Customer Data Remains In-house: Slack AI Lead Amid Concerns Jackie Rocca and the team began designing and developing Slack AI around October last year. They kept three vital things in mind: safety, relevance, and usability

Helping people work smarter and faster is Slack's true intention behind deploying artificial intelligence and generative AI among its customers. The 2009-founded collaboration platform has been leveraging AI and machine learning for several years for its search functionality and channel recommendations. And in February, on Slack's 10th year anniversary, it launched its AI subscription product- Slack AI for businesses of all shapes and sizes.

"We released summarization and search capabilities first because we believe those will help customers get the most immediate value from all of that conversational data," shares Jackie Rocca, VP of Product, Slack with Entrepreneur India.

Every week, about 4.7 billion messages are sent on the platform. Notably, 97 min on average were being saved by the likes of SpotOn, Uber, and Anthropic during the pilot program.

Having joined as a Senior Product Manager in 2018, and being promoted to VP in 2023, it was during Rocca's parental leave when the world got introduced to ChatGPT and its impending capabilities. She was keen on understanding how some of those capabilities could help the entity, "We are focused on looking at ways that AI can be used to address the top user pain points that people have been facing in the workplace for years."

"We know that finding and consuming information efficiently is a major challenge. Plus, we have more tools and ways of exchanging ideas than ever before. All of this adds to our cognitive load and makes it feel difficult to stay on top of the workday," she adds.

Slack AI's offerings can be broken down into personalised intelligent responses to conversational questions; conversation summaries that generate highlights from accessible channels and threads; user catch-up on unread messages; and summaries for the last seven days (or custom date range up to 365 days). "Clear sources are included in each summary, allowing users to dive deeper into a highlight," Rocca adds.

Earlier, she reported that Slack has over 10,000 AI-powered tools that are available through applications or third-party developers, including PagerDuty, Notion, and Perplexity. Since its inception, Slack acquired the likes of Screenhero, Spaces, Astro, Rimeto, and Daylight up till its own acquisition by Salesforce in July 2021.

"Your data is not our product"

Rocca and the team began designing and developing Slack AI around October last year. They kept three vital things in mind: safety, relevance, and usability.

They are ensuring that at the end of the day, humans are in the loop and are always in control. Its Enterprise Key Management (EKM) empowers customers to control their encryption keys – even giving you the autonomy to revoke access whenever you choose. After all, transparency builds trust.

However, a week back, Slack came under fire when Corey Quinn, an executive at DuckBill Group, spotted a loophole that enabled Slack to train machine-learning models on user messages, files, and other content without explicit permission.

One of its privacy principles section read "To develop AI/ML models, our systems analyze Customer Data (e.g. messages, content, and files) submitted to Slack as well as Other Information (including usage information) as defined in our Privacy Policy and in your customer agreement."

Slack clarified how its customers' data was used on X, "(it) has platform-level machine-learning models for things like channel and emoji recommendations and search results. And yes, customers can exclude their data from helping train those (non-generative) ML models. Customer data belongs to the customer. We do not build or train these models in such a way that they could learn, memorize, or be able to reproduce some part of customer data."

Rocca notes customer data remains in-house.

"Our guiding principle as we built this product was that the privacy and security of customer data is sacrosanct. Slack AI uses LLMs hosted directly within Slack's infrastructure, so that customer data remains in-house, is not shared with any LLM provider, and is not used to train LLMs. This ensures that customer data stays in that organization's control and exclusively for that organization's use," Rocca adds.

Slack and the AI scale

Parker Harris, CTO, Slack and Co-Founder, Salesforce at Salesforce's World Tour event in New York City last year said "How do I go as fast as possible and not make mistakes along the way?" "We have to worry because our customers trust us to help them in the best way possible," he remarked.

So, where does Slack stand on the scale of effectively adopting AI? "Our primary goal with Slack AI is to provide immediate value, without forcing desk workers to learn brand-new skills to enjoy its benefits," Rocca said.

"We believe that the most effective way to encourage widespread adoption of generative AI is to integrate it seamlessly into employees' existing flow of work, meaning the experience doesn't require users to context switch or get pulled out of their daily routine, and relevant outputs are surfaced in the place they spend the majority of their workday," she adds.

By 2019, it clocked in over 10 million daily active users. According to Colorlib, as of 2023, the collaboration platform boasted of around 20 million daily active users under its belt. Since entering the Indian market in 2018, Slack counts the country among one of its top 10 markets.

Region-specific plans

At the core, Rocca's vision is that Slack, powered by AI, will be the command centre for work. "We'll continue to develop Slack AI's search and summarisation capabilities to tap into new data sources – like files, third-party apps, canvases, and clips – to enhance the breadth and depth of context that Slack AI can access. This includes the ability for Slack AI to deliver a summary of key takeaways and action items from huddles, and a future integration with Salesforce's Einstein Copilot so users can bring CRM insights directly into Slack," she said.

It is aiming for a broader language support, starting with English, Japanese, and Spanish, "You can expect to hear more from us on this soon."