AGI House

AGI House: Technical Ops & Builder‑Matching Upgrade

A focused build to raise sponsor ROI and builder quality

Presenter: Alexey Bauer — Product, Prompt Engineering & Ops Automation (part‑time)

1) Title — AGI House: Technical Ops & Builder‑Matching Upgrade

Subtitle: A focused build to raise sponsor ROI and builder quality

  • Presenter: Alexey Bauer — Product, Prompt Engineering & Ops Automation (part‑time)

Product — AI Agent Flow: Intake → Enrichment → Scoring → Matching

Zero‑friction intake. Agents fetch public signals, enrich, score and route to matching and nudging.

Applicant
LinkedIn · Twitter · GitHub · Website
→
Fetchers
Policy‑compliant crawlers + OAuth drops
→
Normalizers
Dedup · clean · structure
→
Enrichers
Repos · talks · patents · publications
→
Scoring
Seniority · Domain tags · Visibility
→
Vector DB
Embeddings + tags for fast recall
→
Matching
Co‑builder + sponsor roster fit
→
Nudger
Profile drafts · reminders · warm intros

Product — Personal Profile (auto‑drafted)

Step 1: link accounts → Step 2: AI fills your profile → Step 3: pick events.

Avatar
Alex Chen
Senior ML Engineer · Agents & Robotics
San Francisco · Ex‑DeepMind, Cruise
LinkedIn verified GitHub verified
Summary

Skills & Tags
AI agents enriching…
Kernel Wizards
Goal: ship an optimized kernel
Robovision X
Goal: deploy perception pipeline
SLM Turbo
Goal: launch on‑device SLM agent

This flow mirrors AGI House profile: link accounts → our agents fetch signals, write a draft, and unlock event applications aligned to your tags.

Product — Swipe‑to‑Match (co‑builder & sponsor)

YC‑style card with deep profile. Accept to DM, Skip to see next. Ranked by roster fit.

Avatar
About

1 / 5

Product — Teams & Applications

Track your applied teams, statuses, and actions.

Kernel Wizards
Agents · GPU Kernels · Sponsored
Interview scheduled
Robovision X
Robotics · Perception
Accepted
SLM Turbo
Small Language Models
Waitlist

Product — Sponsor Controls & Quotas

Define individual quotas by roles, apply deep filters and multipliers, and preview a transparent scoring model.

Event totals

Role quotas

Engineers
0/120
Founders
0/40
VCs/Investors
0/20
Media/Press
0/10
Corporate Reps
0/10
Potential Sponsors
0/10
Total targeted: 0 / 200 · Estimated teams: 67

Required roles

Seniority multipliers

Junior (0–3)
Mid (3–7)
Senior (7–12)
Staff/Principal (12+)
Executive/CTO

Developer types

Tag groups and weights

AI/ML
Languages
Frameworks
Infra/DevOps
Databases
Robotics/Hardware
Blockchain/Web3
Other

Stacks & tags

Affiliations

FAANG/MAANG ×
VC-backed ×
Reliable Tech Tier 1 ×
Reliable Tech Tier 2 ×
Reliable Tech Tier 3 ×

Open Source & Academic

Other filters

Caps: Affiliation product A ≤ 5.0; final score clamped 0–100.

Weights (0–100)

Seniority (S)
Domain Fit (D)
Visibility (V)
Engagement (E)
Skill Depth (K)
Role Fit (R)
Score = (wS*S + wD*D + wV*V + wE*E + wK*K + wR*R) × A × (1 + Other)
NameRolesScoreTop tags

Product — Roster & Teams Overview

Sponsors can see upcoming teams, statuses, and remaining applicants.

Kernel Wizards #1
Locked
Robovision X #3
Forming
SLM Turbo #2
Forming

Applicants (unassigned)

Priya Raman
Applied Scientist — SLMs
SLMsCUDARAG
Kenji Watanabe
Robotics Engineer
ROS2SLAM
María Alvarez
Full‑stack + Agents
AgentsTool‑use

2) Problem — Great top‑of‑funnel, weak build & sponsor outcomes

  • Sponsor mismatch: ~70% presenting teams skew students; sponsors (AWS/Trainium, robotics partners) need senior builders.
  • “Kickoff tourists”: Top engineers show up to openers, then don’t build—team formation is hard and time‑consuming.
  • Data is trapped: A homegrown CRM blocks segmentation, analytics, and re‑engagement.
  • Manual ops: Outreach, matching, and comms are ad‑hoc—limits scale and predictability.
  • Organizer blind spots: No unified view of pipeline, matches, or event‑level quotas.

3) Solution — The AGI House Builder Graph & Sponsor Engine

  • Zero‑friction intake with AI agent: parse LinkedIn/Twitter/GitHub into bio, skills, tags, interests; support voice‑captured ideas.
  • Scoring & tags: Seniority, visibility, engagement, founder record; domain tags (ML/SLM/Agents/Robotics/etc.).
  • Vector search & matching: YC‑style co‑builder matching + sponsor‑specific matching; ranked people and suggested teams.
  • Discord automation: Servers/channels, group DMs, and auto‑intros; staged match releases as event nears.
  • Sponsor→Quota→Roster: Agent ingests needs, composes target roster/quotas, runs outreach to fill expertise slots.
  • Organizer dashboard: Real‑time pipeline, announcements, tagging, and basic analytics.

4) Solution Internals — Stack

  • CRM: Migrate to HubSpot (source‑of‑truth). Intercom optional for in‑app nudges.
  • Data & services: Enrichment workers, scoring service, managed vector DB, matching service, event webhooks (→ Discord).
  • Comms: Email + SMS sequences, calendar invites, Discord roles/channels; AI nudger for incomplete profiles.
  • Privacy & compliance: Consent tracking, rate limits, audit logs; policy‑compliant LinkedIn (human‑in‑the‑loop).
  • Minimal friction: User flow stays as‑is; enrichment behind the scenes.

5) Targeted Outreach — From “SpaceX Robotics” to ready teams

  • Sponsor‑driven sourcing: Ranked lists from the Builder Graph + public signals.
  • LinkedIn + email + SMS: Policy‑compliant outreach; templates are human‑approved.
  • Nudges & loops: AI nudger proposes profile edits and partner suggestions to convert attendance → building.
  • Value‑prop personalization: Benefits tuned to tags/history (access, mentorship, hiring, prizes).

6) Team & Cadence — Small, senior, fast

Team (3 engineers total)

  • You (part‑time): Product lead, prompt engineering, ops automation, sponsor/internal workflows.
  • Senior Engineer A (FT): AI automation + enrichment/matching services (available now).
  • Senior Engineer B (FT): Backend/platform & CRM/data integration.

Cadence

  • MVP: CRM schema & import, intake agent, tags/scoring v0, Discord auto‑channels, manual matching UI.
  • Pilot: Sponsor→quota agent, outreach flows, dashboard v1, first pilot event on the new stack.
  • Scale: Scoring v2, analytics, organizer tooling v2, playbooks; expand to multiple sponsors.

Dependencies: access to current DB/Discord, CRM licenses, email/SMS providers, sponsor pipeline.

7) KPIs & Next Steps — Make sponsor ROI obvious

KPIs (first 8 weeks)

  • Quota fill rate ≥ 90% for sponsor‑requested roles (pre‑qualified).
  • Team formation: ≥ 30 presenting teams at AWS‑level events; ≥ 60% include 5+ yrs experience engineers.
  • Conversion funnel: Invite→Profile ≥ 40%; Profile→Matched ≥ 60%; Matched→Team ≥ 45%; Show ≥ 80%.
  • Sponsor & builder NPS ≥ 50; Repeat rate ≥ 30% of elite builders within 60 days.

Decision asks (today)

  • Approve scope and team (you part‑time + 2 FTE seniors).
  • Green‑light CRM (HubSpot) and providers (email/SMS, vector DB).
  • Pick pilot sponsor/theme; grant data/Discord access.
  • Establish weekly steering to review KPIs and ship.

8) Format A — Shipathon (Sponsor‑Paid, Shipped Deliverables)

Sponsor funds real features shipped during/just after the event; engineers on‑call to co‑build & unblock.

How it works

  • Sponsor engineers on schedule: bookable slots (AI agent) for office hours, API help.
  • Deliverable‑first brief: acceptance tests, APIs/SDKs, starter repos.
  • Jury & quotas: category quotas + cash bounties; winners meet working‑code gates.
  • Example theme: “Video‑Farm Shipathon” (e.g., V03, Higgsfeld + partner tools).
  • Multi‑sponsor friendly: parallel tracks (Agents, Robotics, SLM‑training, Video).

Systems we’ll ship

  • Engineer Scheduler (AI booking + Discord sync)
  • Deliverables Portal (PR links, demos, tests, juror scoring)
  • Payout rails (Stripe Connect), IP & work‑for‑hire templates

KPIs

  • # shipped features passing sponsor tests; integration rate (merged PRs ≤ 14 days)
  • Sponsor payout utilization ≥ 85%; builder earnings and elite return rate

9) Format B — VC “Requests for Startups” (Check‑Writing Hackathon)

VCs publish RFS categories and pre‑commit to initial checks for teams formed and validated during the hackathon.

How it works

  • RFS intake: themes, criteria, check count; live Q&A.
  • Cofounder matching: Builder Graph prioritizes mature founders & senior builders.
  • On‑site diligence: demos, pilots, LOIs reviewed by multi‑VC jury.
  • Outcome: X checks on‑site; winners get an AGI House Demo Day slot.

Systems we’ll ship

  • RFS Publishing (themes, criteria, commitments)
  • Selection & KYC flow (basic compliance)
  • Investor Portal (shortlists, judging, decisions)

KPIs

  • # checks written & time‑to‑term‑sheet; founder seniority mix
  • Follow‑on meetings within 30 days; media/social lift

10) Virality & Exclusivity Playbook

Goal

Drive hype, FOMO, and pipeline quality—without spam.

Tactics

  • Auto social cards (HF0‑style); optional “post to confirm” for elite tracks.
  • Scarcity: capped slots per track; visible waitlist and acceptance rates.
  • Referral lift: personalized invite codes for alumni/sponsors; track conversion.
  • Live moments: check signings, feature merges, sponsor engineer AMAs.
  • Afterglow: highlight paid shipments, checks, and offers within 7–14 days.

Guardrails

  • Respect platform policies (no auto‑posting); clear consent/opt‑out; assets for LinkedIn/X preference.

KPIs

  • Tagged posts & impressions; waitlist size & acceptance rate
  • Referral % of elite applicants; press mentions / creator shout‑outs
Star

Want this as a ready‑to‑present deck?

I can generate a clean PPTX with slide notes.