1
00:00:00,000 --> 00:00:04,160
Today's topic is one almost everyone has heard of but few actually understand,

2
00:00:04,160 --> 00:00:06,280
Power BI Copilot.

3
00:00:06,280 --> 00:00:10,080
Microsoft markets it as this AI assistant that writes reports for you and on the surface,

4
00:00:10,080 --> 00:00:13,360
that's true, but the truth is more useful and a little dangerous.

5
00:00:13,360 --> 00:00:16,240
By the end of this episode, you'll know what Copilot actually does,

6
00:00:16,240 --> 00:00:19,000
where it falls short and how to use it without getting burned.

7
00:00:19,000 --> 00:00:23,600
Here's the thing, most beginners try it once, get a wrong answer, and never come back.

8
00:00:23,600 --> 00:00:25,880
You're going to avoid that trap.

9
00:00:25,880 --> 00:00:27,880
What exactly is Power BI Copilot?

10
00:00:27,880 --> 00:00:29,680
Let's start with the simplest definition.

11
00:00:29,680 --> 00:00:33,000
Copilot is Microsoft's AI assistant built right into Power BI.

12
00:00:33,000 --> 00:00:37,040
Think of it like having a coworker who knows DAX and report design sitting next to you.

13
00:00:37,040 --> 00:00:39,600
You tell them what you need and they help you get it done.

14
00:00:39,600 --> 00:00:42,760
Now Copilot uses the same tech as ChatGPT but here's the big difference.

15
00:00:42,760 --> 00:00:47,520
It's trained specifically on Power BI features and it's connected to your actual data model.

16
00:00:47,520 --> 00:00:51,200
So when you ask it something, it's not guessing based on generic internet knowledge.

17
00:00:51,200 --> 00:00:54,800
It's looking at your tables, your columns, your measures and your relationships.

18
00:00:54,800 --> 00:00:56,920
You'll find Copilot in two main places.

19
00:00:56,920 --> 00:01:01,360
In Power BI Desktop, there's a Copilot button in the ribbon that opens a Chat pane.

20
00:01:01,360 --> 00:01:05,360
In Power BI service, the web version, there's a similar Chat pane for asking questions about

21
00:01:05,360 --> 00:01:06,760
your published reports.

22
00:01:06,760 --> 00:01:08,160
But here's the key thing to know.

23
00:01:08,160 --> 00:01:10,640
Copilot can see your tables, columns, measures and relationships.

24
00:01:10,640 --> 00:01:13,040
It works with your data, not some generic data set.

25
00:01:13,040 --> 00:01:16,200
That's what makes it powerful and it's also what makes it tricky if your data model isn't

26
00:01:16,200 --> 00:01:17,200
well-organized.

27
00:01:17,200 --> 00:01:19,280
So what can you actually ask this thing to do?

28
00:01:19,280 --> 00:01:21,800
Let's break down the three main things it's good at.

29
00:01:21,800 --> 00:01:23,320
Copilot's three superpowers.

30
00:01:23,320 --> 00:01:25,800
Let's break down Copilot's three superpowers.

31
00:01:25,800 --> 00:01:28,040
The first one is probably the most popular.

32
00:01:28,040 --> 00:01:30,240
Writing DAX measures from plain English.

33
00:01:30,240 --> 00:01:34,040
You type something like create a measure for year-over-year sales growth and Copilot writes

34
00:01:34,040 --> 00:01:35,360
the formula for you.

35
00:01:35,360 --> 00:01:39,360
It generates the DAX, explains what each part does and gives you the steps to add it to your

36
00:01:39,360 --> 00:01:40,360
model.

37
00:01:40,360 --> 00:01:44,400
Prompts like, show total sales for last year, produce working time intelligence measures

38
00:01:44,400 --> 00:01:45,400
in seconds.

39
00:01:45,400 --> 00:01:46,400
Here's the thing.

40
00:01:46,400 --> 00:01:48,920
Copilot doesn't add measures directly to your model.

41
00:01:48,920 --> 00:01:52,440
It gives you the code and instructions to add it yourself, so you still need to copy

42
00:01:52,440 --> 00:01:53,440
and paste.

43
00:01:53,440 --> 00:01:58,160
This small price to pay for not having to remember the exact syntax for same period last year.

44
00:01:58,160 --> 00:02:01,040
Next up is generating entire report pages from a description.

45
00:02:01,040 --> 00:02:03,480
This is where Copilot really shines for report authors.

46
00:02:03,480 --> 00:02:04,680
You describe what you want.

47
00:02:04,680 --> 00:02:09,120
A sales performance overview with KPIs and a bar chart by region and Copilot builds

48
00:02:09,120 --> 00:02:10,120
it.

49
00:02:10,120 --> 00:02:13,880
It picks visual types, lays them out on the page, applies your report theme and even adds

50
00:02:13,880 --> 00:02:14,880
slices.

51
00:02:14,880 --> 00:02:17,720
Users find this is the feature that saves the most time.

52
00:02:17,720 --> 00:02:20,840
Instead of spending an hour dragging and dropping visuals, you can describe what you

53
00:02:20,840 --> 00:02:23,080
need and get a solid first draft in seconds.

54
00:02:23,080 --> 00:02:27,240
And the third superpower is for business users who don't want to dig through reports, answering

55
00:02:27,240 --> 00:02:29,680
questions about your data in plain English.

56
00:02:29,680 --> 00:02:33,440
Someone can ask what were our top products last quarter and get an answer with supporting

57
00:02:33,440 --> 00:02:34,440
visuals.

58
00:02:34,440 --> 00:02:36,760
Copilot can also generate narrative summaries.

59
00:02:36,760 --> 00:02:39,720
A paragraph that explains what the data on a page actually means.

60
00:02:39,720 --> 00:02:43,440
This is perfect for executive briefings or saving time writing report commentary.

61
00:02:43,440 --> 00:02:47,240
Instead of staring at a chart and trying to put it into words, you ask Copilot to summarize

62
00:02:47,240 --> 00:02:49,400
it and you've got your first draft.

63
00:02:49,400 --> 00:02:52,000
The big gotcha, confident, wrong answers.

64
00:02:52,000 --> 00:02:53,840
And all sounds amazing and it is.

65
00:02:53,840 --> 00:02:56,200
But here's the gotcha people don't talk about enough.

66
00:02:56,200 --> 00:02:59,720
Copilot can sound absolutely certain and be completely wrong at the same time.

67
00:02:59,720 --> 00:03:02,840
And because it sounds so confident, you might not think to double check it.

68
00:03:02,840 --> 00:03:04,120
Let me give you a real example.

69
00:03:04,120 --> 00:03:06,840
Someone asked Copilot what were total orders last month.

70
00:03:06,840 --> 00:03:10,640
And Copilot confidently returned 14,941.

71
00:03:10,640 --> 00:03:11,720
Sounds like a real number, right?

72
00:03:11,720 --> 00:03:13,920
The actual number for February was around 2000.

73
00:03:13,920 --> 00:03:15,720
So Copilot was off by about 13,000.

74
00:03:15,720 --> 00:03:19,120
When the person clicked through to the source, they found Copilot had pulled the all-time

75
00:03:19,120 --> 00:03:23,440
total from a headline on a completely different page, not February's figure at all.

76
00:03:23,440 --> 00:03:27,760
If you'd put that number in an email or a board presentation, you'd have a real problem.

77
00:03:27,760 --> 00:03:28,760
So why does this happen?

78
00:03:28,760 --> 00:03:29,920
It's not that Copilot is broken.

79
00:03:29,920 --> 00:03:32,640
It's that Copilot doesn't understand your data the way a human does.

80
00:03:32,640 --> 00:03:36,040
It's matching words in your question to words it finds in your report.

81
00:03:36,040 --> 00:03:38,760
Column names, page titles, chart labels.

82
00:03:38,760 --> 00:03:43,720
When you say total orders but your report labels the chart order volume, Copilot has to guess.

83
00:03:43,720 --> 00:03:44,960
Sometimes it guess is wrong.

84
00:03:44,960 --> 00:03:46,120
There's another issue too.

85
00:03:46,120 --> 00:03:48,680
The same question asked twice can give you different answers.

86
00:03:48,680 --> 00:03:51,840
Copilot is what they call non-deterministic.

87
00:03:51,840 --> 00:03:54,720
The output varies even with identical input.

88
00:03:54,720 --> 00:03:57,400
One session you ask, which manager is struggling?

89
00:03:57,400 --> 00:04:00,040
And you get a useful breakdown with names and numbers.

90
00:04:00,040 --> 00:04:01,440
Next session, same question.

91
00:04:01,440 --> 00:04:03,280
And Copilot asks for clarification.

92
00:04:03,280 --> 00:04:04,760
What do you mean by struggling?

93
00:04:04,760 --> 00:04:06,800
Third session, it gets blocked entirely.

94
00:04:06,800 --> 00:04:08,400
Same question, three different outcomes.

95
00:04:08,400 --> 00:04:09,400
That's a confusing pattern.

96
00:04:09,400 --> 00:04:11,480
For something like dinner ideas, that's fine.

97
00:04:11,480 --> 00:04:14,640
But when you're dealing with revenue figures or SLA numbers, you're presenting to your

98
00:04:14,640 --> 00:04:17,480
board, you need the same answer every single time.

99
00:04:17,480 --> 00:04:19,880
So how do you get consistent reliable answers?

100
00:04:19,880 --> 00:04:22,920
The key is in how you phrase your questions.

101
00:04:22,920 --> 00:04:24,680
How to ask Copilot the right way?

102
00:04:24,680 --> 00:04:26,080
Here's the key insight.

103
00:04:26,080 --> 00:04:28,120
Use the exact words that are already in your report.

104
00:04:28,120 --> 00:04:32,120
If your report calls something order volume and rag status, then those are the terms you

105
00:04:32,120 --> 00:04:33,440
need to type in.

106
00:04:33,440 --> 00:04:35,120
Don't rephrase into everyday language.

107
00:04:35,120 --> 00:04:37,360
Copilot matches vocabulary, not meaning.

108
00:04:37,360 --> 00:04:41,160
So a specific question using the report's own language gets you a consistent correct answer

109
00:04:41,160 --> 00:04:42,160
every time.

110
00:04:42,160 --> 00:04:43,440
Let me show you what that looks like.

111
00:04:43,440 --> 00:04:46,440
A vague question like, what were total orders last month?

112
00:04:46,440 --> 00:04:48,520
That's you that wrong answer we saw earlier.

113
00:04:48,520 --> 00:04:53,880
But a specific question using the reports own words, what was order volume for March 2026,

114
00:04:53,880 --> 00:04:56,480
gets you the correct answer you can actually verify.

115
00:04:56,480 --> 00:04:59,520
Same intent, different wording, completely different result.

116
00:04:59,520 --> 00:05:02,760
Another tip, be specific about visual types and data fields.

117
00:05:02,760 --> 00:05:07,720
Instead of saying show me sales, try create a column chart showing monthly sales for 2024

118
00:05:07,720 --> 00:05:08,920
with data labels.

119
00:05:08,920 --> 00:05:12,280
The more context you give, the less room Copilot has to guess what you want.

120
00:05:12,280 --> 00:05:15,240
You're basically narrowing the field of possible answers.

121
00:05:15,240 --> 00:05:17,240
And here's something a lot of people miss.

122
00:05:17,240 --> 00:05:18,600
Copilot supports conversation.

123
00:05:18,600 --> 00:05:21,960
If the first result isn't quite right, ask it to adjust.

124
00:05:21,960 --> 00:05:25,800
Change the time period to quarterly or sort from highest to lowest.

125
00:05:25,800 --> 00:05:28,400
Treat it like a colleague who needs clear instructions.

126
00:05:28,400 --> 00:05:30,760
That back and forth is where you get the best results.

127
00:05:30,760 --> 00:05:35,000
But there are also things Copilot simply cannot do, no matter how well you ask.

128
00:05:35,000 --> 00:05:36,600
What Copilot cannot do.

129
00:05:36,600 --> 00:05:41,080
Let's talk about the limits, because knowing what Copilot can't do saves you a lot of frustration.

130
00:05:41,080 --> 00:05:42,840
First, it cannot forecast future numbers.

131
00:05:42,840 --> 00:05:46,080
If you ask how many orders will we get next month, Copilot will tell you it can't predict

132
00:05:46,080 --> 00:05:47,080
the future.

133
00:05:47,080 --> 00:05:49,600
It only works with data that already exists in your model.

134
00:05:49,600 --> 00:05:53,360
It can show historical patterns as a rough guide, but no predictions.

135
00:05:53,360 --> 00:05:57,680
So if you need a forecast, you're still looking at Power BI's native forecasting features

136
00:05:57,680 --> 00:05:59,080
or a separate tool.

137
00:05:59,080 --> 00:06:02,520
Second, it cannot tell you why something happened.

138
00:06:02,520 --> 00:06:04,320
Why did delivery spike in August?

139
00:06:04,320 --> 00:06:05,480
Copilot can't answer that.

140
00:06:05,480 --> 00:06:09,600
It can show you the numbers, but causal explanations require human business context.

141
00:06:09,600 --> 00:06:10,600
Maybe there was a promotion.

142
00:06:10,600 --> 00:06:11,600
Maybe a supplier changed.

143
00:06:11,600 --> 00:06:12,800
Copilot doesn't know that.

144
00:06:12,800 --> 00:06:16,360
It can show you the data, but the story behind it is still your job.

145
00:06:16,360 --> 00:06:18,200
Third, it cannot fix bad data.

146
00:06:18,200 --> 00:06:23,400
If your source data has errors, missing values or inconsistent naming, Copilot can't magically

147
00:06:23,400 --> 00:06:24,400
clean it.

148
00:06:24,400 --> 00:06:25,400
Here's the thing.

149
00:06:25,400 --> 00:06:27,440
Copilot is only as good as your semantic model.

150
00:06:27,440 --> 00:06:29,320
Garbage in, garbage out, same as any tool.

151
00:06:29,320 --> 00:06:33,480
If your data is a mess, Copilot will give you confident wrong answers based on that mess.

152
00:06:33,480 --> 00:06:37,240
And fourth, it cannot make major formatting or layout changes.

153
00:06:37,240 --> 00:06:41,680
It's difficult to change the bar color to blue and it will give you manual instructions instead

154
00:06:41,680 --> 00:06:42,680
of doing it.

155
00:06:42,680 --> 00:06:46,480
Visual customization still requires you to use Power BI's formatting panels.

156
00:06:46,480 --> 00:06:50,440
So for the polished presentation ready stuff, you're still doing the work.

157
00:06:50,440 --> 00:06:53,480
So if Copilot has all these limitations, is it still worth using?

158
00:06:53,480 --> 00:06:54,480
Absolutely.

159
00:06:54,480 --> 00:06:56,840
If you set it up for success.

160
00:06:56,840 --> 00:06:58,600
What makes Copilot work well?

161
00:06:58,600 --> 00:07:00,040
Your data model matters most.

162
00:07:00,040 --> 00:07:03,600
Clean table names, clear columns and proper relationships are what Copilot reads.

163
00:07:03,600 --> 00:07:06,040
If your model is messy, Copilot will be messy.

164
00:07:06,040 --> 00:07:07,040
There's no way around it.

165
00:07:07,040 --> 00:07:08,520
What does that mean for you?

166
00:07:08,520 --> 00:07:09,840
Use names people actually say.

167
00:07:09,840 --> 00:07:12,520
A field called Revenue is useless to Copilot.

168
00:07:12,520 --> 00:07:14,000
Rename it to net revenue.

169
00:07:14,000 --> 00:07:16,840
Now Copilot understands when someone asks about revenue.

170
00:07:16,840 --> 00:07:18,920
This helps every human who uses the report too.

171
00:07:18,920 --> 00:07:21,160
It's one of those fixes that pays off everywhere.

172
00:07:21,160 --> 00:07:24,000
There's a feature in Power BI called Prep Data for AI.

173
00:07:24,000 --> 00:07:26,600
It lets you simplify the schema and add instructions.

174
00:07:26,600 --> 00:07:29,280
You tell Copilot which tables and measures matter.

175
00:07:29,280 --> 00:07:32,640
You add context like Geography is used for borrow analysis.

176
00:07:32,640 --> 00:07:37,720
That dramatically improves answer quality because you're teaching Copilot how your data works.

177
00:07:37,720 --> 00:07:41,760
A small investment of time makes everything Copilot does more reliable.

178
00:07:41,760 --> 00:07:43,720
Here's the most practical advice I can give.

179
00:07:43,720 --> 00:07:45,240
Start small and validate everything.

180
00:07:45,240 --> 00:07:49,040
Don't ask Copilot to build your entire executive dashboard on the first try.

181
00:07:49,040 --> 00:07:51,080
Start with one measure or one visual.

182
00:07:51,080 --> 00:07:52,080
Verify it.

183
00:07:52,080 --> 00:07:53,080
Then expand.

184
00:07:53,080 --> 00:07:56,240
Users who treat Copilot as a drafting assistant rather than a finished product have

185
00:07:56,240 --> 00:07:57,520
much better results.

186
00:07:57,520 --> 00:08:00,240
They don't get burned because they're not trusting it blindly.

187
00:08:00,240 --> 00:08:02,080
This is all stuff you can do today.

188
00:08:02,080 --> 00:08:06,640
There's a bigger question here about what Copilot means for you as someone learning Power BI.

189
00:08:06,640 --> 00:08:10,040
What Copilot means for beginners and citizen developers.

190
00:08:10,040 --> 00:08:13,360
So what does all this mean for you as someone learning Power BI?

191
00:08:13,360 --> 00:08:16,400
The short answer is that Copilot lowers the barrier to entry.

192
00:08:16,400 --> 00:08:17,400
And that's a good thing.

193
00:08:17,400 --> 00:08:21,080
You don't need to memorize DAX syntax to start building useful reports anymore.

194
00:08:21,080 --> 00:08:24,400
You can describe what you want in plain English and see working examples right away.

195
00:08:24,400 --> 00:08:26,960
Think of Copilot as both an assistant and a teacher.

196
00:08:26,960 --> 00:08:27,960
It's both at once.

197
00:08:27,960 --> 00:08:29,200
It teaches you as you go.

198
00:08:29,200 --> 00:08:32,640
When Copilot generates a DAX formula, it also explains what each part does.

199
00:08:32,640 --> 00:08:37,160
You see the code, you read the explanation, and over time you start recognizing patterns.

200
00:08:37,160 --> 00:08:40,600
You learn what calculate does, how filter works, and how time intelligence functions like

201
00:08:40,600 --> 00:08:42,800
same period last year fit together.

202
00:08:42,800 --> 00:08:46,200
You're learning by doing which is the most effective way to pick up a new skill.

203
00:08:46,200 --> 00:08:50,560
Instead of reading a textbook on DAX, you're writing real measures and seeing how they behave.

204
00:08:50,560 --> 00:08:52,080
But here's the part that matters.

205
00:08:52,080 --> 00:08:54,720
Copilot doesn't replace the need to understand your data.

206
00:08:54,720 --> 00:08:57,920
It can write the formula, but you still need to know if the result is reasonable.

207
00:08:57,920 --> 00:09:01,520
It can build the chart, but you need to know if it tells the right story.

208
00:09:01,520 --> 00:09:02,840
Validation is still your job.

209
00:09:02,840 --> 00:09:05,240
Copilot is a tool, not a replacement for thinking.

210
00:09:05,240 --> 00:09:06,760
The risk is over reliance.

211
00:09:06,760 --> 00:09:09,720
Beginners who trust Copilot completely are the ones who get burned.

212
00:09:09,720 --> 00:09:12,880
That confident wrong answer problem we talked about earlier hits hardest when you don't

213
00:09:12,880 --> 00:09:13,880
know enough to spot it.

214
00:09:13,880 --> 00:09:17,880
If you've never seen a revenue report before, you might not realize that 14,1441 is way

215
00:09:17,880 --> 00:09:19,160
too high for one month.

216
00:09:19,160 --> 00:09:23,000
The right approach is to use Copilot as an accelerator, not a crutch.

217
00:09:23,000 --> 00:09:26,280
Let it speed you up, but don't let it think for you.

218
00:09:26,280 --> 00:09:27,280
Practical framework.

219
00:09:27,280 --> 00:09:28,920
How to use Copilot today.

220
00:09:28,920 --> 00:09:32,320
So how do you actually use Copilot in Power BI without getting burned?

221
00:09:32,320 --> 00:09:34,680
Think of it as a brilliant but forgetful junior analyst.

222
00:09:34,680 --> 00:09:38,040
It can draft reports in seconds, but you still need to supervise the work.

223
00:09:38,040 --> 00:09:40,080
Here are four principles to keep in mind.

224
00:09:40,080 --> 00:09:41,840
First, always verify the numbers.

225
00:09:41,840 --> 00:09:45,040
Copilot typically gives you a click through link to the source visual it used.

226
00:09:45,040 --> 00:09:48,360
Click that link, confirm the number matches, every single time.

227
00:09:48,360 --> 00:09:50,520
If it can't show you a source, be skeptical.

228
00:09:50,520 --> 00:09:52,840
That's your first warning sign that something might be wrong.

229
00:09:52,840 --> 00:09:55,240
Next, use Copilot for what it does best.

230
00:09:55,240 --> 00:09:59,160
Using DAX measures, generating starter report pages, creating narrative summaries, those

231
00:09:59,160 --> 00:10:00,560
are its strengths.

232
00:10:00,560 --> 00:10:04,360
Complex calculations, causal analysis or anything that requires business judgment, keep

233
00:10:04,360 --> 00:10:05,600
those for yourself.

234
00:10:05,600 --> 00:10:07,040
That territory is still yours.

235
00:10:07,040 --> 00:10:10,640
Also, invest in your data model before you ask Copilot anything.

236
00:10:10,640 --> 00:10:14,200
Clean up table and column names, set clear relationships between tables.

237
00:10:14,200 --> 00:10:17,240
Use the prep data for AI feature to add instructions.

238
00:10:17,240 --> 00:10:20,680
Every hour you spend here pays back 10 in Copilot accuracy.

239
00:10:20,680 --> 00:10:24,920
A well structured model is the single biggest factor in getting reliable answers.

240
00:10:24,920 --> 00:10:27,960
Finally, treat everything Copilot gives you as a rough draft.

241
00:10:27,960 --> 00:10:29,960
It's a starting point, not a finished product.

242
00:10:29,960 --> 00:10:34,040
Refine it manually, test it with different filters, and make sure it behaves correctly in

243
00:10:34,040 --> 00:10:35,920
all the scenarios you care about.

244
00:10:35,920 --> 00:10:37,400
Then promote it to production.

245
00:10:37,400 --> 00:10:40,680
That one habit will save you from the most embarrassing mistakes.

246
00:10:40,680 --> 00:10:42,440
So here's what you need to take away.

247
00:10:42,440 --> 00:10:46,640
Copilot is a powerful drafting assistant for your DAX, reports and data questions.

248
00:10:46,640 --> 00:10:47,880
But it comes with a catch.

249
00:10:47,880 --> 00:10:51,160
It sounds confident even when it's wrong, and it needs a clean data model and specific

250
00:10:51,160 --> 00:10:52,680
prompts to work reliably.

251
00:10:52,680 --> 00:10:53,920
Your homework is simple.

252
00:10:53,920 --> 00:10:58,560
In Power BI, clean up your main reports column names, and ask Copilot to write one measure

253
00:10:58,560 --> 00:10:59,560
you've been avoiding.

254
00:10:59,560 --> 00:11:01,360
Most people will just watch this and move on.

255
00:11:01,360 --> 00:11:04,000
You have a chance to actually try it and see the difference.

256
00:11:04,000 --> 00:11:07,360
If this helped, subscribe and share it with someone starting their Power BI journey.