| topic | identity | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| type | guide | ||||||||||||
| status | research-complete | ||||||||||||
| last-validated | 2026-05-21 | ||||||||||||
| original-query | How to use Bonfire as Zaal's personal second brain + Obsidian alternative for BCZ and ZAO data (reconstructed) | ||||||||||||
| tier | DEEP | ||||||||||||
| related-docs |
|
Goal: Design how Zaal Panthaki uses Bonfire as a personal second-brain alternative to Obsidian, ingesting BCZ + ZAO + personal life data so Bonfire becomes a useful daily companion for memory, decision-making, and synthesis.
Zaal should use Bonfire instead of Obsidian as his personal memory system, with conditional go-live on pricing (<$300/mo) and rate limits (10+ req/sec, resolve by May 5 via email to Joshua.eth).
Bonfire wins on 3 counts:
- Telegram-native capture (zero friction) - 15 seconds to add note vs 2-5 mins for Obsidian
- AI synthesis - "@bonfires what themes from my reading?" vs manual backlinks in Obsidian
- Agent integration - Hermes/ZOE can query Bonfire natively (MCP proven)
Bonfire loses on 1 count:
- No offline mode - SaaS only (vs Obsidian's local-first). Mitigant: Apple Notes buffer when offline.
7 Personal Data Domains to Ingest:
- ZAO ecosystem (40-60 entities) - weekly refresh
- BCZ consulting (15-25 projects + clients) - event-driven
- Music life (artists, releases, venues) - monthly
- Personal (Ellsworth, family, friends, health) - weekly
- Reading + ideas (articles, podcasts, half-formed thoughts) - continuous
- Decisions + reversals (major choices with rationale) - weekly
- Inbox + capture (links, screenshots, mentions) - real-time
- One entry per day (5-min commitment)
- Telegram to @bonfires: "Journal: [date]. Intent: [what I want to accomplish]. Decision: [what I decided]. Reflection: [what I learned]."
- Bot extracts entities (projects, people, decisions), stores in Daily Journal kEngram
- Query day 7: "@bonfires show me my journal entries about ZAOstock" → traverses graph, returns synthesis
Habit: 5:00-5:05am (morning power sprint per user_zaal_schedule) or evening 6:45pm
- Seed day 1: "Zaal Panthaki. Founder of The ZAO, owner of BCZ Strategies, musician + curator. Location: Ellsworth Maine. Values: decentralization, community-driven, build-in-public. Work: 4:30am-7pm M-F with power sprints 5-5:45am + 4-7pm."
- Bot creates PERSON node with role edges (founder, owner, curator)
- Query: "@bonfires what roles do I have?" → returns The ZAO founder, BCZ owner, music curator
- Seed day 1: ZAOstock 2026 (Oct 3 deadline, $5-25K budget, artist apps close ~Sept 1), BCZ Music Entity (legal + contracts), Cipher Release (Q2 2026 target)
- Bot tracks status, milestones, blockers, decisions, reversals for each
- Query: "@bonfires what are my active projects? Show deadlines & blockers." → returns 3 projects + next milestones
Habit: Update on status change. Check weekly for blockers.
- Per-person nodes: Roddy (venue contact, relationship_health: POSITIVE), Steve Peer (ZAO Stock co-curator, UNKNOWN), Iman (collaborator on Cipher, STRONG)
- Attributes: last_contact, next_step, shared_interests, relationship_health
- Query: "@bonfires who should I reach out to this week?" → people with UNKNOWN status + action items
- Articles read (title, source, key insight, date, relevance tag)
- Podcasts (guest, episode, takeaway, tag)
- Ideas (half-formed thought, status: brainstorm/exploring/paused, source)
- Query: "@bonfires what themes am I exploring?" → returns clusters of related articles + ideas
- Major decision: statement + date + rationale + alternatives + confidence (0.6-1.0) + reversible flag + status
- Example: "ZAOstock lineup cutoff Sept 1 / Rationale: 4-week buffer / Alternatives: Aug 15 (conflicts with BCZ season) / Confidence: HIGH / Reversible: YES / Status: ACTIVE"
- Query: "@bonfires what decisions have I made in April?" → returns 8-12 decisions
Habit: Capture immediately after decision (1-2 min). Friday retro: review reversals.
- Phase 2 (May 20+): bulk-load 20 high-signal docs (271, 432, 542-548, 523, 524, 531, 539, 541)
- Confidence flags: 1.0 (authoritative like doc 432), 0.7-0.8 (aspirational like 506)
- Query: "@bonfires what do our research docs say about agent autonomy?" → searches 20 docs, returns synthesis + citations
- Friday 6:45pm: "@bonfires what are my wins this week?" / "@bonfires what blockers do I need to address?" / "@bonfires what themes emerged in my reading?"
- Capture retro entry: wins, blockers, themes, decisions made, people to follow up
- Habit: 40 min, non-negotiable, pairs with ZAO team meeting Tuesday 10am or personal ritual
Morning (5:00-5:45am):
- 5:00-5:05: Daily journal intent to @bonfires (1 message)
- 5:05-5:30: Inbox clear (0-3 messages: screenshots, articles, person mentions)
- 5:30-5:45: Check one query ("@bonfires what's my priority today?")
Midday (11:30am-12:30pm):
- Ad-hoc capture (if meeting/call, 1-2 messages within 1 hour: "Talked to [person] about [topic]. Outcomes: [X]. Decision: [Y].")
Evening (4:00-7:00pm):
- 4:00-4:15: Post-work retro (if major decision/outcome)
- 6:45pm Friday (40 min): Weekly synthesis + retro
Time investment: ~10 mins daily, 40 mins Friday. Return: Clear head, zero forgotten tasks, weekly patterns visible.
| Aspect | Obsidian | Bonfire | Winner |
|---|---|---|---|
| Capture friction | Open app (30s), type note | Telegram DM (3s) | BONFIRE |
| Mobile capture | Web Clipper (web only) | Telegram native (instant) | BONFIRE |
| Offline mode | YES (local) | NO (SaaS) | OBSIDIAN |
| Search | Backlinks (manual labor) | "@bonfires what do I know about X?" (AI synthesis) | BONFIRE |
| Agent integration | No native (custom MCP hard) | MCP native + REST API | BONFIRE |
| Setup time | 6-10 hours | 30 minutes | BONFIRE |
| Cost | Free + $50/year | TBD ($200-1000/mo est.) | TBD |
| Data ownership | Local (yours) | SaaS (Bonfire/Weaviate) | OBSIDIAN |
3 Wins:
- Telegram-native capture - Zaal already there, zero friction
- AI synthesis - "@bonfires what themes?" beats manual backlinks 10x
- Agent context - Hermes/ZOE can query Bonfire for smarter context (proven MCP per doc 546)
1 Loss:
- No offline mode - Real gap for travel. Mitigant: Apple Notes buffer, sync when online (Phase 2 could add local cache)
- ZAO Ecosystem (40-60 entities, weekly refresh, public) - Already in scope (doc 545 ontology)
- BCZ Personal Brand (15-25 projects, event-driven, semi-private) - CRM use case, Bonfire stores client status + outcomes
- Music Life (25-40 entities, monthly, public) - Artists, releases, venues, collabs. Queryable for A&R decisions
- Personal Life (20-30 entities, weekly-monthly, private) - Family, friends, Ellsworth context, health, goals
- Reading + Ideas (40-60+ continuous, mixed) - Articles, podcasts, half-formed ideas. Bonfire clusters themes
- Decisions + Reflections (15-25 weekly, private-to-public) - Decision log + reversals. Bonfire tracks rationale + confidence
- Inbox + Capture (5-20/day, mixed, real-time) - Links, screenshots, mentions. Bonfire auto-extracts from Telegram
Hit 5/7 to extend to Phase 1; hit 3/7 or fewer to pivot to Obsidian/Neo4j.
- Habit adherence: 95%+ (20/21 morning journals, 4/4 Friday retros)
- Query accuracy: 80%+ correct answers (no hallucination)
- Synthesis quality: Friday retro identifies real themes (not generic)
- Capture velocity: 4+ new entities per day average (120+ total/month)
- Agent integration: Hermes queries Bonfire <3s latency, gets relevant context
- Pricing acceptable: <$300/mo or aligned with budget
- No data loss: Zero crashes, lost entries, data corruption
- Pricing: What's the monthly cost for Genesis tier? If >$500/mo, OSS pivot.
- Rate limits: How many API requests/sec/min? Critical for Hermes (5+ queries/day).
- Export/backup: Can you export knowledge graph if Bonfire shuts down? Escape hatch?
- Document loader: Does Bonfire support ingesting Markdown + PDFs natively (doc 544)? (Phase 2 depends on this)
- Voice capture roadmap: Tentative date for iOS? (Helps unlock gym capture, Phase 2 nice-to-have)
- Confidence scoring: Does API expose confidence attributes + source_kind in queries? (Need to distinguish "Zaal said" 1.0 vs "inferred" 0.7)
- Telegram rate limits: Can @bonfires handle 100+ messages/day if whole team uses it? Per-user or global?
- Day 1 (April 29): Email Joshua.eth (pricing/rate limits/questions). Seed 5 identity entities (Zaal, The ZAO, BCZ, ZAOstock, Franklin St Parklet). Seed 3 active projects. Load alias dedup dictionary (WaveWarZ, The ZAO, ZABAL, BetterCallZaal, etc.)
- Days 2-3: Add 10 core entities (Roddy, Iman, Steve Peer, sponsors, timeline milestones). Test 3 queries.
- Days 4-7: Daily journal (establish 5-min morning habit). Add 2-3 entities/day. Friday retro (40 min).
- End of week 1: Bonfire has 25-30 entities, schema validated. Decision: extend to Phase 1 or pivot?
| Risk | Level | Mitigant |
|---|---|---|
| Pricing TBD | HIGH | Email Joshua.eth by May 5. Budget <$300/mo or pivot to Neo4j. |
| Rate limits unknown | MEDIUM | Ask for limits. If <2 req/sec, Hermes risky. If 10+, good. |
| Accuracy unproven | MEDIUM | 30-day trial. If hallucinations >20%, Obsidian safer. |
| No offline mode | LOW | Not critical (80% online). Workaround: Apple Notes + sync. Phase 2: local cache. |
| SaaS lock-in | MEDIUM | Acceptable vs benefits. Flag export policy for Phase 2. |
| Habit formation | MEDIUM | Depends on Zaal's discipline. Ritual designed around his schedule (5am + Fri 6:45pm). |
| Data loss | LOW | 35+ live deployments (proven). Monitor week 1. |
| # | Action | Owner | By | Type |
|---|---|---|---|---|
| 1 | Email Joshua.eth (pricing, rate limits, export, document loader, voice roadmap, confidence scoring, Telegram limits) | Zaal | May 5 | BLOCKER |
| 2 | Seed 5 founding entities in Bonfire UI | Agent/Zaal | April 29 EOD | GO/NO-GO |
| 3 | Establish morning journal + Friday retro ritual | Zaal | May 1 + May 3 | HABIT |
| 4 | Week 1 retro: query accuracy + velocity + habit | Agent/Zaal | May 5 | EVAL |
| 5 | Decide: Phase 1 commit or pivot to Neo4j/Obsidian | Zaal | May 5-6 | DECISION |
| 6 | If Phase 1: bulk-load 5 high-signal research docs | Agent | May 20 | EXECUTION |
| 7 | If Phase 1: wire Hermes to Bonfire MCP | Agent | May 20-27 | INTEGRATION |
| 8 | 30-day evaluation (June 5): success metrics + Phase 2 decision | Zaal | June 5 | GO/NO-GO |
- Obsidian PARA: Tiago Forte. "The PARA Method." fortelabs.co/para (updated April 15 2026)
- AI Second Brain: Copana. "Best AI Second Brain Tools 2026." copana.ai/blog (Feb 7 2026)
- GraphRAG: Microsoft Research. microsoft.github.io/graphrag (updated April 13 2026)
- Bonfire: Docs 542-548 (April 28-29 2026)
- Zaal's schedule: memory/user_zaal_schedule.md
Recommendation: YES, commit to Phase 0 (week 1) pending Joshua.eth confirmation on pricing <$300/mo + rate limits 10+ req/sec. Non-negotiable: Zaal must commit to daily ritual (10 min morning + 40 min Friday).