Commyt
Founder, Product Lead & Designer — AI Job Search Companion for LinkedIn
Role & Setup
I founded Commyt. I designed it. I shipped it.
Commyt is not client work. It's a product I conceived, validated, designed, and brought to market as solo founder. The build model: me as Product + Design lead, paired with AI as my engineering and prototyping partner across every phase.
My Role
Founder, Product Lead & Designer
Vision, product strategy, end-to-end UX/UI across web app, Chrome Extension, landing page, and pricing model.
Build Model
Solo founder × AI engineering pair
I drove every product decision; AI tools (Cursor, Claude, Gemini, v0) accelerated implementation, scaffolding, and the feedback loop.
What I Owned
- Problem framing & positioning
- Feature prioritization & roadmap
- End-to-end UX, UI, design system
- Chrome Extension UX
- AI prompt design (scoring, messages, CV)
- Pricing model & free/Pro split
- Bilingual EN/ES system
- Launch & measurement plan
Honest scope note: I led the full product setup and implementation, including API integrations, backend configuration, account setup, and system workflows. I guided the process end to end and ensured everything worked correctly, with AI assisting in code generation and execution.
The Why
Built because the existing tools were working against people
I watched too many brilliant people lose their spark during a job search — not because they weren't ready, but because the tools were working against them. Spreadsheets that broke at 30 applications. Generic InMails that got ignored. Fifteen tabs open, no system, no signal, no momentum.
Commyt is the product I wished existed when I was applying: one-click capture, AI-tailored CVs, honest match scoring. The boring parts compressed. The human parts protected.
Overview
What Commyt does
Commyt captures opportunities directly from LinkedIn, scores every job against your CV, and helps you reach the right people with AI-tailored outreach. A web dashboard for the deep work, a Chrome Extension for the moment of capture, and AI as a quiet co-pilot through the whole pipeline.
The Product
Nine features, one focused mission
Every feature in Commyt exists to remove a specific friction in the job-search ritual — or to protect the parts that should stay human. Below: the problem each one solves, the design solution I shipped, and why it works.
One-Click LinkedIn Capture (Chrome Extension)
ProblemCopy-pasting job links into spreadsheets kills momentum — most people quit tracking by job 10.
SolutionSave any LinkedIn job in one click. Job details, company info, and visible contacts are auto-extracted instantly.
WhyFriction has to die at the moment of intent — not five tabs later in another app.
CV-Based Match Scoring
ProblemApplying to everything that vaguely fits is exhausting and reads as desperate.
SolutionUpload your CV once; every captured job gets a fit % with breakdown by skills, experience, and preferences.
WhyPrioritization should be a glance, not a debate. The score makes the decision a 2-second call.
AI-Tailored Outreach Messages
ProblemCold InMails and connection requests get ignored because they're generic.
SolutionAI generates a personalized message per role and company, grounded in your CV and the job description — user always edits before sending.
WhyAI removes the slow part (drafting). The human keeps the part that matters (specificity and voice).
AI CV Tailoring (Gemini)
Problem"Tailor your CV to every job" is the advice everyone gives and nobody follows — it's too slow.
SolutionGemini generates a tailored CV per role in seconds, matching your real experience to the JD. You review and ship.
WhyA good ritual that's friction-free actually happens. This is the most-requested Pro upgrade trigger.
Skill Gap Analysis
ProblemRejections don't tell you why. Users guess what's missing and spiral.
SolutionScoring engine surfaces missing skills per role + actionable recommendations to close the gap.
WhyTurns a "no" into a learning loop — and gives users a roadmap they actually trust.
Find Key Connections
ProblemWarm intros convert ~8× better than cold applications, but finding the right path is buried in LinkedIn.
SolutionSurface your contacts within target companies, filter by relationship strength, suggest the highest-leverage intro path.
WhyReframes networking from "ask for help" to "find the right person fast."
Kanban Pipeline
ProblemSpreadsheets break at 30+ active applications across stages.
SolutionVisual Kanban: Saved → Applied → Interview → Offer. Drag, drop, see the whole pipeline in one screen.
WhyStatus changes are exactly when you most need the bird's-eye view — the linear table format hides it.
Smart Reminders
ProblemFollow-ups die in the gap between "applied" and "next step." Most jobs are lost here.
SolutionTimed nudges based on application stage and last activity — surfaced inside the app, not lost in email.
WhyThe difference between "applied and forgot" and "applied and got the job."
Activity Heatmap
ProblemJob search is demoralizing. Seeing zero output makes you think you're not trying.
SolutionA GitHub-style heatmap visualizes consistency over time — small actions add up to a visible streak.
WhyVisible progress is fuel for emotional persistence. Most users won't quit if they can see they're moving.
In Context
The dashboard at a glance
One screen ties together capture, scoring, the kanban, AI tools, and reminders — collapsible sidebar, borderless cards, glanceable AI score badges per row.
Product Decisions & Tradeoffs
Three product calls where the obvious answer was wrong
Every founder decision is a tradeoff. Here are three places where the easy answer would have hurt the product, and the rationale for what I shipped instead.
Where should the free / paid wall live?
Option A
7-day free trial, then card required
Standard SaaS playbook. But predatory for job seekers who are often without income.
Option B
Hard wall after 20 applications
Forces upgrade at exactly the moment the user has the most momentum. Felt punishing in tests.
Chosen · C
Free with meaningful limits
20 apps, 2 AI CV tailorings/mo, 3 AI messages, basic scoring. Pro unlocks the AI features that get you hired.
Why C: Job seekers can't be coerced — they leave. The Free tier delivers real value (capture, kanban, basic scoring). The Pro tier targets the moment of conviction: "this AI just tailored my CV in 5 seconds — I need unlimited."
How aggressive should the AI outreach voice be?
Option A
Hyper-salesy & familiar
"Hey [first name]! I noticed you posted about..." Recipients see right through it; smells like a bot.
Option B
Generic & safely professional
Defeats the entire point — why use AI if it sounds like a template?
Chosen · C
Warm + specific + concise, user-edited
AI drafts with one concrete reference (a project, a post, a shared connection). User edits before sending.
Why C: AI is the draft, not the author. The "user edits before sending" is non-negotiable design — it preserves voice, accountability, and the human signal that makes outreach actually convert.
How transparent should the AI scoring be?
Option A
Just a number (e.g., "74")
Compact, but cold. Users don't trust what they don't understand — and can't act on it.
Option B
Full breakdown always visible
Maximum transparency, but kills row density — the table drops from 12 visible jobs to 5.
Chosen · C
Color badge + score, breakdown on demand
Glanceable up front; click for skills/experience/preferences breakdown. Auditable but not noisy.
Why C: Trust in AI requires auditability, not opacity. Users need to be able to ask "why is this a 74?" and get an answer in one click — otherwise they ignore the score and the feature dies.
Status & Measurement
What shipped, what we're measuring, and how we'll know
Commyt is live in production and in active iteration. Rather than claim numbers I don't have yet, here's the full picture: shipped scope, the bets the product is making, and the measurement plan.
Shipped · Scope Metrics
What was actually built & delivered
- 3Production surfaces — Web App, Chrome Extension, Landing Page
- 9Core features shipped end-to-end (capture, scoring, AI messages, AI CV, gap analysis, connections, kanban, reminders, heatmap)
- 2AI integrations: Gemini for CV tailoring, GPT for outreach drafting
- 2Pricing tiers (Free / Pro $9.99/mo) with deliberate feature split
- EN / ESFull bilingual system across product & landing
- 1-clickLinkedIn capture latency — the design″s north-star UX metric
Hypotheses · Being Measured Now
The five product bets Commyt is making
One-click extension capture removes the #1 reason users abandon job tracking — people will save 3× more jobs in week 1 than with a manual tool.
Target: median 12+ jobs captured in week 1 · Measured via: extension event logCV-based scoring tightens the saved-to-applied ratio — users apply to fewer, better-fit roles.
Target: > 60% of applied jobs scored ≥ 70 fit · Measured via: kanban transition eventsAI-tailored messages outperform generic outreach — reply rates beat the LinkedIn cold-InMail baseline.
Target: ≥ 15% reply rate (vs ≈ 6% LinkedIn baseline) · Measured via: user-reported reply trackingAI CV Tailoring is the single biggest free→Pro upgrade trigger — users hit the 2/month limit and convert.
Target: ≥ 4% of free users upgrade in first 30 days · Measured via: Stripe + product event funnelSmart reminders keep applications alive past the "applied and forgot" graveyard — more jobs get a follow-up action.
Target: ≥ 50% of "Applied" jobs receive a follow-up within 7 days · Measured via: reminder action eventsMeasurement Plan
How we'll know
- Day 30Quantitative read on H1 & H2 + 5 moderated user sessions
- Day 60Pro conversion analysis (H4) + AI message reply tracking (H3)
- Day 90Full hypothesis review & iteration plan for failed bets
Tools
Stack for evidence
Funnels for activation & Pro conversion, heatmaps for layout validation, Maze for unmoderated 5-second tests, Stripe for revenue truth.
Founder Note
Why I'm building this in the open
I watched too many brilliant people lose their spark during a job search — not because they weren't ready, but because their tools were working against them. Commyt is the tool I wished existed: one-click capture, AI-tailored CVs, honest match scoring. The boring parts compressed. The human parts protected.
AI-Enhanced Workflow
How AI let one founder ship a full product
The premise
Commyt would not exist as a solo project five years ago. AI didn't replace the design or product thinking — it removed the friction that used to make founder-led products require a five-person team. I made every product call. AI executed the boring middle.
Discovery & Validation
Synthesized 14 user interviews into themes; surfaced contradictions across personas.
Recruited & ran the interviews. Killed AI-suggested clusters that flattened nuance. Made the call on what to build.
Product Design & UI
Generated dashboard layout variants, palette options, microcopy alternatives.
Every taste call: final palette, type, component anatomy, micro-interactions, brand voice. AI proposed, I disposed.
Engineering & Build
Scaffolded React + the Chrome Extension manifest, wired Gemini & GPT integrations, generated Tailwind from Figma specs.
Architecture decisions, prompt design for scoring & CV tailoring, every UX detail the AI missed, QA on every shipped feature.
Launch & Iteration
Drafted landing copy variants, generated demo videos with Remotion, summarized user-test recordings.
Pricing model, positioning, landing page direction, prioritized which user feedback becomes a roadmap item.