oll.in · scope · decision instrument

The MVP cut — what we can charge for

The prototype sells the full vision: Ollie finds the jobs, writes everything, keeps you visible, watches the market. We can't build — or charge for — all of it at once. This enumerates every feature and asks one question of each: is it needed to go live and take a stranger's franc? Tap any row to move it Ship → Trim → Defer; your cut persists. Once settled, we build the survivors for real and adjust the design to match.

01The principle

MVP is not "a smaller version of everything." It's the smallest real thing a stranger will pay for, shipped on infrastructure we already have. Everything else is roadmap — kept in the prototype as the promise, built once the money path is proven.

Charge for what the spines already do. Defer what needs a new engine.

Ship-vs-build — the one that matters here

The prototype is a mockup; none of it is live. The fastest stranger dollar is the CHF 29 headshot — foto-service is already scaffolded on Core, no login, one-time. The heavy, risky features (Ollie finds jobs, Career Radar, auto-send, Get-found posting) are net-new infrastructure with scraping / ToS / API friction. Building those before the first franc is exactly the trap. Lead the MVP with what's already real.

The buildability lens (what each feature leans on)

real now — Core (auth + billing), oll-model (Groq / Claude text-ops → CV, cover letter, prep), foto-service (the headshot). Reuse, not rebuild.

new engine — a Swiss job crawler + matcher, an always-on monitor, a social-posting integration, an email/portal auto-sender. None exist; each is a project.

02Every feature — settle the cut

Defaults below are the recommendation. Tap a row to cycle its verdict. Ship = build for v1 · Trim = ship a simpler honest version · Defer = roadmap, after revenue.

v1 surface Ship 0 Trim 0 Defer 0
Get in — onboarding & setup
First-run onboarding wizard
5-step setup. Trim: 2 steps — paste CV, name the role. Core
CV upload / paste
The raw material everything is tailored from. real now
Preferences — role, pay, hours, location
Only "the role you're after" is needed to prepare. Rest is for finding (deferred). Core
Connect Email
Needed only for real sending. integration
Connect LinkedIn / X
Only for Get-found posting. integration + ToS
The Photo — the first dollar
AI professional headshot · CHF 29, no login
One-time, guest checkout. The on-ramp. foto-service exists
Multiple styles / backgrounds
A few looks to pick from — cheap, part of foto-service. real now
Prepare the application — the writing engine
Tailored CV for a role you paste
You supply the role; Ollie re-orders + rewrites to fit. No scraping. oll-model
Cover letter drafting
Written from your real experience against the role. oll-model
Honest gap-flagging
Names where you don't fully fit — the trust differentiator. Cheap. oll-model
Interview prep
Likely questions from the posting. LLM-cheap; ship or trim to a later add-on. oll-model
Dossier assembly (photo + CV + letter)
One reviewable pack — ties Photo and Prepare together. real now
Find the roles — the automation vision
Ollie finds matching roles
Crawl + match Swiss boards. Hard, slow, ToS-exposed. new engine
Fit score (% match)
Meaningless without the matcher above. new engine
Career Radar — always-on watch
Background monitoring; depends on the finder. A subscription hook — later. new engine
Radar alerts (better role / pay opened)
Downstream of the watch. new engine
Salary benchmark vs market
Needs a pay-data source. Nice, not first-dollar. new data
Send & track
Approve & send
Trim: v1 = approve → download / email-to-self / copy. Real send is an integration. Core email later
"You make the final click" + never auto-apply
A stance, not code — the core positioning & trust. Free. copy
Application tracker (pipeline)
Only meaningful once sending + responses are real. needs send
Dashboard stats
Trim to a simple "prepared / downloaded" count; full stats need the pipeline. real now
Get found — inbound presence
LinkedIn / X post drafts
LLM-cheap to draft, but only useful with posting + accounts. integration
Publish posts (official sign-in)
API friction + ToS. integration
"Open to work" signal
Depends on the connected accounts. integration
Presence analytics (views / reach)
Needs platform data access. integration
Account & trust
Plan & billing
Must charge. Photo one-time + a Prepare pack / plan. Core billing
Magic-link auth (no passwords)
For the paid Prepare tier. Photo stays guest. Core auth
Automation & control guarantees
The honesty panel — positioning, free, keep. copy
Integrations management
Nothing to manage until Email / LinkedIn / X land. integration
Language EN / DE
Trim: EN v1, DE fast-follow. Content, not engine. copy

03The recommended slice

With the defaults above, the v1 you can actually build + charge for is two real products on the existing spines:

Product 1 — The Photo · CHF 29, one-time, no login

Guest checkout → AI headshot (a few styles) → delivered. This is foto-service on Core. It is the first stranger franc and needs no new engine.

Product 2 — Prepare my application · the writing engine

You paste a role + your CV → Ollie returns a tailored CV, a cover letter, and an honest gap note (interview prep optional), bundled as one reviewable dossier you approve and download. This is oll-model + Core auth/billing. No job-scraping, no auto-send — so no ToS or integration risk.

What that deliberately leaves out of v1

Deferred to post-revenue: Ollie finding jobs, fit scores, Career Radar + alerts, salary benchmark, real auto-send, the pipeline tracker, and the entire Get-found surface (drafts, posting, presence, LinkedIn/X). Each is a new engine or integration — built once the two products above are earning, in the order the market pulls.

04Design adjustments this implies

Once the cut is settled, the prototype gets trimmed to the shipped surface — the full vision stays as a clearly-labelled roadmap, not as UI promising things that don't run.

keep, simplified Onboarding → 2 steps (CV + the role). Drop Email/LinkedIn connect.

keep The Photo flow (foto-service, guest) and the Apply/Prepare flow (CV · cover letter · gap note · download).

keep, honest Dashboard → "your prepared applications" + the explainer & the "Ollie's on it" reel (they explain the value truthfully). The automation & control honesty panel stays.

remove from v1 UI Dashboard's Ollie found, Career Radar, Get found panels; the Applications tracker; Radar and Get-found screens → move behind a "on the roadmap" marker.

keep Settings → Photo + Prepare plan only; drop integrations until they exist.

The Bold/Clean/Warm switcher already rides the token system, so the trimmed screens re-skin for free — the design decision (which skin) and the scope decision (which features) stay independent.

05Go-live sequence

  1. Photo live first. Wire foto-service guest checkout on Core → CHF 29 → the first stranger franc. No design changes needed to start.
  2. Prepare live second. Ship the CV + cover-letter + gap-note flow on oll-model behind magic-link + a Prepare plan.
  3. Adjust the design. Trim the prototype screens to the two shipped products (above); label the rest roadmap.
  4. Expand on pull. Only after revenue, build the deferred engines in the order users ask for them — likely finding → tracker → radar → get-found.

06Decisions for you

Yes/no; my recommendation on each. Once you settle these (adjust the board above to taste), we build the survivors for real and I trim the design to match.

1 · Lead with the CHF 29 headshot as the first live product? Rec: yes — it's already scaffolded, no login, no new engine.
2 · Product 2 = "Prepare my application" (paste a role → CV + cover letter + gap note), no finding / no auto-send in v1? Rec: yes — real on oll-model, zero ToS risk.
3 · Defer the whole automation vision (finding, radar, get-found, auto-apply, integrations) to post-revenue? Rec: yes — each is a new engine; don't build pre-franc.
4 · v1 "send" = approve → download / email-to-self (not real auto-send)? Rec: yes — keeps the honest "you make the final click" true, no integration.
5 · Trim onboarding to 2 steps, EN-only, no account connects? Rec: yes
6 · Once settled, I trim the prototype screens to the shipped set (full vision → roadmap)? Rec: yes — design follows the settled scope, not before.
Next move: settle the board above + these six, then I wire foto-service checkout for the first franc and simplify the screens to match. Nothing gets built or redesigned until the cut is settled — we go live with real features once settled.