AI-Powered Filtering of Overwhelming Digital Communication

The deluge of modern work communication, particularly Slack notifications, can feel like an insurmountable tide, drowning individuals in a sea of information that demands attention but rarely offers clarity. This conversation with Yash Tekriwal, Head of Education at Clay, reveals a powerful, albeit non-obvious, implication: the true challenge isn't managing the volume, but intelligently filtering and prioritizing it to reclaim agency over one's workflow. Tekriwal demonstrates how custom-built AI tools, leveraging platforms like Perplexity Computer, can transform overwhelming notification streams into actionable insights. This post is for any professional feeling buried by digital communication, offering a blueprint for building personalized systems that not only manage but actively enhance productivity, providing a distinct competitive advantage for those willing to invest in tailored solutions.

The Hidden Costs of an "All-Purpose Inbox"

Slack, while a powerhouse for communication, often devolves into an "all-purpose inbox" where crucial actions get lost amidst casual chatter and FYI messages. Tekriwal highlights this by noting that out of 100-150 daily Slack notifications, only 30-40 truly require his attention. This gap between perceived importance and actual urgency is a systemic issue. The immediate convenience of a unified platform masks a downstream consequence: the mental overhead of sifting through noise.

"Not all notifications are created equal in slack for example I care much more about getting back to you on scheduling our podcast recording then I do about my colleagues' really fun comment on their dog that they posted a photo of in the fun dog channel but I didn't equal notification for them"

This distinction is critical. The system treats every notification with equal weight, forcing the human user to perform the complex, time-consuming task of categorization. Tekriwal's initial solution involved using OpenClaw to build a custom digest. This wasn't just about aggregation; it was about imposing a structured workflow onto an unstructured problem. The digest categorized messages into four buckets: direct messages, group DMs, threads, and group mentions, and then further subcategorized these into "action required," "need to read," and "FYI." This layered approach directly combats the "all-purpose inbox" problem by creating distinct pathways for different types of information, mirroring how a well-designed email client separates promotions from primary messages.

The real insight here is the application of deterministic code for structured tasks. While AI can categorize, the underlying mechanics of notification retrieval and grouping are best handled by code. This hybrid approach--AI for nuanced tasks like categorization, and deterministic code for API interactions and data structuring--is where efficiency gains are found. The initial OpenClaw solution, while functional, still presented a long list of text and emojis, proving that raw data, even organized, can be overwhelming.

From Digest to Dashboard: Engineering Your Mental Workflow

The evolution from a text-based digest to a visual dashboard with Perplexity Computer exemplifies a systems-thinking approach to personal productivity. Tekriwal recognized that the interface and interactivity were as crucial as the content filtering. He envisioned a system that felt like "Superhuman for email," a benchmark for a frictionless, optimized user experience.

Perplexity Computer's strength, as highlighted by Tekriwal, lies in its multi-model orchestration and cloud-native connectors. Unlike single-model tools, it can leverage different AI models for distinct parts of a task--using one for planning, another for coding, and a third for execution. This ensemble approach reduces the frustrating "prompt, fail, re-prompt" loop common with other AI assistants.

"Perplexity Computer is actually not exactly how we initially solved this slack problem we'll come back to that in just a second but i think the framework is also what matters most is i needed to be able to envision what does a better world look like instead of just asking claude or perplexity or openclaw make my slack easier"

The dashboard itself is a masterclass in consequence mapping. It features three Kanban-style columns: "Action Required" (red, urgent), "Need to Read" (yellow, important but no immediate response), and "FYI" (green, informational). This visual hierarchy immediately signals priorities. Crucially, the "FYI" column includes an "Archive All" button that not only clears the dashboard but also the corresponding Slack notifications. This is a direct intervention against notification fatigue, providing a clean, decisive way to manage low-priority information without manual effort. This "archive all" function is a delayed payoff--it requires building the system, but the ongoing benefit is a perpetually cleaner inbox and reduced cognitive load.

The system's ability to link directly back to the original Slack messages is another critical element. This preserves the ability to access full context when needed, bridging the gap between the filtered dashboard and the raw source. This iterative refinement--from initial text digest to visual dashboard with deep links--shows a commitment to optimizing not just information intake, but the entire user interaction loop.

The Micro-Software Revolution: Building the Tools You Actually Need

Tekriwal's approach to solving his Slack problem is a microcosm of a larger trend: the rise of micro-software. He articulates a vision where individuals and small teams can build bespoke tools to address specific workflow inefficiencies, rather than waiting for SaaS giants to cater to niche needs.

"My dream is for someone else to watch this video and say I want to build that app on top of slack and then I can go pay that person 15 a month for this app to be maintained and used and then I can file bug reports with them instead of having to fix it myself because I would happily pay that"

This sentiment underscores a key implication: the cost of building custom solutions is decreasing dramatically. This lowers the barrier to entry for creating highly specialized tools. The "SaaS apocalypse" narrative, which suggests a decline in software-as-a-service, is reframed here as an "explosion of micro-software." Instead of one monolithic tool, we will see a proliferation of small, interconnected applications, each solving a very specific problem.

This has profound implications for competitive advantage. Companies or individuals who can identify and build these niche solutions can gain significant efficiency gains. For example, Tekriwal's team used Perplexity Computer to prototype persona-based learning journeys for Clay University. This wasn't about replacing existing design tools like Figma but about bridging the communication gap between the education team and designers. By creating a visual prototype that demonstrated user journeys for different roles (SDR, BDR, RevOps), they provided designers with concrete, actionable context, accelerating the design process and ensuring the final product met specific user needs. This proactive, custom solution avoids the long tail of customer requests that traditional SaaS products often ignore.

The "anti-to-do list" framework further reinforces this idea. By identifying tasks that are so undesirable they should ideally never be done manually (like deleting spam emails or manually entering meeting action items), individuals can focus AI efforts on automating these high-friction points. This is where immediate discomfort--the effort of building the automation--leads to lasting advantage: freedom from tedious, repetitive tasks.

Key Action Items

• Immediate Action (This Week):

  • Identify your top 1-2 most persistent workflow bottlenecks related to information overload or repetitive digital tasks.
  • Explore the concept of an "anti-to-do list" and jot down 3-5 tasks you would happily automate away forever.
  • Experiment with AI tools (like Perplexity Computer, OpenClaw, or even ChatGPT with specific instructions) to categorize a small, manageable stream of your notifications (e.g., one specific Slack channel).

• Short-Term Investment (Next Quarter):

  • Investigate platforms like Perplexity Computer that offer multi-model orchestration and cloud-native connectors for building custom applications.
  • Prototype a simple dashboard or digest for one high-volume communication stream (e.g., email, specific Slack channels) to filter and prioritize information.
  • Begin documenting your ideal workflow for a specific task and identify where AI could deterministically automate parts of it.

• Long-Term Investment (6-18 Months):

  • Develop a more sophisticated, multi-layered notification management system, incorporating visual elements (like Kanban boards) and direct action capabilities (e.g., archiving, linking).
  • Explore building custom tools to bridge communication gaps with cross-functional teams (e.g., prototyping design systems, summarizing research for stakeholders).
  • Consider how to productize a successful personal automation tool, potentially offering it as a micro-SaaS solution to others facing similar challenges. This requires embracing the discomfort of building something robust enough for external use.