Prior to its big Dreamforce event next month, Salesforce gave a sneak peek as to the next phase of its AI vision today, with the pre-announcement of two new autonomous AI sales agents: Einstein Sales Development Rep (SDR) Agent and Einstein Sales Coach Agent. Generally available in October, both sales agents are built on Salesforce’s Einstein 1 Agentforce Platform, which is its new platform that will include prebuilt agents like these initial two, as well as allow users to create and configure other purpose-driven AI agents.
The announcement hints at a demarcation point in Salesforce's AI strategy. To reduce it to simple terms, the company has now entered the automated and autonomous phase; as it has evolved from Einstein insights, to generative AI copilots, to these new agents. And to be clear, Salesforce is intently pointing out that these are NOT "bots." These new agents are more outcome oriented, and are designed to handle far more complex and personalized interactions that your typical web chatbot. These agents should, in theory, have far more data points to work with to make this a reality.
The SDR agent is designed to engage with prospects and do the heavy lifting that comes along with lead qualification exercises. The vision is to have a virtual sales development team working 24/7 (either on its own or to supplement human SDRs who work typical business hours - and they are multi-lingual and can thus expand a single location SDR team into a global prospecting organization). The agents can interact over multiple channels: email, web chat, WhatsApp, SMS. These agents can answer questions, ask BANT and other qualifying questions, and when integrated with workflow can hand off qualified leads to the appropriate sales rep.
The Sales Coach Agent can simulate role plays in real time, for example prior to an important prospect call, and uses generative AI to create conversations and RAG to emulate potential buyer responses and objections. The “coach” can also sit as a sidebar on Zoom calls (more platforms to come) and provide real time coaching during a video call. The agent can also suggest followups and when integrated with workflow and task management actually start to populate fields and create tasks etc. in Sales Cloud, both driving productivity for sales reps but also ensuring the right post-call activities take place to further the opportunity along.
Salesforce notes that these agents can be implemented quickly by admins, with no-code, simple to follow prompts to get up and running. Empowering RevOps teams (who often are the admins of Salesforce deployments) to think strategically and not just implement but also create custom AI agents for various sales use cases is a compelling wrinkle here, and it will be interesting to see how RevOps teams and other admins start to create a type of marketplace for AI agents based on common jobs to be done.
The initial choice of agents to rollout makes a lot of sense in terms of the "what" the "who" and the "why." I believe Salesforce has chosen two common sales use cases that can be best augmented by AI. Human SDRs are typically doing a lot of repetitive, question-based work - which can be well automated and optimized by AI across channel. In fact, the ability to do this in an asynchronous and multi-channel manner may be an upgrade from human SDRs: Ai agents can collate data across channels and be less intrusive, allowing prospects to respond and provide info at their convenience on the channel they best prefer.
The same goes for the AI coaching. Even human coaches lack the ability to know everything about every prospect, so providing "real time" and personalized coaching is impossible. But AI coaches can analyze literally every single sales opportunity and identify interesting moments upon which to dig in and coach; as well as understand the sales cadence to provide roll playing and other pre-meeting/call exercises to optimize sales performance and improve every engagement. The ability for coaches to suggest follow-ups and execute on them is another huge productivity gain, but in an optimized manner, not simply a time saver. It is a great combination of gen AI and AI-powered automation.
The most interesting pearl here is how the Salesforce partner and customer community takes this and runs with it - in terms of creating configured/custom industry-focused and other purpose-built AI agents. The "mix and match" of agent and gen AI tools will result in some cool iterations.
The big issue for both Salesforce from a messaging perspective, and for buyers as an internal selling perspective, is assuaging the anxiety factor. Buyers need to keep sales support from seeing this as "human replacement" and more about "human optimization." For example: Ramping current SDRs into actual revenue drivers by supercharging their career progression with AI, etc. And assuring sales enablement professionals that these tools allow them to keep better tabs 1:1 with every rep in ways that is humanly impossible.
And of course, pricing is still a bit amorphous. Buyers should look closely to the costs associated with supporting a fully functional AI agent. This means implementing Data Cloud if it is not already in place, as well as undergoing significant data cleansing and quality exercises (as well as data entry best practices by sales reps AND customers) to get the CRM and supporting data into a position to drive the kind of reliable insights that can empower the agents in a useful and non-disruptive manner.
