AI consulting workflows
Summary
Overview
Natalia Quintero (head of consulting at Every) and Nityesh Agarwal (Applied Data Engineer) presented a 15-minute session showcasing how they have transformed their consulting workflow using Claude Code and custom plugins. The session focused on their "Every PPTX" plugin - a vibe-coded solution that generates branded presentations from raw ideas in under 15 minutes, dramatically reducing the time consultants spend on deck formatting.
Session Context
- Timestamp: 2:00:01 in the stream
- Duration: Approximately 15 minutes
- Format: Live demo and discussion
Key Themes
1. Presentation Generation as Workflow Automation
The core demo showcased a custom Claude Code plugin that generates Every-branded presentations. Rather than using generic AI presentation tools like Gamma, they built a tailored solution that:
- Takes raw ideas and conversation transcripts as input
- Generates blueprint markdown files with slide-by-slide planning
- Creates visual direction documents specifying design choices
- Produces final presentations with speaker notes included
- Integrates Gemini API for on-the-fly image generation
2. From Days to Hours
Traditional consulting workflow for deck creation took days to a week. With their plugin, the content-ready version emerges in hours. The design team still polishes final deliverables, but the labor-intensive first draft phase is now automated.
3. Non-Technical Accessibility
Nityesh demonstrated using Claude Code through the desktop app rather than terminal, emphasizing accessibility for non-technical users. The entire workflow happens through natural conversation without command-line interaction.
4. Style Guide Encoding
A critical success factor was encoding Every's 600-line visual style guide into the plugin. This required significant iteration but enables the system to generate presentations that match the company aesthetic rather than producing generic AI output.
Business Impact
- Time savings: Deck creation reduced from days/week to hours
- Client responsiveness: Faster turnaround enables more client touchpoints
- Scalability: Consulting team can handle more clients without proportional headcount increase
- Design team leverage: Designers focus on polish rather than initial layouts
Technical Architecture
- Input: Raw ideas, client meeting notes, conversation transcripts
- Blueprint phase: Generate markdown with slide structure, content, visual direction
- Visual direction: Detailed per-slide specifications including image prompts
- Generation: Claude Code calls Gemini API for images, assembles PPTX
- Output: Branded presentation with speaker notes
Notable Workflow Pattern
The session demonstrated a "content first, design polish second" approach:
- Human has conversation with client to understand needs
- Human has conversation with Claude about the content structure
- Claude generates blueprint and iterates based on feedback
- Claude generates presentation with AI-created visuals
- Design team reviews and polishes for final delivery
Status and Availability
The plugin was not publicly available at the time of the session. The team indicated they may release a public version in the future based on audience interest (visible in chat reactions).
Connection to Broader Themes
This session exemplified several Vibe Code Camp themes:
- Skills over complexity: A focused skill that does one thing well
- Compound automation: Each presentation builds on learned patterns
- Designer-engineer collaboration: AI handles scaffolding, humans handle taste
- Voice/conversation-driven workflows: Natural language for professional deliverables
Key Concepts
Custom Plugin Forking for Brand Consistency
Taking an existing Claude Code plugin (like the standard PPTX plugin) and customizing it with company-specific styling, workflows, and visual guidelines.
Blueprint-First Presentation Workflow
A multi-stage approach where AI first generates a detailed markdown blueprint (slide structure, content, visual direction, speaker notes) before creating the actual presentation.
AI-Generated Presentation Visuals
Using image generation APIs (like Gemini/Nano Banana) directly within the presentation creation workflow to generate custom illustrations on the fly.
Style Guide as AI Training Document
Encoding an organization's visual design principles, brand guidelines, and aesthetic preferences into a document that AI can reference during generation.
Trivializing Non-Worthwhile Tasks
Using AI automation to reduce time-consuming but low-creativity tasks to near-zero effort, freeing humans for higher-value work.
Notable Quotes
"Nitesh and I have been waking up like three and a half hours early every day for the past week just to like work on this stuff that we're so excited about."
"I am using AI to turn producing non-worth-a-task into trivial ones. So like this deck is the example of that. I just generated this deck to show you what this plugin does."
"I could have just screen shared and showed you the actual code and taken you through the cursor and explored the files, but that would have been boring. So instead we have this nice looking deck."
"If you're like a Claude code nerd, then you probably know that Claude has a PPTX plugin which allows Claude to create presentations and PowerPoints just using the power of code. But if you've used it, you'll know that that creates a very generic-looking presentation."
Tools Mentioned
Transcript
NATALIA QUINTERO & NITYESH AGARWAL (Every Consulting) - AI consulting workflows
=== NATALIA QUINTERO & NITYESH AGARWAL (Every Consulting) - AI consulting workflows ===
(02:00:01): uh,
(02:00:01): some internal,
(02:00:02): every people who are going to show you some of the coolest workflows that I have
(02:00:06): seen recently.
(02:00:07): Um,
(02:00:07): so I'm going to add Natalia and Austin to the stage,
(02:00:11): sorry,
(02:00:11): Nitesh and then Austin,
(02:00:13): Austin's first of all,
(02:00:14): Austin's our producer.
(02:00:15): He's also our head of growth.
(02:00:16): He put together this whole thing.
(02:00:19): Everyone just give him a little bit of a round of applause for how well this is going.
(02:00:22): Like, Austin, you're doing amazing.
(02:00:25): Do you want to be on here or should we just leave it to Natalia and Natasha?
(02:00:31): I'll stay on just in case for your little 30 minute break.
(02:00:34): Maybe go outside, touch some grass, drink some water.
(02:00:36): Sweet.
(02:00:37): Okay.
(02:00:38): I'm off for 30 minutes.
(02:00:40): Everybody enjoy these.
(02:00:42): Natalia and Natasha are going to show you some insane stuff and I'll be back in 30 minutes.
(02:00:47): Thanks, Dan.
(02:00:48): All right.
(02:00:48): Natalia and Natasha, it's your show now.
(02:00:51): I'm stoked for this.
(02:00:52): Why don't you both introduce yourselves and then let's get right into the demo.
(02:00:56): Awesome.
(02:00:57): Hey everyone.
(02:00:58): So we are popping in for about 15 minutes just to share some of the stuff that
(02:01:02): we've been working on over the past week.
(02:01:04): I'm Natalia Quintero.
(02:01:05): I run consulting here at Every and
(02:01:09): So Nitesh and I,
(02:01:11): I'll let Nitesh introduce himself in a second,
(02:01:14): but Nitesh and I have been waking up like three and a half hours early every day
(02:01:19): for the past week just to like work on this stuff that we're so excited about.
(02:01:25): And Nitesh is actually going to share one of the skills that he built and then
(02:01:31): we'll take it from there.
(02:01:32): So Nitesh, over to you.
(02:01:35): Thanks, Natalia.
(02:01:36): Hey, everyone.
(02:01:37): I'm an Applied Data Engineer.
(02:01:39): I help Natalia with everything consulting and automating the day-to-day work.
(02:01:46): So that work is, you would put that in a non-technical bucket.
(02:01:52): So I've been using
(02:01:55): tools like lot code and other wide coding tools to wide code different parts of our
(02:02:02): workflow,
(02:02:04): one of which involves creating presentations.
(02:02:07): So every,
(02:02:09): as you can imagine,
(02:02:10): with all the aesthetic that we've been going on here,
(02:02:12): every has a very distinct aesthetic.
(02:02:16): which translates into all the presentations that we're doing.
(02:02:18): So traditionally,
(02:02:20): we use Figma as our presentation building tool,
(02:02:25): but that takes quite a bit of time.
(02:02:27): So I just write coded a plugin that helps us use Cloud Core to make presentations.
(02:02:35): Let me show you what it does.
(02:02:49): One second.
(02:02:50): Let me just show my screen.
(02:02:52): You're good.
(02:02:58): There we go.
(02:02:59): OK.
(02:03:00): All good.
(02:03:01): Let it rip.
(02:03:02): OK.
(02:03:05): Great.
(02:03:06): OK, so this is a presentation.
(02:03:09): And what I built is called the EveryPPDX plugin.
(02:03:12): So if you're like a Claude code nerd,
(02:03:15): then you probably know that Claude has a PPPDX plugin,
(02:03:19): which allows Claude to create presentations and PowerPoints just using the power of
(02:03:27): code.
(02:03:28): But if you've used it,
(02:03:29): you'll know that that creates a very generic-looking presentation,
(02:03:33): not something that looks like this.
(02:03:36): So I created this, like, forked this plugin.
(02:03:40): I created every version of that plugin so that it can generate presentations that
(02:03:45): look like this straightaway.
(02:03:48): And I created this deck to be the demo.
(02:03:51): So this presentation,
(02:03:53): all the backgrounds,
(02:03:55): the colors,
(02:03:56): the font placements,
(02:03:58): everything,
(02:03:59): including the actual flow of slides and the whole plan of how we're going to
(02:04:05): discuss this was generated by Claude in Claude code.
(02:04:12): And this is huge because what this is basically,
(02:04:19): it means that I am using AI to turn producing not worth a task into trivial ones.
(02:04:24): So like this deck is the example of that.
(02:04:27): Like I just generated this deck to show you what this plugin does.
(02:04:32): I could have just screen shared and showed you the actual code and, you know,
(02:04:38): It took you through the cursor and explored the files, but that would have been boring.
(02:04:43): So instead we have this nice looking deck.
(02:04:47): There's a question in the chat,
(02:04:48): which I have to actually is like,
(02:04:50): are you using this just to make like a nice looking deck from a design perspective?
(02:04:55): Or are you also having kind of like not very clean copy or like kind of
(02:05:00): disorganized thoughts and asking it to organize really good copy for the deck as
(02:05:05): well at the same time?
(02:05:06): That's a great question.
(02:05:08): So let me show you.
(02:05:09): So I am doing everything inside Cloud Core, including the copy.
(02:05:13): And not just the copy, but actually taking me from a raw idea.
(02:05:18): So this is the chat.
(02:05:19): I hope you can still see my screen, right?
(02:05:21): Yeah, you're good.
(02:05:23): OK, so you can see my Claude desktop app.
(02:05:27): And I actually just started a new chat here in code mode where I started discussing
(02:05:34): the plugin that I created.
(02:05:36): And then I gave it this idea that I had, a very rough idea of what I wanted to do.
(02:05:43): And then in the chat, I just built this framework of what we want to build here.
(02:05:51): And then there's a generate blueprint command included in this plugin,
(02:05:56): which actually uses whatever we discussed above.
(02:06:01): to generate a rough markdown file which discusses the entire Blueprint.
(02:06:06): So this is only markdown,
(02:06:08): and this is a slide-by-slide description of what will be in the deck,
(02:06:12): along with the visual direction that we will be taking,
(02:06:17): along with the speaker notes that are going to be here.
(02:06:20): So this deck, if I share the presenter view, you'll be able to see
(02:06:26): speaker notes here.
(02:06:27): So every slide has a speaker notes, which are to be used by the speaker.
(02:06:32): And then like the plugin,
(02:06:35): what it allows me to do is once I have this,
(02:06:38): this is essentially a plan for the presentation.
(02:06:41): So I work on the plan, like give it feedback and make any edits that I need.
(02:06:46): And then there's another command called visual direction.
(02:06:49): Now,
(02:06:51): This is where the magic happens because this is where we use Claude to basically
(02:07:02): give a detailed slide-by-side description of how we will generate each side.
(02:07:08): And this includes a command or a slide for any illustration that we want to add.
(02:07:17): So you see this illustration?
(02:07:20): This was generated by AI while creating this deck.
(02:07:24): And how was Flora able to do that?
(02:07:26): Flora is able to do that because inside Cloud Code,
(02:07:30): I have given it access to my Gemini API keys,
(02:07:36): which it is using to call the Nano Banana API and send it like prompts to generate
(02:07:44): images.
(02:07:45): And those images are generated on the fly to add to this presentation.
(02:07:52): All the visuals that you see here,
(02:07:54): so these things along with any visual that you see here,
(02:08:00): most of them have been generated on the fly with AI.
(02:08:05): And in order to do this,
(02:08:09): what I have to customize here is in the every PBDX present skill,
(02:08:14): I basically gave it every visual style document and told it to use this to
(02:08:22): essentially understand how we think about design.
(02:08:26): Now, getting this 600-line style guide
(02:08:31): the majority of the time getting it right uh it was a lot of back and forth and
(02:08:34): iterations but um but now that uh that work is done we uh it's it's like creating a
(02:08:42): presentation is as easy as just starting a chat inside cloud code um and i'm even
(02:08:48): doing it in the desktop app so there's no um terminal business here uh a lot of
(02:08:54): people want to know if this is a public skill they can use
(02:08:58): Uh, it's not public yet, but, um, I think we can, we, we may be able to do that soon.
(02:09:03): So, um, we'll, uh, we'll follow up with that.
(02:09:05): So people can potentially use this.
(02:09:07): Natalia, I'm curious, like, um, you run the consulting business.
(02:09:11): Uh,
(02:09:11): and I think you've also recently gotten deeply pilled on by coding and cloud code
(02:09:16): in a bigger way.
(02:09:17): Um, how are y'all using this and how is it like, uh, changing how you approach the business?
(02:09:23): Yeah,
(02:09:23): I mean,
(02:09:23): for anyone who's ever had a job where part of their job was making decks,
(02:09:29): you'll be very familiar with just how time consuming it is to format a deck,
(02:09:34): to think about how you lay out the information in a way that is really compelling.
(02:09:39): I think it's why so many of us were so excited about Gamma.
(02:09:42): and still are so excited about Gamma as a product because you have this information
(02:09:46): that you're trying to convey.
(02:09:47): And for us, we're working with tech companies, private equity funds, hedge funds.
(02:09:54): And the way that they digest information really varies quite a bit depending on
(02:09:59): what their strategy is or what their business goals are.
(02:10:02): And instead of us spending so many hours of our day just formatting decks so that
(02:10:06): we can be sharing the information that we want to be sharing,
(02:10:09): we can just have a conversation with our clients understand what their ai sort of
(02:10:14): needs and enablement they need and then we have a conversation with Claude that
(02:10:18): automatically can kind of like lay out this v1 of the decks um we have an
(02:10:23): incredibly talented design team so these version of the slides are not the final
(02:10:28): version that our clients would get but it makes it so that between the time that we
(02:10:33): would usually go
(02:10:34): have those conversations and have it be one of a deck,
(02:10:37): which might be a few days to a week because it's so time intensive.
(02:10:40): Now that time is like a few hours, which is really, really transformative and so impactful.
(02:10:47): So it saves us so much time.
(02:10:49): It makes it so that it works in the way that we hope AI works really well,
(02:10:53): which is that it just makes it so that we have more time to be really responsive to
(02:10:57): the needs
(02:10:58): of our clients.
(02:10:59): And yes, I see the chance of public, public, public.
(02:11:03): We'll see if there's a version of this that we can get out that can be useful to other people.
(02:11:08): But I mean,
(02:11:09): Natasha is just,
(02:11:10): he's so creative in the way that he thinks about the possibilities of how we could
(02:11:14): be using AI.
(02:11:16): And this is just one example, just making our lives easier.
(02:11:20): So are you all at the point yet that you are pretty close to one-shotting this of
(02:11:27): we meet with the client,
(02:11:28): we do our meeting,
(02:11:29): we debrief,
(02:11:31): we have meeting notes,
(02:11:32): we drop it into the plugin,
(02:11:33): and then the thing that comes out is a thing you would email to the client.
(02:11:38): Is there one rev?
(02:11:39): How close is the one-shotting to being completely done here?
(02:11:44): that's a good question i think like we're uh there's a few sort of like pieces here
(02:11:49): right there's like the content piece and i feel like we're pretty close to
(02:11:52): one-shotting like the content that is most relevant to clients which is maybe the
(02:11:57): most valuable part of this um i think the design piece where like you know all of
(02:12:01): this is iterative right like uh when you vibe code something like you think it's
(02:12:05): the final version of course it's not it's just like a thing that you're sort of
(02:12:07): like continuing to improve over time i think the next version of this for us is
(02:12:11): like how do we
(02:12:12): um integrate our workflow with our design team's workflow so that this there's like
(02:12:16): a more beautiful version of this that we can create so that we're not um going back
(02:12:20): to them to like make edits over and over so uh so on the design piece i think
(02:12:26): there's like some more work for us to do but on the content piece i think we're
(02:12:29): like almost there what do you think that's awesome
(02:12:33): Absolutely, yeah.
(02:12:34): This version was generated like the presentation that I showed you.
(02:12:38): I timed it.
(02:12:39): It was generated in under 15 minutes,
(02:12:41): all the way from idea to the version that I just showed you.
(02:12:46): And we are working with our design team to further improve the instructions and the
(02:12:49): design style that we created.
(02:12:51): So it's only going to get better from here.
(02:12:55): My favorite version of this that I've found that works because I'm a terrible
(02:13:00): designer and our designers are very good is that since they work in Figma,
(02:13:04): I'm very Figma make pilled because it can reference those Figma designs directly.
(02:13:08): And I'm very excited to see like as you keep pushing this,
(02:13:11): like how custom you can get different plugins to like take the work our designers
(02:13:16): do.
(02:13:17): and make it so that we don't have to use their bandwidth on just recreating the
(02:13:20): same templates over and over again for the work that like our other teams are
(02:13:24): trying to use.
(02:13:27): Totally.
(02:13:28): Yeah, we're so there with you.
(02:13:29): And I saw a comment from Edgar, some of the best designers I know are coding for us now.
(02:13:34): I mean, I think that's how I think of our design team, right?
(02:13:37): They're almost like engineers with like an incredible curatorial eye.
(02:13:41): And,
(02:13:41): uh,
(02:13:42): I think the challenge for us here and something I'm excited for us to think about
(02:13:44): is like,
(02:13:45): how do we,
(02:13:45): how do we build that vision and that curation,
(02:13:47): uh,
(02:13:48): to the work that we're doing on consulting?