A single fabricated detail unravels months of customer trust. Churn is not a post-sales problem.
knowledge/patterns/status-quo-is-the-real-objection-outbound
knowledge/contradictions/save-everyone-vs-let-the-wrong-fit-go
I've done this work for 10+ years inside B2B SaaS teams from $1M to $50M ARR.
The substrate, my AI-native workflow, is open source and on GitHub.
Every number on this site has a citation. Trust, but verify. :)
Wingman homepage, repositioning shipped in 47 days.
Fincent homepage, rebuilt on a Value Proposition Canvas.
JustCall homepage, six refreshes from category label to buyer fear.
Research synthesis, competitive intel, win-loss reads, ICP and messaging variants, fact-check, citation, drift detection, launch comms, sales enablement, CRO writeups, help-center taxonomy.
A pipeline runs all of it now.
What's left is judgment: which angle, which buyer, which bet to commit the company to.
Retainers and fractional decks still charge by the hour for work the pipeline finished before the meeting started.
Q1 2026 at JustCall I deployed Commons across five operators on the team.
One shared substrate, write-back enabled, every claim sourced.
The team shipped landing pages 2 to 5x faster the next quarter, same headcount.
Substrate is the open-source refinement of three systems that came before it.
MessagingGPT (the precursor customGPT). Commons (the team-shared brain I shipped inside JustCall). Flywheel (the operator workflow I built personally, never shipped to a team).
MIT, on GitHub. Skim the principles, read one skill end to end.
Thirty minutes is enough to tell you whether the way I think matches the way your team wants to ship.
Each rule below was learned in a chair, not on a slide.
Each one is now a principle, a pattern, or a skill in substrate.
A single fabricated detail unravels months of customer trust. Churn is not a post-sales problem.
knowledge/patterns/status-quo-is-the-real-objection-outbound
knowledge/contradictions/save-everyone-vs-let-the-wrong-fit-go
Your customer's first words are almost never the problem. That's two questions in.
skills/tactical-empathy-discovery
routines/frontline-contact-routine
knowledge/patterns/frontline-as-pmm-substrate
Whoever owns the source system owns the metric. Everyone else is reading.
knowledge/patterns/metric-tree-not-metric-stack
skills/metric-tree-design
skills/analytics-pull
The fastest way to lose a launch is to
let your angriest customers write PR for you.
skills/social-listening-themes
knowledge/patterns/launch-as-coordinated-distribution
Principle · honesty under pressure
Quality content and distribution are your moat.
skills/help-docs
skills/help-content-gap-detect
knowledge/patterns/quality-as-growth-lever
The buyer panel is the signal, not internal applause.
routines/customer-conversation-rhythm
skills/positioning-forge
knowledge/patterns/diagnose-before-execute
AI is not the multiplier. The context AI reads is.
case-commons.html
case-miniu-abcodex.html
No dashboard screenshots. No logo walls.
The number is rendered from the data underneath, the before-and-after is the type itself,
and the source is always one click away.
I've worked both shapes. Solo embedded operator at Fincent and PreSkale. Built and led teams at Wingman and JustCall.
Phase-1 engagements are built for the solo shape, the team shape composes from it when you want it to.
Author and sole maintainer · v1.0 to v1.2 in 4 days · MIT, open source
Read the case →git clone github.com/k3sava/substrate
Patterns I shipped at Wingman, Fincent, and JustCall, distilled into a system the next engagement gets pre-installed.
Thirty minutes tells you whether the way I think matches the way you ship.
Substrate · 2026-05 to live · three AEO skills, one weekly dispatcher · honest measurement status on the page
Read the case →Aug 2021 to Oct 2022 · sole PMM for the first year · positioning, homepage, win-loss, launches
Read the case →"Conversation intelligence for fast-growing sales teams"
"Your Revenue Accelerator."
Wingman · 2021 to 2022 · 11 comparison + alternative LPs · "we actually like Gong" as the opening concession
Read the case →JustCall · 2024 to 2026 · positioning, hero, social proof, vertical entry points
Read the case →Six homepage refreshes, 21 months. Every hero in Wayback.
AIVA on the homepage stage. +37% signups, +18% demos.
JustCall · H2 2025 · Phase 1 rollout · BPS 3.58 to 5.99% CVR, AI SDR 1.78 to 2.44%, ad A/Bs +26.2% and +27.1%
Read the case →JustCall · 2025 · integration pages as a pipeline channel · Wingify-verified A/B
Read the case →Generic integration listing. Feature grid, single CTA.
+48% signups, +25% SALs, +86% footer scrolls.
JustCall · FQ2 2025 · deal-level attribution · battle cards, demo flows, objection libraries
Read the case →Outdated one-pagers. No utilization tracking. The question "did enablement close anything" was unanswerable.
Named deals, named AEs, honest losses on the same sheet.
JustCall · 2025 to 2026 · Claude Code pipeline · slash-command runnable · zero LLM spend
Read the case →$ /fact-check landing-page.md
→ 42 claims · 5 BLOCK · 12 FLAG · 4 FIX · 1 MATH
Numerical claims shipped without verification.
Every claim labeled. Every page gated. Zero LLM spend.
JustCall · 2024 to 2026 · launch cadence, naming, packaging, positioning
Read the case →18 launches in 21 months. Cadence held when the GTM owner walked mid-flight.
AI SDR · #18 Product of the Day, Sep 2025.
JustCall · Q1 2026 · architecture: static context library to live operator substrate · auto-called from every Claude session
Read the case →Read-only. Snapshots. Five operators copy-pasting the same brief five ways.
Write-back. Auto-called by every Claude session. Akhil went from 2 to 5 LPs a week.
First marketer · Dec 2022 to Feb 2024 · positioning and homepage rewrite · Wayback-verifiable
Read the case →Pricing-table homepage. No hero, no narrative. ~1,700 chars.
VPC-driven hero, three testimonials, four benefit sections. ~4,300 chars.
2023 · advisory · positioning and GTM · what the founder shipped, what I contributed, named separately
Read the case →2026 · self-evolving agent and Brier-scored insights · abcodex.iamkesava.com
Read the case →2024 to 2026 · homepage · integrations · enablement · competitive · AI systems · product GTM
Open the index →If you've read this far, the next step is the offer.
Three phases. Two bounded. One priced on the lift.
Three phases, each one a decision gate.
Diagnostic is free. Scoping is fixed.
The outcome contract carries the fee on the metric we both signed off on.
"I'm not selling you my time. I'm selling you the move that produces the lift, with skin in the game." consulting-offer.md · ln 14
Tell me what you're trying to move. I'll send you a slot.
Got it. Pick a time that works for you:
Pick a 30-min slot →You'll hear from Kesava within one business day either way.
Prefer email? hello@iamkesava.com · LinkedIn: linkedin.com/in/k3sava · the shape of the engagement: offer.html
Read a principle, read a pattern, read a skill.
If anything doesn't check out, that's the conversation we should have, not the engagement.