Owning the Workflow: The Next Wave of B2B AI Applications

Owning the Workflow in B2B AI Apps

This article explores the evolution of B2B AI applications, moving from content generation (Wave 1) to information synthesis and workflow automation (Wave 2), often termed 'SynthAI'. The core idea is that B2B AI applications should aim to "own the workflow" by integrating seamlessly into users' existing processes, thereby increasing customer stickiness and enabling future expansion.

The Shift to Wave 2: SynthAI

Last year, Kristina Shen and the author predicted a second wave of B2B AI applications focused on synthesizing information. Unlike the first wave, which concentrated on creating new content (e.g., emails, marketing copy), Wave 2 emphasizes condensing information to save users time. This is achieved by capturing, storing, and processing information efficiently, ultimately aiming to automate entire workflows.

Understanding Workflows

A workflow is defined as a sequence of steps taken to complete a task. In knowledge work, this typically involves gathering information, applying context, and producing an output like an insight or decision. Software aims to accelerate these workflows by making information management easier and by performing tasks on behalf of the user.

AI and the Input→Output Paradigm

Large Language Models (LLMs) popularized a prompt-based, input→output mechanism. This is evident in terms like "text-to-speech" or "image-to-video." This paradigm maps well to workflows, where the goal is to transform an input of context and information into an output of action or insights. The workflow is the process that bridges the input and output.

Challenges in B2B Applications

A key challenge arises when applying the chat UX/prompting mechanism to B2B applications that are already designed with specific workflows in mind. Requiring users to "chat" with the AI to navigate context can disrupt their native workflow, creating an unnecessary layer of work. The ideal scenario is to integrate the input→output process natively into the product, ideally making the entire workflow a single click.

Owning the Workflow: The SynthAI Advantage

SynthAI offers the potential to achieve this by converting complex workflows into product features. This approach leads to greater customer retention and facilitates the expansion of services.

Example: FigJam

FigJam, Figma's online whiteboard, exemplifies workflow ownership. In a brainstorming session, users generate sticky notes with ideas. The typical workflow involves:

1. Sorting similar sticky notes.
2. Identifying and defining clusters.
3. Summarizing themes and takeaways.

Traditionally manual, these synthesis steps are now automated with a click in FigJam, saving users significant time (up to an hour per session).

Example: Macro

Macro, an AI-powered document editor, addresses the workflow of comparing multiple document versions with edits and comments. The process typically involves:

1. Identifying changes in each version.
2. Comparing conflicting changes.
3. Summarizing the impact and conflicts.

Macro's "AI Compare" feature automates these steps, which could previously take lawyers hours, streamlining the editing and negotiation process.

Example: Claygent

Claygent, Clay's AI web scraper, automates the workflow of researching company-specific attributes (e.g., competitors, pricing, POS providers). This often involves:

1. Navigating company websites.
2. Searching for specific information.
3. Repeating the process for multiple leads.

LLMs are adept at guided information retrieval, making SynthAI ideal for this task. Claygent automates this information scavenger hunt, saving significant time for sales professionals dealing with large lead lists.

Future Trends in Workflow Automation

The trend towards AI-driven workflow automation is expected to continue, with two primary evolutions:

1. More Proactive Automations: AI will increasingly execute workflows without explicit user action, recognizing when a task is needed. This could range from automatically notifying a solutions engineer about a customer objection to an AI avatar participating in meetings to answer questions.
2. Reimagined User Experiences: Reliable, proactive AI capabilities will fundamentally change how users interact with products. For instance, an AI-native CRM might move beyond relational databases and static fields, leveraging embeddings to capture nuanced customer relationships and proactively guide actions or surface information.

The Evolution of User Experience

AI-native applications will likely shift from traditional tabular data representations to more dynamic, context-aware systems. Embeddings will play a crucial role in capturing complex relationships, leading to user experiences that prioritize summary dashboards and notifications over static views. AI will also identify previously unrecognized workflows by analyzing patterns across different data modalities (text, audio, video), leading to automations that users might not even conceive of.

Conclusion

The B2B AI application landscape is rapidly evolving, with a strong focus on workflow ownership. As AI capabilities mature, we can expect more proactive automations and entirely new user experiences that fundamentally reshape how businesses operate. Startups building products to own their respective workflows are encouraged to connect with the author.

The article acknowledges contributions from Kristina Shen, Conor Woods, Shannon Toliver, Jacob Beckerman, Kareem Amin, and Adam Eldefrawy for their insights and feedback.