Chay Patil C2X · Context-to-Execution

01/Signal

Chay Patil

Chay Patil

AI workflow consultant

Buying the best AI model isn’t enough. It needs to know how your business works.

Most teams already have AI. Few have the through line that makes it actually work.

Mumbai, India

[Let'stalk]

02/Selected Work

2022–2026

Selected work, shaped as systems.

Hey, I'm Chay. I turn your team's scattered AI tools and workflows into a system that actually knows your startup and runs without hand-holding.

Everis

AI Workflow & Growth Consultant · Prosp and Pebble

Feb 2026 – Sep 2026

Featured · Prosp

The AI setter

Every post pulled hundreds of comments. Each one now gets scored and answered personally, about 30 seconds a lead instead of 5 minutes, across 1,000+ warm leads every month.

Same 10 comments on a “comment AI” post. Two ways to answer them.

By hand00:00

Right now

Waiting for comments

 

    Answered 0 / 10
    AI setter00:00

    Right now

    Waiting for comments

     

      Handled 0 / 10

      Illustrative leads and names. Timings from the real system: about 5 minutes a lead by hand, about 30 seconds with the setter, and a human still sends every message. The clock runs fast.

      • AI setter1,000+ warm leads a month, scored and answered in about 30 seconds each
      • OnboardingRebuilt around the real drop-off point, found in 30+ support chats
      • AffiliatesFound the 41 partners (6.7%) driving 80% of revenue; win-back email hit 51% opens
      • WebsiteRedesigned Prosp’s site, built with Claude
      • AI videoA content pipeline shipping 20+ review-ready pieces a week
      • HiringAI-assisted hiring flows and tests on Testlify

      The setter, plus everything else I built at Everis.

      [ See how it works ]

      Peoplebox.ai S22Y Combinator

      Lead, Data Management & Annotation

      Feb 2025 – Jan 2026

      Featured · AI recruiter

      Teaching an AI to read engineering roles

      An AI recruiter has to know a Staff SRE from a junior QA before it can match anyone. My team tagged roles by domain and seniority, reviewed every label, and fed it back until the model hit 98%.

      98%Accuracy after fine-tuning, 5% above Gemini 2.5 Pro at the time
      <3 moPlanning to production
      5People on my team
      • DataEngineering roles tagged by domain and seniority, every label reviewed
      • ClientsUsed by teams like Hasura and Razorpay to hire
      • LoopClient feedback turned into labelling rules and review standards

      How the data, review and fine-tuning loop worked.

      [ See how it works ]

      Ticombo

      Content Marketer

      Oct 2023 – Feb 2025

      Featured · Event automation

      Live events, listed automatically

      Resellers can only sell what's on the marketplace. Instead of someone copying new events over by hand, automations watched live event sites and listed each new one for resellers.

      • AutomationNew events spotted on live sites and listed for resellers automatically
      • ContentWebsite and Instagram content with AI built into production
      • OpsMeta campaigns, plus hundreds of content tickets closed across teams in Jira and HubSpot

      How the event watcher and the content pipeline worked.

      [ See how it works ]

      Earlier

      • Agasthya E-cellCo-founded Christ University’s first student-led E-cell; Pitchfork drew 60+ teams nationwide
      • SoileumGrowth and community for a web3 startup, working with the marketing lead
      • Results.ioRan Discord communities for 6 web3 clients, engagement up 25%+

      My own venture

      fallØut

      Co-founder

      Oct 2025 – Now

      Built and run with AI

      One of the leading hard techno platforms in India

      Two people, no developer, no agency. The site, the ticketing, the finance tooling and the ad engine are all built with AI.

      • Reach600K+ organic monthly impressions and a 1M+ view video
      • PartnersPromotions partner for major events and organizers across India and Europe
      • StackOffline-capable QR ticketing on Postgres, finance tooling and ads, under $250 to build

      The venture I run alongside client work.

      [ See fallØut ]

      03/Practice

      In focus 01 / 05

      The model is the easy part. The system is the job.

      I build both: the AI that does the work, and the system that makes its output safe to ship.

      1. Find the leakI pick one workflow that eats time or keeps getting redone, and find where it actually breaks. Context break
      2. Redesign itI map what should happen: the steps, the inputs, what AI handles, what a human decides. Workflow logic
      3. Build it with what you haveI build into the tools the team already uses, so there is no new platform to adopt. Existing stack
      4. Keep a human in the loopAI drafts and prepares. A person approves anything that matters, and rejections become feedback. Review gate
      5. Hand it offThe team keeps the workflow, the rules, and the docs to run it without me. Operating system
      Context

      That's the through line. It holds from context 2 execution, and stays intact after I hand it off.

      Execution

      04/Contact

      Got a workflow that's more manual than it should be?

      [Let'stalk]
      or write directly build@beingchay.com