State of AI in Business Presentation Creation 2026: Conversation with Akash Choudhry

(2 votes, average: 5.00 out of 5)

Loading...

    Akash Choudhary is a Senior Product Marketing Manager at think-cell, the presentation software trusted by over a million professionals worldwide. At think-cell, he bridges product development, customer insights, and go-to-market strategy, leads product positioning, messaging, launches, and thought leadership initiatives, with a particular focus on AI-powered presentation workflows.

In this conversation, Akash discusses findings from think-cell’s State of AI in Business Presentation Creation 2026 survey.

Geetesh: Akash, what made think-cell want to run a survey on this scale, 1,200 professionals across 12 countries, and what were you hoping to learn about how people really use AI when building presentations that nobody had properly answered before?

Akash: We have spent over two decades sitting inside the presentation workflows of consultants, bankers, and analysts, and primarily everyone who creates high-stakes business presentations. So, we had a strong intuition about where the pain points were. But intuition isn’t data, and the AI conversation right now is full of anecdotes and unrepresentative voices from either the “AI will replace all of this” camp or the “AI is useless for real work” camp. Neither extreme matched what we were actually seeing from PowerPoint users. So, we wanted to go directly to 1,200+ professionals across 12 countries and ask them, plainly, how they are actually using AI to create or iterate on presentations today.

What we were really hoping to uncover was the gap between adoption and value. An 80% adoption number alone tells you almost nothing useful. We wanted to know which specific tasks people trust AI with, which tasks they have tried and abandoned, and why. That’s the kind of granularity that actually helps us build better tools. Presentations (and PowerPoint) are a genuinely unique environment for LLMs, and we didn’t want to assume that broader AI trends would just carry over.

Geetesh: People seem happy letting AI help with writing, but far less impressed with AI-built charts or full decks. Why do you think that gap exists, and does it match what your team has seen building AI into a chart-focused tool?

Akash: One of the clearest findings from the survey was that professionals have already embraced AI as a thinking partner. They are using it to review content or brainstorm ideas. All the tasks where large language models naturally excel.

A chart, however, is a different kind of object entirely. Its an important call about what to emphasize, what to leave out, how to guide someone’s eye to the one number that actually matters for the decision being made. There is also a trust dimension here that is very specific to charts inside decks. When AI gets a chart wrong, imagine in a boardroom or client setting, the credibility cost is much higher, so people are more skeptical, and I believe, rightly so. What we found interesting is that this is not really an “AI is not good enough yet” problem, it’s an “AI is not integrated into the right layer of the workflow” problem. Most AI tools are trying to generate a finished slide from scratch, when what professionals actually need is AI operating on real data, so the human is refining a high-quality baseline instead of fixing a hallucinated draft.

Geetesh: With three out of four people expecting to use even more AI for presentations over the next year, where do you see the biggest opportunity to close that satisfaction gap? And more broadly, how might these findings influence the direction of tools like think-cell Assist?

Akash: The 77% figure telling us people expect to increase their AI usage next year is probably one of the most important number in the whole report, because it means the appetite isn’t the problem. The biggest opportunity is exactly where the survey data points towards: quality and relevance of outputs, and stronger data security guarantees, which are cited far more often than cost as the thing that would drive more adoption. That points you to something significant that people are not price-sensitive here, rather trust-sensitive.

That directly shapes how we think about think-cell Assist. Our approach has never been “let an AI chatbot generate a whole deck and hope for the best.” It’s giving professionals AI support exactly where the friction actually lives, and making sure every AI-assisted chart still meets the boardroom-ready bar our charting engine has always been built around.

That is really the whole philosophy behind think-cell Assist: make your best slides more effective with AI. Concretely, that means building boardroom-ready slides, with automated chart generation from verified data sources or your own Excel files. It means using AI where you already work, inside PowerPoint and Excel, seamlessly integrated into your slide workflow, rather than forcing you into a separate tool and a copy-paste routine. That deep integration matters beyond just convenience: because Assist has purpose-built automations for the most common presentation tasks, the LLM only gets called in where it’s actually needed. A lot of the work happens through dedicated, on-machine automations that handle an entire step end-to-end, rather than routing everything through an external model. That means fewer external calls, lower token usage, and faster execution, which is a meaningfully different architecture than tools that treat every single action as another prompt. And above all, it means you own your outputs: charts you can annotate, text you can approve, slides that reflect your human expertise.

More info can be found here.

The views and opinions expressed in this blog post or content are those of the authors or the interviewees and do not necessarily reflect the official policy or position of any other agency, organization, employer, or company.