settings.
But here's the thing. Even with the most
advanced AI, the quality of your answer
depends heavily on how you ask the
question. That's where prompting
techniques come in.
Think of it like this. Grock is an
incredibly powerful tool, but you're the
one steering it. The better you steer,
the better your destination.
Be clear and specific.
Let's start with the golden rule,
clarity.
A well-phred, specific question gets you
a much better answer than something
vague or generic. Why? Because AI models
respond based on the input you give
them. Vague in, vague out, specific in,
specific out. Here's an example. If you
ask, "Tell me about climate change,"
you're going to get a broad answer that
might not address what you actually care
about. It could cover causes, effects,
politics, science, too much all at once
and probably not in depth on the part
you wanted. But if you ask, discuss the
economic impacts of climate change on
agriculture in Africa over the next 20
years, now you've given Grock a clear
target.
The answer becomes focused, detailed,
relevant.
How do you build specificity into your
prompts? Include details about what
exactly you want. If there are multiple
parts to your question, break them out.
Specify the format if needed in three
bullet points as a step-by-step guide in
simple terms.
Give concrete parameters.
Instead of some examples, say three
examples.
Instead of recent, say since 2020.
The less guesswork Grock has to do, the
better the output. Here's another
comparison.
Vague.
Tell me about neural networks. Could go
anywhere. History, technical details,
applications,
specific. Explain how neural networks
work in simple terms using an analogy
and keep it to one paragraph.
Now, Grock knows you want a basic
analogydriven explanation briefly. Or
consider this.
vague. Write a report on renewable
energy.
Specific. Write a 200word summary of the
benefits and challenges of solar and
wind energy in Europe, comparing their
growth over the last decade.
Clear direction, specific scope, region,
length, all defined. The benefit is you
direct the model's focus and reduce
irrelevant information.
Grock produces more targeted useful
responses and you avoid
misunderstandings.
If Grock doesn't have to guess what you
meant, it's less likely to wander off
track. One guide puts it perfectly. The
granularity of your input is directly
proportional to the utility of the
output. In other words, the more
specific you are, the more useful the
answer becomes.
Don't be afraid to spell out exactly
what you need.
Provide context and constraints.
Next up, context. Grock has been trained
on a massive amount of information, but
it doesn't inherently know what's
relevant to you right now. By providing
context, you frame the question and
guide how Grock approaches it. It's like
talking to a person. If you ask a
question out of nowhere, they might not
know which angle matters to you. But if
you set the scene first, the
conversation flows much better. How do
you add context?
A few techniques. One, set a role or
persona. Tell Grock who it is for this
task. You are a financial adviser with
expertise in retirement planning. Given
this client profile, suggest an
investment strategy.
This frames the response from that
expert perspective.
Two, give background information. If
your question depends on previous data,
include it. Based on the summary I just
provided, what are the key risks? Or
using your last answer about X, now
explain why.
Three, specify your audience or style.
explain this in layman's terms or give a
business executive summary or the
classic explain it like I'm five years
old.
You can also include constraints, limits
or formats for the answer.
List five bullet points. No more than
three sentences. If the information
isn't available, say you don't know
instead of guessing.
Constraints help shape the output so it
fits your needs right away, saving you
from having to edit it afterward.
Here's a real example with context.
Instead of just asking, when is the best
time to visit New England for fall
colors, add this.
You are an experienced wildlife
biologist specializing in trees.
Based on recent weather patterns,
predict the best time this year for peak
fall foliage in New England and explain
it in simple terms. Now you've given
Grock a role, a task, and a style.
The answer becomes more insightful,
probably referencing how weather affects
leaf color. Because you set that
context, the benefit, more accurate,
more relevant answers tailored to your
scenario.
The tone and complexity come out just
right for your audience.
If you say explain to a kindergarten
class versus explain to a panel of PhD
scientists, you'll get vastly different
answers, each appropriate for those
listeners. Constraints ensure the format
fits your requirements. Context and
constraints act as guard rails, steering
gro in the exact direction you want
without it having to second guessess
your intentions. Use follow-ups and
refinement. One of the best things about
chatbased AI is that you don't have to
nail the prompt perfectly on the first
try. You can iterate. Ask something, get
an answer, then refine or dig deeper
with follow-up prompts.
This iterative process can significantly
improve the quality of information you
get. And honestly, it feels more
natural, like having a real
conversation. Here's how it works. Grock
remembers the context of your
conversation within its memory limits.
So, if you ask a question and then ask a
related follow-up, it knows what you're
referring to. For example, you might
say, "Explain how solar panels work."
Grock gives you an explanation, then you
follow up.
Great. Now, how does weather affect
their efficiency? And make it a bit
simpler.
Notice you didn't have to say solar
panels again. Grock knows we're still on
that topic. And you added a new
instruction to simplify, so the answer
adjusts accordingly. Some strategies for
effective follow-ups. Ask for
clarification if something's confusing.
Can you clarify what you meant by X? Or
could you explain that in more detail?
Request a different format. Could you
list those points as bullet points or
put that information into a table? Dig
deeper on a subtopic.
You mentioned quantum efficiency in
solar cells. Could you elaborate on
that? This turns a general answer into a
deep dive exactly where you're curious.
Correct or refocus if the answer missed
the mark. That's not quite what I was
looking for. I meant more about the
policy side. Can you answer from that
angle? Grock will adjust and try again.
You can even use make it better prompts.
Things like make it funnier, simplify
this further or give me an example to
illustrate that.
These piggyback on the previous answer
and tweak it in the way you specify got
a dry explanation.
Make it funnier might add an analogy or
a joke to lighten it up.
The benefit of this approach is you
sculpt the answer to be exactly what you
need piece by piece. It's like having a
dialogue with an expert. You keep asking
until you're satisfied. Often the first
answer is decent, but the second or
third exchange is where Grock really
nails it because you've given more
direction or allowed for course
correction.
And here's the thing, you don't have to
cram everything into one mega prompt,
which can be hard to construct. Start
simple, add detail iteratively.
It feels more natural and often works
better. One tip, if your conversation
gets really long or shifts topics
significantly, you might need to provide
context again or even start a fresh chat
to avoid confusion. But for a focused
thread, Grock keeps the history in mind.
And if Grock produces a really long
answer, use a follow-up to condense it.
Summarize the above in three key points
or explain the last part in simpler
words. You're essentially using Grock to
refine its own output. Think of
follow-ups as an interview. You're the
interviewer. Keep probing until you get
that perfect insight or explanation
you're after. Keep experimenting. And
here's a bonus tip for the AI
enthusiasts out there. Prompting is an
evolving art. Grock is updated
frequently and gains new capabilities.
So, my final piece of advice is stay
curious and keep experimenting.
Try those preset prompt buttons in the
interface, things like brainstorm or
summarize to see how they structure
queries. Learn from them. Adapt their
style to your own needs. If Grock
introduces new features like video
generation or advanced voice modes or
whatever comes next, give them a shot.
Each new tool might open up creative
ways to interact that you hadn't
considered before. And remember, prompt
engineering is part science, part art.
the guidelines we've covered. Be clear,
provide context, iterate, use the right
tools. These set you on the right path,
but you won't always get it perfect on
the first try, and that's fine. Keep
refining, keep adjusting, and use those
features we've talked about over time.
You'll develop an intuition for what
works, and Grock will become an even
more powerful collaborator in your work.
So, there you have it. The complete tour
of Grock's interface and how to use it
like a pro.
We walked through the search bar, your
gateway to every interaction we
explored.
Deep search for those moments when you
need serious sourced research delivered
fast. We played with create image to
turn your ideas into visuals.
We broke down the model selector auto
fast expert Grock 4 fast heavy giving
you control over speed and intelligence
based on your task.
We talked about the Super Grock upgrade,
which unlocks advanced models, think
mode, expanded memory, and a whole suite
of premium features that turn Grock from
good to exceptional. And we covered
custom instructions, your chance to
personalize how Grock responds so every
answer feels tailored to you. Then we
dove into prompting techniques, being
clear and specific with your questions,
providing context and constraints to
guide the AI, using follow-ups to refine
answers iteratively, and leveraging
Gro's modes and tools strategically for
different tasks. These aren't just tips.
They're the difference between getting
generic responses and getting answers
that genuinely solve your problems,
inspire your creativity, or deepen your
understanding. Look, whether you're
brainstorming, learning something new,
solving a complex coding challenge, or
conducting research, Gro's interface
gives you the tools. Your prompting
provides the direction. When you combine
the two skillfully, it's like having a
genius collaborator available 24/7.
Someone who never gets tired, never
judges your questions, and is always
ready to help you push boundaries and
explore ideas. So, go ahead, experiment
with these features.
Click deep search when you've got tough
questions. Generate some images to
visualize your ideas or just for fun.
Tweak those custom instructions until
Grock responds exactly how you want.
Switch between model modes based on
whether you need speed or depth. And
most importantly, ask your questions
with precision and context. Grock is
designed to learn from your input and
iteratively help you reach better
outcomes. Use that conversational flow
to your advantage.
If this walkthrough helped you
understand Grock better, consider
dropping a like or leaving a comment
with what feature you're most excited to
try. And hey, if you've got questions or
want to share your own Grock tips, the
comment section is all yours.
Let's keep the conversation going. With
these tools and techniques in hand,
you're ready to master Grock and unlock
everything this AI assistant has to
offer. Happy Grocking, and I'll see you
in the next video.