Comentiq

Approved campaign stories

Story angle

Building the tool the cohort actually needs: a project manager that earns engagement through real accountability, not hollow gamification

The problem Mission Control solves is precise and honest — generic task boards either bore cohorts into abandonment or reward activity theater instead of shipped work. The Kardashev-inspired progression system is only credible because level-ups are gated on auditable outputs (merged PRs, unique contributors, MVP completions), not points that can be inflated. That design integrity is the story. The recent fix to unlimited peer voting shows the team is already stress-testing the integrity of that system in practice.

Most cohort PM tools pick a lane: useful but boring, or engaging but fake. Mission Control (built for @Hult Summer 2026) is trying to hold both. Real PM: assignable tasks, due dates, status boards. Real motivation: civilization levels that only unlock when actual work ships — merged PRs, unique contributors, MVPs completed. Not vibes. Auditable gates. This week: patched unlimited peer voting. Because if peer recognition can be gamed, it means nothing. Still early. Still building. Looking for a technical collaborator and first users inside the cohort. #HultSummer2026 #MissionControl #BuildInPublic

Partner summary

Mission Control is an early-stage project management platform being developed for the Hult Cohort Developer Program's Summer 2026 pilot. It addresses a documented failure mode in cohort tooling: platforms that are either too generic to sustain engagement or reward activity over output. The platform combines standard PM functionality (task assignment, due dates, status tracking, filtering) with a Kardashev-inspired progression system where cohort-wide 'civilization levels' advance only when transparent, auditable thresholds are met — merged pull requests, unique contributors, MVP completions, and adoption. This design ensures the motivation layer is downstream of real work, not a substitute for it. The team recently resolved an unlimited peer voting issue in the individual leaderboard, reflecting active iteration on system integrity. Built on Next.js, Supabase, Tailwind CSS, and deployed on Vercel. Currently seeking a technical collaborator and early users for the Summer 2026 pilot. Organizations running developer programs, accelerators, or multi-week build cohorts who face the same engagement-versus-accountability tradeoff are a natural fit for partnership or early adoption conversations.

Story angle

A builder shipped the platform that runs its own campaign — Comentiq is live in production, evidence-grounded, and demonstrating its own loop in real time.

The most credible proof that an AI campaign tool works is watching it document its own shipping. Every feature in Comentiq's update — auth, approvals, peer amplification, partner enquiries — is live and verified, not promised. That closes the gap between builder claim and public evidence in a way that generic 'building in public' posts rarely do.

I built a platform that turns builders' weekly progress into evidence-grounded campaign content — then used it to document shipping the platform itself. Comeniq is live in production as part of the Hult Summer Cohort 2026. Here's what's working end-to-end right now: → Signup, profile, and project creation → Progress updates that feed an AI drafting layer (Claude, via Anthropic API) → Every AI draft is grounded only in evidence the builder actually supplied — the system prompt enforces this → Mandatory human approval before anything goes public → Peer amplification: cohort members can 'Boost' each other's campaigns with personal endorsements → A public showcase where partners can browse projects and submit enquiries directly to builders The hardest parts weren't the features — they were the edge cases. An RLS policy that was silently dropping public amplification content from the UI. A git branch mixup that buried a week of commits in the wrong submission branch. A Vercel deploy misconfigured across three settings simultaneously. All diagnosed, fixed, and in production. The lesson that's stuck: test the unhappy path deliberately. Assuming a passing build means a working feature is how silent bugs stay silent. Stack: Next.js 14, TypeScript, Tailwind CSS, Supabase, Anthropic API, Zod. What's next: GitHub OAuth, an analytics dashboard on top of the event log already in place, and an embeddable widget for partner sites. If you're a Hult builder who wants to turn your weekly standups into something the world can actually see — or a partner curious about what this cohort is building — the showcase is public and the enquiry pipeline is open. hult-cohort-program-one.vercel.app #HultCohort2026 #BuildingInPublic #AITools #IndieHackers #NextJS #Supabase

Shipped: Comentiq is live in production — and it ran its own campaign to prove it. The full loop works end-to-end: ↳ Builder submits a progress update ↳ Claude drafts channel-specific campaign content, grounded only in evidence they supplied ↳ Builder approves before anything goes public ↳ Peers boost it. Partners find it. What actually made it hard: • An RLS policy silently dropping amplification content from the UI — found by clicking the unhappy path, not reading the logs • A week of commits on the wrong git branch — recovered by an untracked-files check • Three Vercel misconfigs stacked on each other causing production 404s Stack: Next.js 14 · Supabase · Claude · TypeScript · Zod Built for Hult Summer Cohort 2026. Public showcase is live. Partner enquiries open. → hult-cohort-program-one.vercel.app #HultCohort2026 #BuildingInPublic #AITools

Partner summary

Comentiq is a production-ready platform built within the Hult Summer Cohort 2026 that converts builders' weekly progress updates into evidence-grounded, AI-drafted campaign content — reviewed and approved by the builder before publication. The platform is fully operational, with a verified end-to-end loop covering user authentication, project management, AI content generation (via Anthropic's Claude), mandatory human approval, peer amplification, and a partner-facing public showcase with a direct enquiry pipeline. Built on Next.js 14, TypeScript, Supabase, and Tailwind CSS, with Row-Level Security enforced across all data tables. The codebase is public on GitHub. Relevant to partners who are: hiring engineers with demonstrated full-stack and AI integration experience; seeking pilot opportunities with cohort builders and their projects; or interested in embeddable content infrastructure for their own community or accelerator programs. Live platform: hult-cohort-program-one.vercel.app Public codebase: github.com/Lorra-V/hult-cohort-program Enquiries via the partner pipeline on the platform, or direct outreach to the builder.

Story angle

The judgment gap: why AI's real bottleneck is human discernment, not machine output

As AI floods workplaces with generated answers and plans, the scarce resource is no longer information — it is the capacity to evaluate, question, and take responsibility for what AI produces. This project names that gap precisely and builds a structured response to it. The angle is credible because it is grounded in a real, observable shift in how knowledge work fails, not in hype about AI capability.

Here is a problem worth naming precisely. AI now generates answers, plans, and content faster than most people can evaluate whether those outputs deserve trust. The bottleneck in AI-assisted work is no longer production — it is judgment. That is the premise behind The Effective Facilitator, a project I am building as part of the Hult Summer Cohort 2026. It is an assessment-led learning platform built around nine disciplines: the capacities people need to direct, question, and take accountability for AI-generated work — rather than defaulting to whatever a model hands them. The disciplines address how we respond to influence, uncertainty, responsibility, and control when a machine is doing the heavy lifting. This is not another AI literacy course about prompting or tool selection. It is about what happens inside the decision-maker — the facilitator, the manager, the team leader — when AI is in the room and speed creates pressure to accept rather than think. The platform is live in early form at tef-deploy.vercel.app, built on Next.js, Supabase, and Vercel. Right now I am looking for: → A pilot organization willing to run the programme with a real team → A research partner interested in human agency and AI-mediated work → Early users who make decisions with AI daily → A technical collaborator to develop the platform further If any of those descriptions fit you or someone you know, I would genuinely welcome a conversation. #HultSummerCohort2026 #TheEffectiveFacilitator #AILiteracy #WorkforceDevelopment #HumanJudgment #ResponsibleAI

AI generates answers faster than most people can judge whether those answers deserve trust. That gap — between production speed and human discernment — is what The Effective Facilitator is built to close. It's an assessment-led platform with nine disciplines for people who make decisions with AI: facilitators, managers, team leads who need to direct and question machine output rather than default to it. The platform is live (early stage). Built on Next.js + Supabase. Looking for: → Pilot organization → Research partner → Early users → Technical collaborator Building this inside @HultPrizeCohort Summer 2026. If you work in L&D, run a team using AI tools, or study human agency in AI contexts — let's talk. tef-deploy.vercel.app #TheEffectiveFacilitator #HumanJudgment #AI #WorkforceDevelopment

Partner summary

The Effective Facilitator is an assessment-led learning platform designed to develop human judgment in AI-mediated work. Built on nine structured disciplines — covering how people respond to influence, uncertainty, responsibility, and control when using AI — it serves facilitators, team leaders, and managers who make consequential decisions with AI-generated outputs. The platform is in early-stage development within the Hult Summer Cohort 2026, with a live deployment at tef-deploy.vercel.app and an open-source repository on GitHub. The technology stack includes Next.js, Supabase, Vercel, custom JWT authentication, and Railway. The project is actively seeking four types of partners: 1. Pilot organization — to run the programme with a real team and provide structured feedback on the nine disciplines in practice. 2. Research partner — to co-investigate human agency, AI literacy, and decision-making quality in AI-assisted environments. 3. Early users — individuals who make daily decisions with AI tools and want to develop more deliberate judgment. 4. Technical collaborator — to advance platform development alongside the founder. This is an early-stage project with a clearly defined problem, a structured curriculum architecture, and working infrastructure. It sits at the intersection of AI, workforce development, and human-centered learning design. Partners engaging now will shape the programme's direction and methodology.

Lorraine Villaroel

About

Published projects

The Effective Facilitator

Conexus

RoomReserve

Mission Control

Comentiq

Approved campaign stories

Building the tool the cohort actually needs: a project manager that earns engagement through real accountability, not hollow gamification

A builder shipped the platform that runs its own campaign — Comentiq is live in production, evidence-grounded, and demonstrating its own loop in real time.

The judgment gap: why AI's real bottleneck is human discernment, not machine output