Send an email
Future of AIAI
Beyond chatbots: Embedded AI is the real enterprise game-changer
By Eilon Reshef, CPO and Co-founder, Gong
Back to top button
It wasn’t too long ago that voice assistants like Siri and Alexa felt like a window into the future. Ask a question out loud, and you’d be met with a natural (enough) sounding response. However, these virtual assistants were limited in the kinds of information they could draw from – who hasn’t been met with ‘search the web?’ when a question has stumped their assistant.
The emergence of generative AI connected many of these gaps, allowing people to pose any question, in any way they wanted, and receive a tailored response they could probe and modify further – redefining how people access information.
It wasn’t long before technology leaders began to take notice and imagine the possible business use cases – think connecting an AI chatbot to company data to create an autonomous copilot that could answer business questions in real time. No need for specialised apps, menus or dashboards, just talk to the machine for any business information queries.
That hasn’t been realised, at least not yet. Recent research found chatbots haven’t had a marked impact on earnings or working hours, and that’s because business processes need more than just a connected interface – they demand structure, consistency and context. The idea of replacing software with chat, while appealing on the surface, misses the bigger opportunity: embedding AI where people’s work actually happens.
One of the biggest challenges with chat-based AI in enterprise settings is what’s known as the “blank canvas” problem. Chatbots are open-ended by design: ask them anything, and they deliver a response. While that’s part of their appeal for consumers, it becomes a bottleneck for businesses.
Imagine a salesperson juggling multiple customers and prospects. Rather than needing to toggle away from an email client to ask a chatbot about an account, refine the prompts and return once they’ve reviewed the information, a business with AI embedded directly into its tools and systems could surface those insights immediately. In that instance, a chatbot simply adds unnecessary friction.
Too often, chatbots operate outside the systems where employees actually do their work. If a sales leader wants to understand pipeline risk, they might have to bounce between a CRM and a separate chat interface. Ask a question, switch applications, copy, paste, repeat. This doesn’t just slow things down, it increases the risk of error and keeps institutional knowledge siloed across tools.
There’s also a fundamental difference between reactive and proactive intelligence. Chatbots wait for input – they only respond when asked. But in the fast-moving world of sales, finance, or customer support, teams can’t afford to wait until someone thinks to ask the right question. They need to know when a deal is at risk, when a forecast is off or when customer sentiment dips, before they even think to ask.
Embedded AI doesn’t wait. It detects issues in real time, surfaces them within the tools employees already use, and recommends next steps. It’s not just answering questions, it’s steering outcomes.
Another issue with chat-first interfaces is consistency. If five employees are given the same brief but all query a chatbot for answers, they’ll receive five different responses. One user might prompt a chat interface with highly specific details while another might phrase their questions in a more topline way. This kind of variability doesn’t work in enterprise environments, where teams depend on shared processes and standardised outputs.
AI that’s woven into business systems helps enforce structure. It ensures that all employees are guided through the same best-practice workflows from directly within the tools like Microsoft Teams they’re already using. They can come off a call and be given actionable next steps that align with their organisations priorities and policy.
Bringing intelligence directly into workflows is the only answer. When AI is embedded within a sales platform, for example, it can automatically flag issues, suggest next steps and update forecasts in a predictable, consistent way. It doesn’t rely on prompts to be perfect, it just works.
Finally, there’s a real limitation to relying solely on text-based interfaces. Business decisions often rely on complex, multidimensional data that’s typically housed in multiple places. A chatbot can summarise the information it’s given, but it can’t offer the interactive experience of a dashboard that surfaces regional data, information about different customers or generate real time predictions.
Business users need more than words, they need insight that’s easy to digest, relevant to their context and actionable. Knowing what’s happening is less useful to employees under pressure than being instructed on what they should do next to optimise results.
While the novelty of chat interfaces is undeniable, business leaders cannot afford to confuse the medium for the message. While chatbots will no doubt continue to prove popular in consumer applications, the real opportunity in enterprise AI lies in how it can drive better decision making and outcomes, not its linguistic fluency.
When intelligence is embedded in the platforms workers already use, be it CRMs, their company’s messaging platforms and even other AI copilots, it allows for easier access within their workflows to information they can trust. That means automating mundane tasks like researching potential customers so employees can spend more time meaningfully engaging with them, armed with insights that set them up for success.
