JTBD Positioning Analysis — gblock-party
Skill: /positioning (Step 1 of 4: JTBD) · Product: gblock-party · Date: 2026-06-01 · Mode: Alignment Review
Sources: research/icp.md, research/competitive-analysis.md, research/journey-map.md
Primary Job Statement
“When I’m running multiple AI coding agents across projects and need to step away from my desk, I want my agents to keep working on infrastructure that’s managed for me, that I can monitor and control from any device, so I can ship more code without being chained to a laptop.”
Job Structure
| Element | Content | Evidence Source |
|---|---|---|
| Situation | Running 3–8 AI coding agents across 1–3 repos on remote infrastructure | ICP: “Starts 3-8 AI agent sessions across 1-3 repos” |
| Motivation | Need to step away from desk without losing agent state or visibility | ICP: “Close the laptop, pick up where I left off” (#1 stated desire) |
| Expected Outcome | Agents keep working on persistent infrastructure, controllable from any device | Journey Map: aha moment = “closed laptop, checked from phone, 3 PRs ready” |
| Desired Progress | Ship at team velocity as a solo developer | ICP: “Like having a co-founder who never sleeps” |
Gate 1: Job Statement Validation
Does this job statement accurately capture the core “job” the Agent Conductor is hiring gblock-party to do?
Functional Dimensions
What the product must do to fulfil the job. Each dimension is grounded in ICP pain points, competitive gaps, or journey evidence.
| Dimension | Job Requirement | Evidence | Competitive Gap |
|---|---|---|---|
| Session persistence across devices | Agents survive laptop close; resume from any device without state loss | ICP #1 stated desire: “Close the laptop, pick up where I left off.” Journey Map Critical Moment #2: persistence proof. | Every first-party mobile tool (Claude Remote, Codex Mobile, Copilot Remote) dies when laptop sleeps. Grass solves persistence but locks to Daytona cloud UI (no BYO-client), is iOS-only, and early-stage — gblock-party’s persistence is unique because it pairs managed infra with BYO-client flexibility. Unique |
| Unified multi-agent status visibility | One view across all agents — not 8 tmux panes | ICP: “See all my agents in one view” — second most-stated value driver. Pain Map: “No unified multi-agent status view” rated Critical. | Desktop ADEs (Emdash, T3 Code) offer this locally but not persistently across devices. No remote-first unified view exists. Unique |
| Mobile approval routing | Approve or reject risky agent actions from phone in <30 seconds | ICP: “Approve from my phone.” Journey Map Journey 2: “Can I do useful work from my phone in under 30 seconds?” 62% of Copilot Remote approvals happen from notification banners. | Copilot Remote has push approvals but single-agent only. Nimbalyst has iOS app but desktop-dependent. No agent-agnostic mobile approval routing on managed infra. Unique |
| Managed infrastructure | Zero DIY VPS sysadmin — turnkey server provisioning and maintenance | ICP: “Zero setup, just connect.” Current State Journey: 10-step DIY process (provision VPS, install CLIs, set up tmux, configure Tailscale…). Pain Map: “DIY sysadmin overhead” rated Medium. | Coder is enterprise-only ($173M funded). VPS tutorials are 5-10 step guides. Nobody owns solo-dev managed agent infra. Differentiated |
| BYO-client flexibility | Use existing front-end tools (T3 Code, Multica, Claude Code CLI) — not forced into a new UI | ICP: “BYO-client is a must-have.” Deal-breaker: “Vendor lock-in.” Competitive Analysis: front-end layer is commoditizing (5+ OSS tools). | Cursor, Codex, Copilot all lock to vendor UI. Grass locks to Daytona cloud UI. Only Emdash has SSH remote but no persistent daemon. Unique |
| Zero-inbound security | No public ports on the server — all connectivity via outbound tunnels (Tailscale, Cloudflare Tunnel) | ICP: Deal-breaker: “Required public inbound ports.” Competitive Analysis: “Zero-inbound is table stakes” for VPS workflows. | Every VPS tutorial teaches zero-inbound via Tailscale. The differentiation is orchestration UX on top of zero-inbound, not zero-inbound itself. Table stakes |
Emotional Dimensions
How the product makes the user feel — the emotional states it creates, eliminates, or transforms.
| Emotion | Before (Current State) | After (With gblock-party) | Evidence |
|---|---|---|---|
| Relief from desk-tethering | Anxiety that agents will stall, need approval, or lose state if the user steps away. Chained to the laptop. | Confidence to close the laptop knowing you can stop, adjust, or approve from anywhere, anytime. The relief isn’t fire-and-forget — it’s control-at-a-distance. Agents work unsupervised, but you’re always one push notification away from intervening. | ICP #1 desire: “Close the laptop, pick up where I left off.” Journey Map: laptop close is the aha moment trigger. |
| Confidence in unsupervised agents | Worry that agents will do something destructive unsupervised. 93% auto-approve rate = security theater. | Trust that risky actions route to mobile for approval; safe actions proceed automatically. Tiered risk classification. | Competitive Analysis: 93% auto-approve rate. Gap 4: “Smart Approval Routing.” yoloAI exists to bypass approval friction. |
| Delight at the aha moment | No equivalent experience. Agents require continuous babysitting at the desk. | “I started 5 agents, closed my laptop, had dinner, checked from my phone — 3 PRs were ready for review. Opened my other laptop the next morning and it felt like I’d never left.” The aha isn’t just persistence — it’s seamless continuity across control planes (laptop → phone → different laptop), where each device picks up exactly where you left off with no re-authentication, no lost state, no context gap. | ICP: core aha moment quote. Journey Map: Critical Moment #2 (persistence proof) + #3 (phone proof). Cross-device continuity extends the aha from single-session persistence to full control-plane portability. |
| Frustration eliminated | tmux pane cycling (“blind cycling through 8+ identical panes”), SSH on phone (“universally painful”), 30-50 lines of tmux config. | Unified status view replaces pane cycling. PWA replaces SSH on phone. Zero-config managed infra replaces tmux scripts. | ICP Pain Map: “No unified multi-agent status view” (Critical). Competitive Analysis: “Terrible on touchscreen” re tmux. ICP: SSH on phone “intolerable.” |
Desired Outcome State
The desired outcome: AI coding agents run 24/7 on managed infrastructure. Status is visible from any device — laptop, phone, tablet. Risky approvals route to mobile in <30 seconds. A solo developer ships at team velocity without being chained to a desk, without managing VPS infrastructure, and without sacrificing security.
Gate 2: Dimensional Weighting
Are the functional dimensions correctly weighted? The analysis ranks session persistence, unified visibility, and mobile approval routing as the three highest-priority functional dimensions, with managed infrastructure and BYO-client as strong differentiators and zero-inbound security as table stakes.
Positioning Around Job Outcome
Lead with the outcome, not the features. The product is not “a dashboard” or “an agent orchestrator” — it’s the reason your agents survive laptop close. The headline is “your agents never stop,” not “monitor agents from a web UI.”
Winning Dimensions vs. Competitor Categories
| Dimension | tmux + SSH + VPS (DIY) | Desktop ADEs (Emdash, T3, Air) | Mobile Apps (Copilot Remote, AgentsRoom) | Cloud Agents (Cursor, Codex) | gblock-party |
|---|---|---|---|---|---|
| Session persistence | Partial (tmux survives SSH drop, not laptop sleep without VPS) | No | No (host must stay on) | Yes (vendor lock-in) | Yes (managed, BYO-client) |
| Unified multi-agent view | No (tmux panes) | Yes (local only) | Partial (AgentsRoom multi-agent) | No (single-agent) | Yes (remote, any device) |
| Mobile approvals | No | No | Yes (single-agent only) | No | Yes (multi-agent, risk-tiered) |
| BYO-client | Yes (full control) | No (vendor UI) | No (vendor UI) | No (vendor UI) | Yes (API-first) |
| Managed infrastructure | No (full DIY) | No (local) | No | Yes (vendor cloud) | Yes (turnkey) |
| Zero-inbound security | DIY (Tailscale manual) | N/A (local) | Varies | N/A (vendor cloud) | Yes (built-in) |
Positioning Implication
gblock-party wins on session persistence + unified visibility + mobile approvals + BYO-client simultaneously. No competitor covers all four. The closest competitor by dimension is DIY (VPS + tmux), which has BYO-client and partial persistence but no unified view, no mobile approvals, and no managed infrastructure.
The positioning formula: “Your agents never stop. Monitor from any device. Approve from your phone. Use your own tools.”
The anti-positioning: “Not competing on ‘terminal in browser.’ Competing on ‘orchestrate 10 agents safely from your phone, with your favorite tools, on managed infrastructure.’” — from competitive analysis signals.
Alternatives Considered
Alternative job framings considered and why they were rejected in favor of the primary job statement.
Rejected: “I want a better terminal multiplexer”
Frames the job as improving tmux, not replacing the workflow. Leads to building a TUI (Claude Squad already exists, 5,800 stars). Misses the mobile, persistence, and managed infrastructure dimensions entirely. The user isn’t hiring a better tmux — they’re hiring a way to not need tmux.
Rejected: “I want a web-based IDE for AI agents”
Frames the product as a CDE/IDE competitor (Codespaces, Replit, Kiro). Front-end UI is commoditizing — 5+ OSS tools already exist. The job is not “give me a web editor” but “keep my agents running when I walk away.” Competitive analysis explicitly warns: “Not competing on ‘I can open a terminal in a browser.’”
Rejected: “I want to run more agents in parallel”
Focuses on agent count/throughput as the primary job. Evidence shows 3-5 agents is the sweet spot before coordination overhead exceeds gains. The bottleneck is review speed, not agent count. The real job is managing and monitoring the agents you already have, not spawning more.
Rejected: “I want enterprise-grade agent governance”
Frames the product for the Security-Conscious ICP (ranked 4th of 4 in ICP prioritization). Enterprise governance is a later-stage expansion, not the primary job for the solo founder who is the launch ICP. Coder ($173M funded) and Ona own this space.
Evidence Matrix
| Claim | Source | Evidence | Inference | Confidence |
|---|---|---|---|---|
| “Close the laptop, pick up where I left off” is #1 desire | research/icp.md | Stated unprompted across VPS+agent articles and ICP interviews; VPS tutorial ecosystem opens with this promise | Session persistence is the job’s functional anchor | High |
| “See all my agents in one view” is #2 value driver | research/icp.md | Every tmux complaint circles back to multi-pane management; Pain Map rates “No unified status view” as Critical | Unified visibility is the second functional anchor | High |
| “Approve from my phone” is a top-3 value driver | research/icp.md, research/competitive-analysis.md | ICP states this directly; Copilot Remote GA with push notifications validates pattern; 62% approve from banners | Mobile approvals are the conversion trigger vs. free DIY | High |
| BYO-client is a must-have / vendor lock-in is a deal-breaker | research/icp.md | ICP Deal-breakers list; front-end commoditization analysis (5+ OSS tools); T3 Code, Emdash, Claude Squad are free | Product must be infrastructure-first, not UI-first | High |
| Aha moment = “closed laptop, checked from phone, 3 PRs ready” | research/icp.md, research/journey-map.md | Exact quote used by ICP research as the aha moment; Journey Map grounds it in Critical Moments #2 and #3 | The job is complete when persistence + mobile check-in both work | High |
| No product combines all 5 key dimensions | research/competitive-analysis.md | Feature matrix of 27 competitors; each covers 2-3 of 5 (BYO-client, managed infra, multi-agent, mobile, zero-inbound) | The combination is the positioning, not any single feature | High |
| Solo dev ships at team velocity with parallel agents | research/icp.md | “Like having a co-founder who never sleeps”; “maximize parallel AI agent throughput” | Velocity is the emotional payoff, not a functional claim | Medium |
| 93% auto-approve rate = approval fatigue | research/competitive-analysis.md | Claude Code auto mode data; yoloAI project exists to bypass friction; Molten.Bot analysis | Smart approval routing (not just push notifications) is needed | Medium |
| “Agent Conductor” role as identity upgrade | research/icp.md, research/competitive-analysis.md | ICP persona named “Agent Conductor”; Pragmatic Engineer: “only senior+ engineers succeed”; role shift coder → conductor | Social dimension is real but secondary to functional | Medium |
| Price sensitivity: must be under $30/mo vs. free DIY | research/icp.md | ICP budget analysis; tmux/T3 Code/Emdash are free; $9-29/mo sweet spot stated | Pricing must be justified by outcomes DIY can’t deliver | High |
Gate 3: Evidence Sufficiency
Is the evidence base sufficient to support this JTBD analysis, or are there critical gaps that need additional research before proceeding?
Confidence & Assumption Register
| Item | Status | Confidence | What Would Change It |
|---|---|---|---|
| Job statement captures primary hiring reason | Evidence-backed | High | User interviews showing a different primary job (e.g., cost savings, not mobility) |
| Persistence + mobile are the top two functional dimensions | Evidence-backed | High | Users ranking a different dimension (e.g., managed infra) higher than persistence |
| BYO-client is a must-have, not a nice-to-have | Evidence-backed | High | Users willing to abandon their tools for a superior integrated UI |
| Social dimensions (Agent Conductor identity) matter to the ICP | Provisional | Medium | User interviews showing identity/status is irrelevant to purchase decision |
| Emotional relief from desk-tethering is a real purchase driver | Provisional | Medium | Users describing the problem as purely functional with no emotional weight |
| The combination of dimensions (not any single one) is the positioning | Evidence-backed | High | A competitor ships all 5 dimensions simultaneously |
| Price must be under $30/mo to compete with free DIY | Evidence-backed | High | Willingness-to-pay testing showing higher price tolerance |
Proposed File Changes
File: research/positioning-jtbd.md
Upon approval, the JTBD analysis content from this alignment page will be written to research/positioning-jtbd.md. This file will contain:
- Primary job statement with structured elements (situation, motivation, outcome, progress)
- Functional dimensions table (6 dimensions with evidence and competitive gaps)
- Social dimensions table (2 dimensions)
- Emotional dimensions table (4 before/after states)
- Desired outcome state and metrics
- Positioning formula and anti-positioning
- Winning dimensions vs. competitor categories matrix
- Alternatives considered (4 rejected framings)
- Evidence matrix and confidence register
Gate 4: Proposed File Changes
Approve writing the JTBD analysis to research/positioning-jtbd.md?
Compile Review
Use Compile Feedback to send concerns or clarification requests before answering every gate, or Compile Answers for final approval once all required gate questions are answered. The compiled YAML always includes any optional section feedback you set.
Social Dimensions
How using the product changes how others perceive the user and how the user perceives their professional identity.