Podcast Summary: How I AI

Episode: "I built a custom Slack inbox. It was easier than you’d think."

Host: Claire Vo

Guest: Yash Tekriwal (Clay)

Date: April 8, 2026

Episode Overview

This episode features Yash Tekriwal, Head of Education at Clay, sharing a step-by-step deep dive into how he built a custom, AI-powered Slack inbox to efficiently triage his overwhelming daily deluge of notifications. Host Claire Vo and Yash discuss practical AI workflows for productivity, unique approaches to personal software customization, how Perplexity Computer leaps ahead of its competitors for automating daily work, and what the future of SaaS looks like when AI enables users to solve personalized problems themselves. Listeners get both technical how-tos and philosophical reflections on demystifying AI beyond buzzwords.

Key Discussion Points & Insights

1. The Problem: Slack Notification Overload

• Yash's daily chaos: Wakes up to 100-150 Slack notifications—most are low-urgency, but it’s hard to spot what’s critical ([00:00]).
  • “Not all notifications are created equal in Slack... I care much more about getting back to you on scheduling our podcast recording than I do about my colleague’s really fun comment on their dog...” (Yash, [03:42])
• Core challenge: Slack gives the same level of urgency to all notifications, making inbox triage mentally taxing.

2. First Solution: DIY Digest with AI Assistants

• Initial prototype: Built an automation to pull just the actionable Slack messages into a digest, using OpenClaw as a coding agent in Discord ([06:06]-[10:44]).
  • Segmented notifications into DMs, group DMs, threads, @mentions, and further subcategorized by “action required”, “need to read”, “FYI”.
  • “Those 100+ notifications now come in to this Jarvis Digest channel in Slack... grouped into direct at mentions, then DMs, group mentions, and threads.” (Yash, [07:45])
• Limits of this approach: Still a long text scroll; overwhelming after a week ([11:51]).

3. Building the Visual Dashboard: The Slack Kanban UI

• Evolution: Used Perplexity Computer to quickly build an interactive Kanban-style UI for Slack inbox ([11:51], [18:42]).
  • Action Required (Red): Needs Yash’s input now
  • Need to Read (Yellow): Worth attention, not urgent
  • FYI (Green): Informational only, one-click archive removes from Slack and dashboard
  • “The best thing about this dashboard for me is that... the FYI notifications—I can just go ahead and click this Archive all button. They’ll disappear from the dash and then those notifications will also disappear on my Slack.” (Yash, [19:38])
• This mirrors a personalized “superhuman for Slack” workflow—an individualized productivity overlay.

4. Why Perplexity Computer? Next-Generation AI Tooling

• Parallelism and connectors: Yash details why Perplexity Computer excels for automations over Claude, Codex, or OpenClaw.
  • Can run multiple tasks in parallel and leverages many built-in connectors (Google Drive, Gmail, Notion, Asana, Slack, etc.).
  • “Concurrent runs and long running tasks are really nice in Perplexity Computer... because it’s in the cloud, it is much more natively connected to all these different tools. I don’t have to go give it... access to different skills; it can just go do it.” (Yash, [14:58], [15:25])
• Composable orchestration: Perplexity uses best-of-stack AI models (Sonnet, Gemini, Opus) within each workflow.

5. The Future of SaaS and Personalized Productivity

• Micro-SaaS and AI: Explores how custom software will flourish, not because AI’s replacing SaaS, but because individuals can now bridge gaps for their own workflows.
  • “You will see an explosion in software being created and used because of all these tools. I don’t think the average person is going to start custom coding all these tools...I’d happily pay $15 a month to someone else who builds and maintains this.” (Yash, [21:26])
• Opportunity: Low-code, AI-generated apps can thrive even with smaller user bases—doesn’t have to be “VC scale”.

Notable Quote:
“This is such a better way to just get through your queue... You don’t have to change how you work—you just get to work better.” (Claire, [19:51])

Detailed Walkthroughs & Timestamps

How to Build a Custom Slack Digest

• Prototype in Discord/OpenClaw ([06:06]-[11:14])
• “I try to clear my DMs ASAP... But even within those categories, I want to subcategorize by what requires real action from me, what do I need to read, and what are more FYI.” (Yash, [03:42])
• Iterative prototyping—thousands of messages back and forth with the agent

Moving to Perplexity Computer ([11:51]-[18:42])

• Building the UI: Kanban dashboard explained and demoed ([18:42])
• “80% of what you just saw...was built in like the first four messages.” (Yash, [12:48])
• Orchestration: AI chooses different models for subtasks ([13:30])

Real Connectors in Practice ([17:38])

• Demo: Google Drive, Gmail, Notion, Asana, Slack, Forms, Zoom, Airtable, and more ([17:38])

Bonus Use Cases

• Other Command Centers Built by Yash ([26:32])
  • Merges Slack, email, and news into one digest
  • Quick iteration with AI: From simple text to interactive UI with deep links
• Team Usage
  • Clay University persona-based onboarding prototyped visually with Perplexity ([32:08]-[35:13])
  • “Being able to build a visual bridge between [design and content] is incredibly valuable.” (Yash, [34:39])

Fun and Personal AI Use Cases

• AI for Board Games and Party Planning ([36:25]-[39:04])
  • Brainstorms unique team games and social activities
  • “I had a long brainstorming session with Jarvis about what are all the activities we’ve done before... It gets me in the thought process... then I make modifications myself.” (Yash, [38:13])
• Claire’s “anti to-do list” philosophy: Use AI to prioritize and automate tasks you don’t want to do ([24:04])

AI Prompting Strategies When Things Go Wrong ([40:39])

• Three approaches:
  1. Recognize when the task is a bad fit for the model; sometimes code or API is better
  2. Be “strict” with the AI: use all caps, extreme examples to elicit better results
     • “I threaten the model all the time. All the time. It’s so unusually effective.” (Yash, [43:05])
  3. Skill-building: have the AI explain its logic, iterate on prompts/skills through dialog
• Iterative improvement: “It’ll take 10, 12, 20 messages, but you’ll notice the improvement gradually.” (Yash, [42:36])

Notable Quotes & Memorable Moments

• “If someone else watches this, builds this app—I’ll pay $15 a month just to never see FYI’s again.” (Yash, [21:26])
• “Slack is great for sending messages. I don’t know if it’s great for reading messages.” (Claire, [19:51])
• “Let’s go do more. There’s so much useful software we could have because it’s been so expensive to build and not worth people’s time.” (Claire, [23:47])
• “I threaten the model all the time. All the time. It’s so unusually effective.” (Yash, [43:05])

Practical Takeaways

• Personal productivity isn't just about tools—it's about aligning workflows with your own priorities, with AI assistance.
• Modern AI platforms allow non-engineers to quickly build complex, integrated automations across multiple apps.
• AI can power not just work automations, but ideation and fun projects as well.
• The SaaS landscape will be transformed by micro-solutions built by and for “power users”, unlocking value without needing billion-dollar scale.

Episode Timestamps

| Segment | Timestamp | |------------------------------------------------|----------| | Slack Overload: The Pain | [00:00]-[03:42] | | AI-Powered Digest: The Early Prototype | [06:06]-[10:44] | | Dashboard UI: Kanban for Slack | [18:42]-[19:51] | | Why Perplexity Computer? | [14:29]-[18:21] | | Future of SaaS/Micro-SaaS explosion | [21:26]-[24:04] | | Connectors and Multi-Source Digests | [17:30]-[28:33] | | Clay University Persona Use Case | [32:08]-[35:13] | | Fun Use Cases: Board Games, Social Events | [36:25]-[39:04] | | Prompting Strategy When AI Fails | [40:39]-[43:05] |

Closing

Yash’s journey is a masterclass in the practical application of AI—not just building for novelty, but creating systems that reclaim time and sanity. Between technical breakdowns and philosophical takes, the episode inspires listeners to reimagine AI as a tool to mold software around themselves, not the other way around.

Find Yash:

• LinkedIn or Twitter/X: “Yash Tekriwal” or @yashtuck
• Always happy to help and share!

Recommended for:
Anyone drowning in notifications, curious about hands-on AI automation for productivity, or looking to ride the next wave of personalized, user-driven software creation.