Productivity software firms pivot to genAI by leaning on legacy strengths

A recent survey by MIT found that 95% of generative AI (genAI) projects failed to have a meaningful financial impact. But for companies making productivity tools, adopting the technology has become a matter of survival.

Everyone from big-name companies like Microsoft and Google to others like Zoom, Slack and Twilio, shifted gears (and business practices) in recent years to make seemingly overnight changes to their wares, plugging in AI and adding agents atop well-laid workflows and legacy business models.

Now, those genAI tools are widely used to make their products easier to use, with AI helpers providing new insights by analyzing old data that until now lay siloed and dormant.

Microsoft and Google are respectively adding Copilot and Gemini features to their productivity suites at a quickening pace. It’s a matter of survival that “will depend on how quickly companies adopt AI tools as a new way of doing things, and how quickly their customers adopt the new services,” said Jack Gold, principal analyst at J. Gold Associates.

Other companies have taken note: OpenAI, which spurred the genAI revolution with the release of ChatGPT in 2022, is planning its own productivity offering. And Perplexity is rethinking browsers, envisioning browsing apps that include e-mail composition and other productivity features.

“OpenAI is increasingly seeing itself as a productivity tool, and that would include the need to address actual creation tools like Office tools does,” said Gold.

Why the rush to add genAI seemingly everywhere? Because doing so adds a level of intelligence that can accelerate enterprise productivity not possible at a human scale, corporate execs said.

“Your strategy is delivered by and enhanced through AI,” said Zachary Hanif, head of AI, machine learning, and data at Twilio. “AI itself is not the fundamental strategy.”

Twilio, a cloud-based customer engagement company founded in 2008, has a history of tearing down products to evolve its offerings. It started off focused on SMS and email communication but is now looking at multimodal forms of communication with AI.

“The benefit of those transitions is that you develop a lot of muscle memory for how to be able to change and adapt as the world changes and adapts around you,” Hanif said. He characterized genAI technology as delivering added context and speed — the “right message, right time, right context.” 

Taking a different tack, Slack, which is part of Salesforce, converted its popular communications and messaging platform into what Chief Product officer Rob Seaman called a “work operating system.”

The company looked for problems AI could solve within the Slack interface and is acting on what it found, Seaman said. For example, digital labor tools inside Slack can orchestrate communication and processes and newer features offer enterprise search, summaries, meeting notes, and translations.

“Can they be materially better because they’re in Slack? If not, we’re just wasting time trying to recreate or reinvent the wheel,” Seaman said.

At Zoom, AI features are embedded in the company’s videoconferencing software, not bolted on; the goal is to deliver real value to users, said Kim Storin, chief marketing officer at Zoom.

“It’s truly part of the overall workflow, versus just a side project of AI,” Storin said, adding, “What we’re seeing a lot in the market right now is AI washing.”

Like Slack, Zoom is adding AI agents to automate online and offline tasks. (Both companies help customers discover agents through search, directories and galleries.)

Even with the push to infuse AI in software, tech vendors and buyers alike face universal challenges, analysts said.

enAI

While AI is indeed changing business, software vendors are rushing out new features without real insights from their buyers, said Craig LeClair, vice president and principal analyst at Forrester Research. “It’s kind of like the traditional process methodologies are going to be blown up by this. And there’s nothing that’s jumping out there…to replace it,” LeClair said.

Agents are being designed as co-workers to humans, but measuring their productivity remains an open question. And some workers might resist AI if cultural changes aren’t put in place.

“It’s going to go from human in the loop to human on the loop,” LeClair said. “There needs to be monitoring to attribute accountability to people to manage these agents.”

Trusting newer frameworks and models to interact with older systems is also a concern, LeClair said. “They’re worried about explainability issues, they’re worried about security and data leakage and prompt injections,” he said.

Relying completely on AI and computers to make critical decisions could hurt companies if guardrails aren’t in place, said Bob Parker, senior vice president at IDC.

“You’re introducing some reasoning into some decision making that the AI is able to do and you can’t always predict the outcome,” he said.