How developers, sales and marketing professionals use Generative Artificial Intelligence in 2025

This is the transcript of our latest live session “Artificial Intelligence in Tech: usage, adoption and challenges in 2025” which you can watch in the following video.

Intro & welcome 

Moschoula

Hi everybody, welcome back to SlashData's webinar series for 2025. For those who aren't familiar with us and are joining for the first time, SlashData is a market research firm active in the technology community for nearly 20 years. We serve the technology community, helping companies make data-backed, high-impact decisions with confidence. We help you understand your customers, your users, and your decision-makers, and understand how to do everything from product design to marketing strategies with data.

We will continue this series throughout the year, so stay tuned and join our newsletter to get invited to the next ones. For housekeeping, before I hand off to our featured speakers, we will be open for questions. The live chat that you should see to the right of your screen is available, and we will be reading through that at the end of the presentation.

We have two senior analysts here with us today: Bleona Bicaj and Alvaro Ruiz. They will address the most topical subject we are all dealing with and learning more about each day—AI and tech usage, adoption, and challenges. Without further ado, I'll hand it over to Alvaro.

Alvaro RuizHello everyone, and welcome again. I'm Alvaro from the research team at SlashData. Today, in this first half of the webinar, we will explore how developers are working with AI and integrating it into their applications.

Here’s a quick overview of what we’ll cover. First, we’ll look at how developers are actually working with ML and AI—whether that’s using AI tools in development workflows, adding AI functionality to applications, or building AI models.

Next, focusing on the second group—those adding AI functionality to applications—we’ll explore the types of models they’re using and do a deep dive on open and open-source models to understand why developers choose to use them and the challenges they face.

Finally, we’ll look at the type of AI functionality developers are adding to their apps—generative versus non-generative—and how the proportion of developers adding GenAI functionality varies based on experience, region, and company size.

Developers using AI tools in their workflows

According to our data from the 27th edition of our Developer Nation survey, fielded in Q3 2024, about two-thirds of developers are already using AI tools in their development workflows. The most common use case is AI chatbots for coding questions, with 46% of developers doing this, followed by 32% using AI-assisted development tools like GitHub Copilot.

Another 21% use AI to generate creative assets for their projects, such as 3D models. When it comes to adding AI functionality directly to applications, 21% of developers are doing so—15% through fully managed AI services or APIs, and 10% using self-managed or local models.

Finally, 15% of developers are involved in creating AI models themselves—customizing with their own data, building and training models, or fine-tuning hyperparameters. That leaves only about a quarter of developers not yet working with AI, highlighting just how integrated AI has become in software development.

For the rest of this presentation, we’ll focus on those adding AI functionality to their applications. In the next presentation, Bleona will share insights on those using AI tools in development workflows.

How developers bring Artificial Intelligence into their applications

Now let’s take a closer look at how developers are bringing AI into their applications. Here, we see the types of AI models developers use. Of the 21% of developers adding AI functionality, 66% indicate they use open or open-source AI models, which equates to around 6.3 million developers.

As these are the most popular types of AI models, we’re going to explore developers’ experiences using them. It’s worth noting that while 58% within this 66% rely exclusively on open and open-source models, a substantial portion—42%—also use in-house or proprietary models.

Use cases of developers adding AI models to their applications

Now, moving on to use cases—modern AI models are opening up a world of possibilities. We asked developers what kinds of AI features they’re building using these models. 

Here’s what we found:

Text generation leads with 37% of developers using open or open-source AI for this. Right behind are conversational interfaces such as chatbots at 36% and text summarisation at 34%. This is no surprise, as natural language processing powers many of today’s most useful AI features—from creating content to streamlining customer support.

But the story doesn’t end there. Developers are also using AI for predictive analytics (30%) and personalisation or recommendation systems (29%). Image generation is equally popular at 29%, reflecting demand for creative, visual AI tools. Many other functionalities also show substantial adoption, highlighting how AI is shaping the next generation of smart applications.

Why developers use open source models

Now let’s explore why developers use open or open-source models. Top reasons include ease of integration, customisation, flexibility, and belief in the open-source model—cited by 34% of developers. This shows that developers want models that fit into their workflows and can be adapted to their needs.

Community support is another major factor, cited by 33%, tied closely to the open-source philosophy—developers can share knowledge, get help quickly, and contribute improvements. No licensing costs (26%) and transparency (25%) are also key. Developers gain visibility into how models work, which is critical for trust, compliance, and addressing ethical concerns.

Other reasons, each cited by fewer than 25%, include algorithm suitability, alignment with organisational values, and avoiding vendor lock-in.

However, using these models comes with challenges. According to our data, 86% of developers using open or open-source models face at least one challenge.

Top among these is security and privacy, cited by 25%. Developers must ensure that AI models don’t compromise user privacy or create vulnerabilities.

Finding the right model is another major issue (23%), especially for those adding conversational interfaces, where this rises to 29%. These use cases generate added complexity, as models may not meet the nuanced needs of conversational AI.

Other challenges include ensuring accuracy (21%), lack of specialised support (19%), and difficulties with fine-tuning or customisation (19%). Many developers also cite limited training resources, knowledge gaps, or compatibility issues (18%).

To complement this, we asked developers why they avoid open or open-source models. The top reasons closely match the challenges discussed earlier. However, 19% say they opt for managed services simply because they’re more convenient. And 25% avoid open or open-source models due to security and privacy concerns.

While this doesn’t mean open-source models are inherently insecure, they may lack the guarantees offered by proprietary solutions.

Now, for the last part of the presentation, let’s see what types of AI functionality developers are adding and profile those using GenAI.

Types of AI functionality developers are adding and who the developers using GenAI are

According to new Q1 2025 data, 25% of developers are now adding AI functionality to applications, up from 21% in Q3 2024. This shows rapid growth.

Breaking this down, 20% of developers are adding generative AI, while 11% are adding non-generative AI for tasks like analysis, prediction, or classification.

Looking at experience levels, developers with less than a year of experience are least likely to build GenAI-powered apps—only about 1 in 10 have done so. This makes sense, as newcomers are often focused on learning the basics.

Developers with 6 to 10 years of experience lead the way at 26%

In contrast, developers with 6 to 10 years of experience lead the way at 26%, followed by those with 3 to 5 years at 23%. These mid-career professionals have built enough expertise to handle complex projects and are often tasked with experimenting with new technologies.

Interestingly, adoption drops among developers with over 10 years of experience—only 17% are adding GenAI features. Many senior developers focus more on oversight, refining workflows, or mentoring.

Regionally, North America leads with 27% of developers integrating GenAI, thanks to its strong tech ecosystem and funding environment. Eastern Europe and South America have the lowest rates, at 11% and 12%, respectively. Contributing factors include weaker infrastructure and economic barriers.

Looking only at professional developers, company size also plays a role. Freelancers and those at very small companies are least likely to integrate GenAI—13% and 16%, respectively, likely due to limited resources.

Mid-sized companies show the highest adoption at 29%, striking the right balance of resources and agility. At large enterprises, adoption drops to 24%, likely due to legacy systems, regulatory concerns, or segmented team responsibilities.

So that’s all for today. We’ll take your questions during the Q&A session at the end of the webinar. And now I’ll hand it over to Bleona to cover how developers—and non-developers—are using AI in their daily work.

The users of Artificial Intelligence in 2025 

Bleona Bicaj

Thank you, Alvaro. I'm Bleona, and I’m also part of the research team here at SlashData. Now that Alvaro has walked us through the builder side of AI-enabled apps, let’s switch to the people who use them.

I’ll open with a snapshot of how developers are working with AI-assisted coding tools. These figures are not from our most recent data set, so think of them as a baseline—we’ll be collecting fresh numbers soon.

According to our data, 32% of developers are already using tools like GitHub Copilot, DeepCode, or Source3.

Looking at experience, those new to software development are least likely to use these tools—only 22% of those with under a year of experience. That’s not surprising, since beginners tend to be cautious about suggestions they can’t yet debug.

But usage rises quickly. By the six-year mark, it reaches 37% as productivity starts to matter more than practice. It levels off and even dips for developers with 16+ years of experience—28%. These veterans may be more selective or focused on tasks like architecture or mentoring, which don’t benefit as much from code generation.

One use case clearly dominates: code generation, reported by 55% of AI tool users. The more seasoned the developer, the stronger the uptake—75% of developers with 16+ years rely on AI to generate code, compared to 37% of those with less than a year.

When asked for their top three reasons to adopt AI tools, 51% mentioned increased productivity. That priority grows with seniority, as senior developers handle larger projects and more responsibility.

Related reasons—like automating repetitive or time-consuming tasks—follow the same pattern, resonating most with experienced professionals.

As I said, this is just a snapshot. We’re collecting new data and will share updates soon.

How decision makers use Generative Artificial Intelligence

Now, shifting from engineers to decision-makers—earlier this year, in January, we interviewed 10 leaders in large tech companies (five in the U.S. and five in Europe), all heading marketing or sales teams. We focused on these functions to explore how GenAI is reshaping non-technical work.

For marketing and sales, GenAI’s promise is clear: it can amplify human effort and streamline operations. Over the past few years, these teams have used GenAI for content creation, customer support, and lead generation. But they’re also learning its limitations.

In our interviews, we asked: 

Why did you introduce GenAI? What tasks does it handle today? What benefits or risks are you seeing? 

Their responses gave us a well-rounded picture of GenAI’s impact.

Interest in GenAI spiked as soon as the wider industry started talking about its potential. Early adopters launched pilot projects 2–3 years ago to streamline workflows, deepen engagement, and extract better insights from data.

As one sales strategy manager put it: 

“GenAI became a topic in our company ever since OpenAI came into existence and the world started talking about it.”

However, other firms moved more recently, potentially encouraged by a new generation of easier and more capable tools. Whatever the timing, we can see a very clear pattern. What began as a small-scale experiment has now shifted to the strategic core of many organisations. And for most of these organisations, Gen AI is no longer just an optional R&D project—it is a priority for staying competitive in this rapidly evolving digital market. Across every interview, three motives came up again and again for using Gen AI: greater efficiency, relief from repetitive work, and sharper decision-making. Sales teams turned to AI for lead qualification, customer segmentation, and personalised outreach—tasks that once took hours but now convert faster with far less manual effort.

How marketing teams use generative AI 

Marketing teams use Gen AI to generate blogs, social posts, and email campaigns at scale, while also keeping tone and quality consistent, which is very important for marketing firms in particular. Most organisations began with small, low-risk pilots, trialing tools like ChatGPT, Gemini, or Copilot before committing at an enterprise level. As one sales enablement manager told us:

“One or two years ago, we started playing around with co-pilots to author materials both internally and externally within a very small group. Based on our input, we decided to implement a pilot throughout the company.”

This start-small-and-then-scale approach was pretty common—launching Gen AI in one team, learning fast, and only then extending it across the organisation.

When we asked where Gen AI shows up in day-to-day work, three main buckets emerged. We have certain use cases for sales, others for marketing, and some that span both.

How sales teams use generative AI 

Sales teams are using Gen AI to zero in on high-potential leads, automate initial outreach, and tailor follow-up messages. It also helps crunch historical data for sharper forecasting and takes care of routine tasks like drafting sales briefs or updating the CRM. That way, sales representatives can spend more time building relationships and closing deals.

Marketing tells a similar story. Gen AI drafts blog posts, social copy, and even edits images in minutes while keeping brand tone intact.

Marketers feed Gen AI campaign data to fine-tune and personalise their messages, and they rely on it for quick-turn visuals such as infographics or short videos. Some tasks, however, cut across both functions.

AI tools record and summarise client calls, draft email replies, and generally serve as an idea sparring partner during planning sessions. By offloading these repetitive jobs, sales and marketing teams alike can redirect their time toward higher-value strategies, creativity, and high-level conversations.

Gen AI is proving to be a genuine workflow changer. Across all of our interviews, sales and marketing leaders highlighted four recurring payoffs: speed, personalisation, cost control, and sharper decisions.

Generative AI: time-saving, personalising and cost-saving

Starting with time saved—AI tools summarise reports, draft sales briefs, and generally clear away low-value admin work in minutes rather than hours. One sales manager put it this way: “Something that used to take two hours now takes 20 minutes.” Yes, there are accuracy issues, but for many tasks, AI dramatically improves efficiency. The result is more calendar space for strategy, creativity, and client conversations.

Then there’s hyper-personalised content. By crunching customer data on the fly, Gen AI tailors ads, emails, and pitch decks to smaller segments—but at volume. A marketing manager said, “With more and more use, AI is starting to learn the tone of our brand and how we communicate to our audience. Now I barely need to tweak anything.” Sales teams see the same benefit: targeted messages that land better and convert faster.

Next, we have cost savings. A major upside of Gen AI is straightforward savings. Marketing leaders told us they are trimming agency fees, especially around media planning and creative production, because AI now builds assets and places ads in real time.

One head of marketing said, “We’ve cut down on agency costs significantly because AI allows us to automate creative production and ad placements in real time.” Sales teams see a similar impact—AI automates lead generation, keeps the CRM updated, and sharpens forecasts, reducing manual effort and freeing budget for higher-value activities.

Better, faster decision-making—Gen AI not only automates but also improves the quality of choices. AI-driven analytics pull insights from live data rather than just last quarter’s reports, so strategy adjusts in real time. Automated transcripts and summary notes capture meetings, customer calls, and performance reviews verbatim.

AI removes corporate amnesia

“AI removes corporate amnesia,” one head of marketing told us. “It records exactly what was said, reducing confusion and ensuring clarity in decision-making.” By reducing human error and preserving a reliable record, Gen AI supports compliance and provides a solid data-driven foundation for next steps.

Adoption is no longer the question—scaling is. Most companies we spoke with have Gen AI running in at least one part of the business. Yet rolling it out to new use cases is proving tricky, and the obstacles after the pilot stage tend to be the same.

Trust sits at the top of that list. Gen AI still hallucinates—producing confident-sounding but incorrect answers. In data-driven roles where accuracy is key, employees hesitate to act on or even vet AI output. A director of sales operations said, “AI generates outputs that sound highly convincing but aren’t always factually correct. The challenge is that someone might trust this information without verifying it.”

Security and privacy are also major concerns. Many firms handle sensitive or proprietary data, and sending that to external AI services raises the risk of leaks, third-party access, and compliance breaches. As a result, some limit AI use to low-risk tasks, while others are building in-house models or imposing strict governance frameworks.

A sales enablement manager noted, “We handle a lot of sensitive data. We can’t afford to upload proprietary information into an external AI system without knowing how the data is secured.”

A third roadblock is know-how. Some employees experiment readily with Gen AI, while others hesitate due to lack of trust or uncertainty about value. Without formal upskilling, adoption stalls—people don’t feel confident using AI in daily work. Companies that invested early in structured training saw faster uptake and a smoother transition.

A marketing expert observed, “There is still a large portion not adopting or unwilling to adopt. It’s more about lack of education and confidence. Training needs to happen across the organization.”

Even with skills in place, standardization is another hurdle. In many firms, one team races ahead with AI while another sticks to manual workflows. The lack of clear company-wide guidelines—when to use AI, how to validate output, who signs off—keeps adoption uneven and dilutes the impact.

An advanced analytics manager said, “Some staff are pushing these tools, while others don’t care. It’s not affecting our roles much yet, but when AI catches up, we’ll need to rethink training.” There’s a clear pattern that more senior or experienced staff are usually more hesitant to adopt AI.

The future of Generative AI for sales and marketing professionals

Looking ahead, most leaders see Gen AI as an augmenter—not a replacement. Its value lies in speeding up routine work, lightening admin loads, and sharpening decisions, while humans supply context, judgment, and ethics. Adoption will expand across business functions, but human oversight will remain essential.

Customer-facing roles, especially in sales, illustrate this well. AI can qualify leads and send automated follow-ups, but complex negotiations and relationship building still need human intuition. In other words, AI handles high-volume, low-value touchpoints—humans own the moments that matter.

A global alliance lead from a sales team said, “I could see a world where maybe smaller deals have an AI rep selling to small clients, but for now, sales reps are still necessary.” According to them, we haven’t reached the point of replacement. Accuracy remains a top priority. Teams are refining models with better training data and validation loops. This “human-in-the-loop” approach builds trust and tamps down hallucinations.

Beyond the tech, companies need to re-engineer workflows and invest in upskilling so people and AI can work side by side. Organizations that invest in training and thoughtful integration will capture the biggest gains. Those that don’t risk stalled adoption and employee pushback.

In short, Gen AI’s future is as a productivity multiplier and strategic ally—if businesses strike the right balance between automation and human strengths like creativity, critical thinking, and relationship management.

A head of marketing said, “AI isn’t going away—it’s only becoming more embedded in how we work. The key is using it responsibly and strategically.” Work alongside AI—not let it replace us.

Key takeaways from professionals in sales and marketing using Generative AI

Now let me wrap up with five key takeaways:

1. From trial to strategy – Two years ago, Gen AI was an experiment. Today, it’s on the board agenda. The shift from pilot to priority happened faster than any tech we’ve tracked.

2. Sales and marketing see the fastest wins – Leads qualified, campaigns drafted—often six times faster. Agency spend drops as creative moves in-house.

3. Trust and security still matter – Every AI output still needs a human eye. Many firms prefer private models to keep data locked down.

4. Skills gap is the choke point – Tech only scales as fast as people’s skills. Without structured training, adoption stalls.

5. The future equals augmentation – Gen AI takes the high-volume, low-judgment tasks, but humans stay accountable for complex decisions.

   
   

These are the headline lessons from the report. Ready for your questions.

Q&A from the audience

Moschoula: 

Thank you so much, Bleona. Thank you, Alvaro, as well.

Here’s a question for Alvaro: Do you think that finding the right model for the job will remain a barrier to AI adoption, or will it decrease over time?

Alvaro Ruiz:

Good question—it could go either way. The open-source AI ecosystem is expanding rapidly. There are now hundreds of models, architectures, and fine-tuned variants, making it hard for developers—especially less experienced ones—to identify the best fit. If growth continues like this, it might get harder. But as the ecosystem matures, we may see more user-friendly platforms and automated model selection tools, making it easier. So, I think the proportion of developers citing this as a barrier will decrease, but it won't vanish, especially for niche domains or junior developers.

Moschoula:

Thank you, Alvaro. Now a question for Bleona. If AI boosts productivity by six times, doesn’t that reduce the need for more staff?

Bleona Bicaj: 

That’s a fair point, and it came up in interviews too. But the six-fold boost mainly applies to mechanical parts of a task—drafting boilerplate code, summarising meetings, first-draft copywriting. Instead of making roles redundant, it frees people to work on backlogs of higher-value tasks—shipping more features, localizing campaigns, strategic conversations. Since we must review AI output for accuracy and bias, the reclaimed time is repurposed, not cut. AI removes busywork, not brain work.

Moschoula:

Yes, I’ve seen that too. One more question for Alvaro: Are developers specialising in just one or two use cases, or are they integrating multiple functionalities?

Alvaro Ruiz:

According to our data, developers are using, on average, 3.8 out of 14 AI functionalities. So yes, most are working with multiple use cases.

Moschoula:

Last question for Bleona—what training formats are delivering the fastest results?

Bleona Bicaj: 

Companies use several formats, but the most effective is a blended approach: short self-paced modules for basics, followed by live workshops with real tasks, then reinforced through monthly micro-sessions and co-worker collaborations. The last part—peer support—proved especially helpful. Companies that relied only on written guidelines or video courses saw slower adoption.

Moschoula:

That's really helpful. Thank you both, Alvaro and Bleona. We look forward to the next session—stay tuned for announcements in our newsletter. And let us know if there are topics you want us to cover.

Bleona Bicaj:

Thank you. Bye.

Alvaro Ruiz:

Thank you. Bye.