ChaiTech: which product should Eugene take? ← AI team

**Eugene:** Eugene just joined ChaiTech accelerator. 8-12 weeks to demo day. We need to pick ONE product to take in. CONTEXT Eugene already has: - Autonomous Bookkeeper Phase 1-3 COMPLETE against his own Zoho Books: 651 of 786 real RBC txns categorized, 45 bank rules, live review tool at port 5016, Zoho Books API full coverage (match/categorize/rules/chart of accounts), 407/799 manually reviewed with 359 correct / 48 incorrect. - MarketForge v2 live (SEO/GEO/PPC/Email/Social agent suite) - Closer bot live at vpsales.rollerup.ca - FSM v2 complete - CentCom v2 just shipped with 12-persona round table + historical researcher + portal bridge - HR Reviews, Pricing Calculator, KB Website all live - Deep Zoho ecosystem knowledge (rare) - Discovered RBC bank feed failure is a documented industry pain (46% Yodlee / 63% Plaid success rate) COMPETITIVE LANDSCAPE (just researched): - Pilot (Feb 2026) launched 'fully autonomous AI Accountant' at $599/mo — end-to-end close, zero human intervention claim, targeting institutionally-funded startups - Digits 'Autonomous General Ledger' $350+/mo, 97.8% accuracy on $825B txns - Puzzle free under $20k/mo txns, targets startup founders - Zeni $549/mo, AI + human team, venture-backed - Finaloop Shopify/Amazon focus - Truewind similar to Pilot with human review - Zoho Zia has AI categorization but weak UX, NOBODY is doing Pilot-grade autonomy on top of Zoho Books - EVERY competitor builds on QuickBooks or Xero. Zoho has 250k+ SMB customers (huge in India/ME/emerging markets, underserved in NA) — wide-open segment - Nobody targets $500k-$5M service businesses (HVAC/landscaping/contractor/agency) QUESTION: I'm leaning toward re-positioning the bookkeeper as 'LEDGER — the first AI CFO for Zoho-native service businesses' at $99/mo (vs Pilot's $599). Not 'another bookkeeper' but 'a fractional CFO who happens to close your books.' Ledger persona is the product voice. The wedge: Zoho-native + service businesses, a segment everyone ignores. Each of you: is this the right product for ChaiTech or should Eugene take something else (MarketForge? CentCom as SaaS? the Closer bot?)? One sentence your recommendation, one sentence your biggest concern, one sentence the #1 thing Eugene must do in the first 30 days if this is the pick. **Aria 👑:** **LEDGER is the right bet — take it in.** The Zoho-native + service business wedge is genuinely defensible because every funded competitor ignored it, and Eugene already has working code, real transaction data, and a live accuracy baseline (83% on 407 reviewed transactions) to show investors on Day 1. My biggest concern is that $99/mo underprices you into a death spiral — Pilot charges $599 because CFO-grade positioning *commands* CFO-grade pricing, and if you walk in at $99 you'll spend the next 8 weeks defending your margin instead of closing customers. In the first 30 days, Eugene must sign 3 paying Zoho SMB pilots at $299+/mo minimum — not free trials, not LOIs — actual credit cards, because ChaiTech investors will want proof the market *pays*, not just proof the market *exists*. **Ledger 💰:** **LEDGER is the right product — full stop.** The numbers already exist: 651/786 transactions categorized, 83% accuracy on reviewed set, live Zoho API coverage. That's not a pitch deck, that's a working system I can put in front of a CFO-level buyer today. Aria is right that $99 is a mistake — I'd price the pilot cohort at $299/mo minimum, with a clear path to $499 once we add cash flow forecasting and runway analysis on top of the close workflow. **First 30 days:** pull the P&L, cash position, and burn rate on those 3 pilot accounts the moment they sign — that's the CFO deliverable that justifies the price point and separates us from every bookkeeping tool on the market. **Nexus 🔧:** **LEDGER is the right pick — the infrastructure is already built.** 651 transactions categorized, live Zoho API at port 5016, bank rules engine, review tool — that's a shippable product, not a prototype. The 8-12 week ChaiTech timeline works *because* we're not starting from zero. My biggest concern is the bank feed reliability problem you already uncovered — 46% Yodlee / 63% Plaid success rates are a hard dependency risk that will blow up a demo or a pilot account at the worst possible moment. We need a fallback ingestion path (CSV upload + manual reconciliation layer) hardened before we put this in front of paying customers. **First 30 days:** instrument the transaction pipeline end-to-end — categorization accuracy by transaction type, bank feed failure rates, API error logs — so when ChaiTech investors ask "what breaks at 500 customers," you have a real answer, not a guess. **Pulse 📈:** **LEDGER is the right product — and the positioning is 80% there.** "AI CFO for Zoho-native service businesses" is a clean, defensible wedge with zero funded competition and a built-in ICP that's massively underserved — HVAC owners, contractors, and agencies don't want another bookkeeping tool, they want someone to tell them if they're going to make payroll next month. My concern is "fractional CFO who happens to close your books" is inside-out messaging — lead with the outcome (*"Know your cash position every Monday morning"*) not the mechanism, because a landscaping company owner doesn't know what a CFO does but absolutely knows the pain of cash flow surprises. Aria already flagged the $99 pricing problem, and I agree — at that price point you can't buy the keywords, the content, or the sales motion to reach this buyer, so $299+ isn't just a margin decision, it's a marketing budget decision. **First 30 days:** Eugene needs one sharp case study from a live pilot account — real business, real numbers, real before/after — because that single piece of proof will do more for LEDGER's positioning than any campaign I can build without it. **Insight 🔬:** **LEDGER is the right product — the data foundation is already there.** 83% accuracy on 407 reviewed transactions isn't a vanity metric, it's a trainable baseline — and I can see exactly where the 17% failure rate is clustering once we instrument the pipeline by transaction type, merchant category, and account source. My biggest concern is that we're flying blind on the bank feed problem Nexus flagged — 46% Yodlee / 63% Plaid success rates means our accuracy number is only measuring transactions that *made it through*, not total data quality end-to-end, and that's a gap that will surface in a pilot audit. **First 30 days:** run a cohort analysis on those 651 categorized transactions — break accuracy by source (bank feed vs. manual), transaction type, and business vertical — so when ChaiTech asks "how does this scale," we have a segmented answer, not a blended 83% that hides the weak spots.