Sophia — Product Overview
As of: 2026-09-22 · Spec: PRODUCT_CONTEXT.md · Tracker: implementation-tracker
What is Sophia?
Sophia (Sofia) is a standalone, public, self-service AI interview SaaS: signup, prepaid interview-minute credits (Stripe), Positions, Candidates, private links, English voice interview via shared Pipecat bot + Daily, scored evidence-backed report with honest assessed flags. Same engine as CleverHire, sold without an ATS.
Target market and users
| Segment | Role | Phase 1 focus |
|---|---|---|
| SMB recruiters / hiring teams | Primary buyers | Not enterprise; ATS deferred |
| Workspace admin | Billing, team, all interviews | Creator starts as admin |
| Workspace member | Own interviews only | No private billing |
| Candidate | Private link | No account |
| Highbrow operator | Internal recovery/audit | Not customer-visible |
Pricing model
- Unit: 1 credit = 1 interview minute (integer credit-seconds in ledger).
- Packs (USD): 50 / $19 · 250 / $79 · 1,000 / $299.
- Billing: Reserve max duration at schedule; charge actual connected seconds at settlement.
- Release gate: Blended variable cost ≤ $0.15 per paid minute (or signed exception).
- Phase 2: Subscriptions, auto top-up, enhanced tiers.
Phase 1 scope (target 7 October 2026)
Standalone SaaS (AC16); signup + admin/member; positions, candidates, résumés; day-window + email/ICS; candidate lobby (screen share, live AI); credits/Stripe; results + assessed; FR19 proctoring/credibility; retention, sweepers, 10/3 concurrency, 100-session qual; contract docs; minimal public pages.
~30–40% built. Gaps: email/ICS, FR19, report UI/PDF, sweepers, notifications, ops console, load/cost qual, AC/WAL/PRO.
Out of scope (Phase 1)
SSO; Apple/Microsoft sign-in; candidate accounts; mobile/coding/multilingual; guest scores; subscriptions; white-label; CleverHire migration/adapter; ATS; calendar; templates; phone QR; lockdown browser; gov ID.
Release gate
- AC01–AC21 after re-baseline (calendar/templates/bulk dropped; PDF AC06–AC08 stale).
- Zero critical/high defects; no billing/isolation/data-loss defects.
- Blended cost ≤ $0.15/paid minute (or Kevin-signed exception).
- Kevin release decision on tested build.
Key decisions (selected)
| Decision | Choice | Notes |
|---|---|---|
| D1 | Standalone SaaS | AC16 independence hard gate. |
| D2 | Shared bot, not forked | Coupling risk to CleverHire. |
| D3 | Single MongoDB, org-scoped | — |
| D5–D6 | Reserve max, charge actual | Billing trust model. |
| D7 | Snapshot at schedule | Reports stable if position edits. |
| D10 | assessed per skill | No silent missing criteria. |
| D13–D14 | Drop templates + calendar | Day-window + ICS remains. |
| D16–D17 | Credibility + full detector scope | Top delivery risk; not implemented. |
| D19 | Position required | — |
Identity: Spec Cognito; code Firebase Auth (Google + email).
CleverHire relationship
- Shared Pipecat bot (
cleverhire-bot), not forked. - Separate deployment, DB, customer journey; no CleverHire hybrid tenant port.
- Future: CleverHire as consumer of interview service (G5);
packages/must stay CleverHire-usable — strategy tension (§23.3). - No historical migration Phase 1.
Timeline — 7 October 2026
| Milestone | Planned | Status (Sep 2026) |
|---|---|---|
| G0 Kickoff | 9 Sep | Done |
| G0b Detector feasibility | 11 Sep | Critical — FR19 not productized |
| G1 Independent vertical slice | 15 Sep | Partial — bot E2E + email incomplete |
| G2 Feature freeze | 29 Sep | At risk — remaining-work.md |
| QA week | 30 Sep – 6 Oct | Not started |
| G3 Acceptance | 6 Oct | Not started |
| G4 Release | 7 Oct | Re-baseline + qualification evidence |
Capacity (§18.6): ~270 build-hours; 120 QA-hours final week.
See also: implementation-tracker.md, remaining-work.md, nfr-and-quality-gates.md.