From: redpointai
The landscape of email is undergoing a significant transformation, driven by the advent of artificial intelligence (AI) agents. Superhuman, a prominent email tool, is at the forefront of integrating AI to redefine the user experience and productivity in email communication [00:00:30].
The Enduring Nature of Email
Despite perennial claims of its demise by journalists, email continues to be a cornerstone of professional communication [00:01:36]. This resilience stems from the email address itself serving as a fundamental, company-owned unique identifier, essential for corporate addressing and single sign-on (SSO) solutions [00:01:45]. While the email address remains, the nature of email communication is poised for dramatic change [00:02:40].
The Future of Email and AI
The future of email, as envisioned by Superhuman, is characterized by two major themes: AI and collaboration [00:02:47]. The most significant shift will be towards an agentic AI future, where AI agents operate autonomously around a user’s email, performing tasks on their behalf [00:03:00].
Superhuman’s AI Features
Superhuman’s integration of AI has evolved through distinct phases:
Phase One: On-Demand AI Features
The initial approach focused on “on-demand” AI features that users actively invoked [00:06:26].
- Write with AI: This feature allows users to jot down a few words, which are then transformed into a fully written email that matches the user’s voice and tone from previously sent emails [00:04:21]. It can also shorten, lengthen, improve writing, fix mistakes, and change language [00:04:32].
- This feature was launched in June or July of the previous year [00:04:39] and led to significant virality for Superhuman [00:15:24]. This phase served as a way to understand the technology, assess its quality, and gauge user adoption [00:06:43].
Phase Two: Always-On AI Features
Following the success of on-demand features, Superhuman transitioned to more ambitious, “always-on” AI capabilities that continuously work for the user [00:06:56].
- Auto Summarize: Launched in October [00:04:47], this feature provides a one-line summary above email conversations, which updates instantly as new emails arrive [00:05:00]. Users can also access a bullet-point summary [00:05:11]. This often eliminates the need to read the full email, enabling faster processing [00:05:21].
- Instant Reply: Released recently, this feature drafts replies for every incoming email, allowing users to simply edit and send, or sometimes send without any edits [00:05:30]. Users of Instant Reply are writing emails twice as fast, effectively cutting reply time in half [00:05:57]. The product design of Instant Reply was carefully considered, opting for always-visible short, snappy replies to reduce user activation energy [00:18:12]. Key to its usability is the seamless “Tab and Enter” interaction [00:22:17].
These features are more complex and expensive to operate due to their continuous nature and the massive scale at which they process emails (e.g., 4 billion emails processed since launch for Auto Summarize and Instant Reply) [00:07:05].
The Agentic AI Future
The ultimate vision for Superhuman is an agentic AI future, which forms the foundation for phase three of their AI roadmap [00:07:50].
- Users will likely have multiple AI agents, including a dedicated email agent [00:08:00].
- These agents will be largely autonomous, capable of handling ambiguity by asking questions and interacting with other systems or even other agents [00:08:08].
- The goal is for AI agents to handle routine tasks, freeing up humans for more creative, strategic, and impactful work [00:08:32].
Behind the Scenes: AI Development
Superhuman’s rapid AI development was achieved through a specific operating model:
- Prioritization: The company made an existential decision to prioritize AI, operating under a “Theta mode” where an executive (the CEO) is embedded directly in the project team, ensuring rapid progress and existential commitment [00:14:00].
- Prompt Engineering: Superhuman relies heavily on advanced prompt engineering and multi-shot learning (using past emails to match voice) rather than fine-tuning models, which avoids time-consuming data labeling and training [00:25:06]. Their prompt for Instant Reply is notably complex and large [00:23:22].
- Testing and Iteration: Due to the unpredictable nature of LLM updates, they built a robust regression testing and evaluation framework using Brain Trust to ensure features like Instant Reply function correctly across models and prevent issues like irrelevant replies [00:26:41].
- Model Selection: Superhuman primarily uses OpenAI models (GPT-3.5 Turbo for Instant Reply) due to their perceived 6-9 month lead in model quality, commitment to aggressive cost reduction, and strong collaboration [00:29:28]. Throughput and latency are more critical than raw model intelligence for bursty, real-time features like Instant Reply [00:31:44].
Teaching New Workflows
A core challenge for AI companies is teaching users new workflows [00:34:36]. Superhuman’s strategy is to:
- Start with “always-on” features (like Auto Summarize and Instant Reply) which have 100% reach [00:35:06].
- Connect these to “on-demand” features (like Write with AI) [00:36:09]. For example, after an Instant Reply is inserted, buttons like “lengthen” or “custom” could lead to the full Write with AI prompt, subtly educating users about the on-demand feature [00:36:26]. This approach leverages the inherent utility of always-on features to naturally introduce users to more powerful, but less immediately obvious, capabilities.
Broader Implications of AI Agents
The discussion extends beyond email to the broader role of AI in future business operations.
- Cross-Agent Interaction: The future envisions an ecosystem where personal AI agents interact with other agents (e.g., HR AI, benefits AI) to resolve complex requests autonomously, transforming multi-step human interactions into instantaneous, natural language conversations [00:43:24].
- Platform for Agents: Products with high daily engagement (like email and chat platforms) are well-positioned to become the “pane of glass” through which users interact with their various AI agents [00:44:50].
- Impact on Job Functions: AI is expected to significantly impact various jobs, particularly entry-level roles in customer support and sales, where AI agents can achieve 80% of human performance at a fraction of the cost [00:56:02]. This shift will extend to many professional roles, including even executive positions, leading to a redefinition of what “work” entails [00:56:55].
Conclusion
The integration of AI agents into email management, as demonstrated by Superhuman, is not merely an incremental improvement but a fundamental re-imagining of productivity and interaction. This shift towards autonomous AI agents promises to free up human potential for more strategic endeavors, changing the very nature of work and daily professional life.