What is the Watcher agent in the Agentic AI CFO Playbook?
+
The Watcher is the cash-position monitoring agent at the core of the Agentic AI CFO Playbook. It polls QuickBooks, Xero, or NetSuite on a defined cadence — hourly for early-stage startups, every 15 minutes for post-fundraise operators — and emits a structured event when cash crosses a threshold you set, typically 2x monthly burn as buffer, 1x monthly burn as critical. The Watcher does not draft narratives or take action; it only emits.
How often does the Watcher poll for cash position?
+
By default the Watcher polls every hour — long enough that bank-API rate limits are never exhausted, short enough that you are alerted inside the same business day when cash crosses a threshold. Post-fundraise operators set the cadence to every 15 minutes because the cost of a missed cash dip is a missed fundraising trigger. Pre-seed founders can stretch to every 4 hours because cash swings are slower at low burn. The cadence is a single parameter in the agent prompt.
What thresholds should the Watcher monitor?
+
Three thresholds cover 90% of cash monitoring cases without alert fatigue: a soft alert at 2x your monthly burn (your buffer zone), a hard alert at 1x monthly burn (you are inside one payroll cycle of running out), and a critical alert at 0.5x monthly burn (you have a fundraising deadline inside 4 weeks). Soft alerts route to the Editor for context, hard alerts route to the Narrator for a written briefing, and critical alerts trigger the Referee escalation loop directly.
Can the Watcher connect directly to a bank feed?
+
Yes. The Watcher accepts any webhook source, including Plaid direct bank feeds, Mercury, Brex, Ramp, Wise, and any bank that exposes an API for transaction polling. Plaid is the fastest path for non-tech-enabled banks because it abstracts the bank-specific OAuth and customer-permission flow. If you already have a read-only API key in NetSuite or a Stripe Connect read-only key, the Watcher will poll those instead.
What happens when cash crosses a threshold in a single day?
+
The Watcher emits a single structured event — payment reference, amount, before/after cash, threshold crossed — and downstream subscribers decide what to do with it. The Referee escalates if the dip is anomalous (a vendor refund reversal you did not expect, a duplicate ACH), the Narrator drafts a one-paragraph brief if the dip is material, and the Strategist reruns the 12-month forecast if the dip is large enough to change your runway window. Nothing happens automatically that you cannot override.
Does the Watcher replace a CFO daily cash review?
+
Yes, for daily cash monitoring the Watcher replaces what a controller or fractional CFO would otherwise poll manually each morning. A human-driven process is consistent only as far as the human remembers to do it, and most cash-management misses happen because the daily review lapsed during a travel week or a hiring push. The Watcher runs every hour without a human in the loop, and routes the things that need attention to your inbox in plain English. A human still owns the strategic decisions.
How is the Watcher different from a standard bank alert?
+
A standard bank alert fires when your balance crosses an arbitrary number you typed into the bank UI; the Watcher fires on cash position after reconciliation, against your monthly burn ratio, with full transaction-line context attached. Bank alerts suffer false positives because the alert is fired before reconciliation. The Watcher polls after reconciliation so the alert you receive is the alert that actually matters. It is also a structured event, not an SMS — the output feeds the Editor, Narrator, and Strategist as a single object they can subscribe to.
What does the Editor agent do in the Agentic AI CFO Playbook?
+
The Editor is the transaction categorization agent. It reads uncategorized transactions from the Watcher output, applies your chart-of-accounts rules, and either auto-approves the categorization or routes the transaction to a human for review. Vendor spend, payroll, refund reversals, and intercompany transfers — every transaction gets routed through the Editor before it lands in your P&L. The Editor output feeds the Strategist and the Referee, so the forecast and the anomaly loop only see categorized data.
How does the Editor categorize transactions?
+
The Editor applies three layers of rules: vendor-name lookups (every Stripe or AWS transaction is sorted based on a vendor dictionary you maintain), chart-of-accounts mapping (every merchant category maps to a GL code), and an LLM layer that catches the long tail of ambiguous transactions. The LLM layer is the safety net — the deterministic rules cover 80–85% of transactions; the remaining 15–20% are the ones where the LLM reasoning earns its place.
What auto-approve thresholds does the Editor support?
+
Three tiers: full auto-approve (transactions under a dollar threshold you set, typically $500–$2,500, with vendor-name matches in your dictionary), conditional auto-approve (transactions on the long-tail LLM categorization but below a confidence threshold, typically 0.85), and route-to-human (anything above the auto-approve thresholds or with low LLM confidence). Most teams route 60–80% of transactions through full auto-approve, 15–25% through conditional, and 5–10% to human review.
Can the Editor handle refunds and reversal events?
+
Yes — refunds, reversals, and credit memos are first-class citizens in the Editor. The agent treats them as their own category and triggers a Referee lookup for the corresponding original transaction. If the original transaction was already categorized, the reversal is auto-tagged as a paired event. If the original transaction was not found, the reversal routes to the human review queue with a flag that says paired reversal not found — investigate.
How does the Editor keep the chart of accounts clean?
+
The Editor maintains a small normalization layer between raw bank transactions and your chart of accounts. Vendor name AWS and Amazon Web Services and AMAZON.COM/AWS all map to the same GL code; vendor name Stripe and STRIPE PAYOUT and STRIPE TRANSFER all map to the same code. The normalization dictionary lives in your agent prompt or a small JSON file alongside the playbook — when you onboard a new vendor, you add one line to the dictionary and the Editor learns. The dictionary is version-controlled so categorizations are auditable.
What happens to a low-confidence categorization?
+
The Editor queues the transaction into the human review queue and emits a structured event with everything needed to make the decision: vendor name, amount, memo, date, the suggested GL code, the LLM confidence score, and the top 2–3 alternate categories the LLM considered. Most low-confidence transactions are resolvable in 5–15 seconds by a glance. The review queue is delivered as a daily digest email; you can also wire it into Slack, Notion, or any task tool.
What does the Strategist agent do in the Agentic AI CFO Playbook?
+
The Strategist is the forecasting agent. It ingests categorized transactions from the Editor and runs a rolling 12-month cash flow forecast with three scenarios — base, upside, downside. The Strategist regenerates the forecast on trigger rather than on a fixed schedule, so the forecast you read at 9am on Wednesday reflects every cash event that cleared overnight. Pairs naturally with /tools/cash-flow-intelligence as the always-on version of a single-shot 12-month forecast.
How does the Strategist generate a 12-month forecast?
+
The Strategist applies a base scenario derived from your trailing 90 days of cash position and burn rate, an upside scenario with stronger inflows (+20% inflow or -15% on big-ticket renewals), and a downside scenario with slower inflows (-20%) or higher costs (+15%). The three scenarios run simultaneously and the Strategist emits both the headline week-by-week lines and a probability distribution of month-12 cash outcomes — not just three single numbers.
What inputs does the Strategist need?
+
The Strategist needs current cash position, trailing 90 days of categorized inflows and outflows (courtesy of the Editor), known large one-time costs coming up (vendor renewals, debt service, planned hires), and your revenue baseline (MRR, growth rate, conviction-weighted pipeline). The Strategist also accepts scenario overrides per quarter — bumped headcount, a deferred equipment purchase, a new contract signing — that trigger a forecast regen without affecting the base scenario history.
How often does the Strategist run?
+
By default the Strategist runs weekly on Monday morning alongside the weekly cash briefing. It also runs on trigger: whenever the Watcher fires a hard or critical threshold crossing, the Strategist reruns to confirm the threshold event remains a real one against the latest data. The trigger-based regen is what makes the playbook feel agentic — the forecast is the freshest possible view whenever a significant cash event has cleared. For most teams, the trigger regens happen 4–6 times per month.
Can the Strategist model M&A or fundraising scenarios?
+
Yes — the Strategist accepts scenario overrides per quarter, including hypothetical fundraising rounds (a $5M Series A closing in month 4), debt service (a $500K line of credit draw in month 3), and large M&A inflows. The overrides are explicit parameters in the agent prompt; they never affect the base scenario history. You can run as many override scenarios as you want, side by side, without polluting the main forecast.
How does the Strategist handle revenue volatility?
+
The Strategist applies convolution against your revenue distribution — not a single MRR point estimate. If your MRR is $110K with a 12% standard deviation, the Strategist draws 1,000 sample paths and reports the month-12 cash distribution as the median, the 10th percentile, and the 90th percentile. Companies with stable recurring revenue see tight distributions; companies with project-based or contract-heavy revenue see fat tails. The Strategist adapts to your revenue shape automatically.
What does the Narrator agent do in the Agentic AI CFO Playbook?
+
The Narrator is the board-reporting agent. It converts the Strategist forecast, the Watcher cash events, and the Referee anomaly findings into a one-page weekly CFO briefing and a monthly board pack. The Narrator locks a tone (CFO-voice, no marketing language, no hyperbole) and a structure (cash position, runway, key events, recommended actions) so every briefing reads like it was written by the same person. Pairs with the Financial Analysis tool at /tools/financial-analysis for benchmark-bearing board packs.
How does the Narrator write a weekly CFO briefing?
+
The Narrator pulls the week events from the Watcher, the week categorization summary from the Editor, the latest forecast from the Strategist, and the week anomalies from the Referee, then drafts a 4-paragraph briefing: cash position snapshot, runway impact, key events and categorizations, and recommended actions. The brief is calibrated to fit in a 3-minute read because most CFOs and founders read their briefings on the way to the next meeting. The Narrator tone is locked to neutral CFO-voice.
What does a Narrator-drafted board pack look like?
+
The monthly board pack is a 6–10 page document with cover slide, executive summary, cash and runway chart, burn rate trend, revenue and growth summary, key events list, anomaly findings, key initiatives for the next quarter, asks of the board, and appendix metrics. Every slide has speaker notes drafted in CEO-voice so a board member can read the deck in 5 minutes and walk into the meeting briefed. The pack is delivered as a PDF on the first business day of each month.
Can the Narrator write investor updates?
+
Yes — the Narrator also drafts monthly investor updates using the same data the board pack uses, but in a different structured format: a 3-paragraph executive summary written for an investor audience (more commercial, less technical), the month KPI dashboard (ARR, growth, NRR, burn multiple), the cash and runway snapshot, and a closing list of what we are focused on next month. The investor update is delivered with the same cadence as the board pack.
How does the Narrator lock tone?
+
The Narrator reads your existing firm voice from a small reference text (3–5 example emails, board memos, or investor updates you provide) so every briefing it writes feels like it came from your team rather than from a generic AI. Tone controls cover prose style (formal vs conversational), vocabulary (financial jargon vs plain English), hedging language (we expect vs we anticipate), and emoji or absence thereof. Once locked, every briefing, board pack, and investor update reads like the same author wrote them.
Can the Narrator output to PDF or Google Slides?
+
Both. The default output is a PDF — copy-paste ready for email distribution to your board list or investor list. Google Slides export is available on the paid CFO Tech Stack tiers because Slides export requires the same Google OAuth setup as the rest of the platform slide features. The free playbook output is PDF-only. If you need Slides export today, run the playbook with a copy-paste into Google Slides — it is 5 minutes of work.
What does the Referee agent do in the Agentic AI CFO Playbook?
+
The Referee is the anomaly detection agent. It cross-checks the Editor categorizations, the Watcher cash events, and the Strategist forecast against historical baselines and emits a structured alert when something does not match. A vendor payment that is 3x the historical average for that vendor, a refund reversal that has no paired original transaction, a forecasted month-12 cash that is 25% below the trailing 90-day pattern — the Referee catches these without a human monitoring. Each alert routes to your inbox with a recommended action.
How does the Referee detect anomalies?
+
The Referee runs three layers of detection: per-vendor statistical thresholds (any spend on a known vendor above the 95th percentile of historical spend triggers an alert), cross-period variance (any line item with week-over-week or month-over-month variance above your threshold — typically 20–25% — triggers an alert), and pattern anomalies (a forecast that diverges from the trailing 90-day pattern by more than your tolerance). Most teams set 2 of the 3 layers active at first and add the third once the playbook is mature.
What does the Referee escalation loop do?
+
The escalation loop is the Referee most important behavior. When a high-severity anomaly fires, the Referee emits a structured alert into your escalation channel (email today, Slack tomorrow) with a recommended action, then waits a defined period (typically 4 hours) for human acknowledgment. If no acknowledgment arrives, the Referee escalates to the next tier and waits again. The loop continues until a human acknowledges or the anomaly is auto-resolved.
Can the Referee catch duplicate payments?
+
Yes — duplicate payments and accidental overpayments are first-class anomaly types in the Referee. When the Editor categorizes a transaction that has a near-duplicate (same vendor, same amount, within a defined window — typically 7 days), the Referee emits a duplicate-payment alert with both transaction IDs and your recommended action. Most teams recover 0.5–2% of annual spend through duplicate-payment detection in the first year the Referee goes live.
How does the Referee avoid alert fatigue?
+
The Referee operates on a tiered severity model: low (informational, no human action required), medium (review within the week), and high (acknowledge within 4 hours). Low-severity alerts roll up into a weekly digest; only medium and high alerts route to your immediate inbox. The tier assignment is calibrated to your historical alert distribution — once the playbook has 60 days of history, the Referee auto-tunes the thresholds.
What is the Referee recommended-action output?
+
Every Referee alert is paired with a recommended-action string chosen from a small vocabulary: review, void, investigate, escalate to <name>, mark for board discussion, or no action. The recommendation is generated alongside the alert, never as a separate step. Most alerts can be resolved with a 5-second glance at the recommended action and the rate at which humans follow the recommendation (the acceptance rate) is the Referee primary health metric.
Can the Referee integrate with Slack or Microsoft Teams?
+
Yes — the Referee accepts any webhook URL for alert routing, so a Slack channel, a Microsoft Teams channel, or a Discord channel can be your escalation endpoint. The free Agentic AI CFO Playbook ships with email escalation as the default because email is the lowest-friction path for most operators. Wire up Slack or Teams by adding your channel webhook URL into the agent prompt. Most teams wire Slack within 10 minutes.
Is the Agentic AI CFO Playbook suitable for a pre-seed founder?
+
Yes — the playbook is built for pre-seed founders who are running finance themselves and do not yet have a fractional or full-time CFO. The Watcher handles daily cash monitoring without a finance hire, the Narrator writes the monthly investor update, and the Editor keeps the chart of accounts clean even when you are also handling product, sales, and hiring. Pair the playbook with the /outsourced-cfo-services path when you are ready for human strategic advice on the next round.
How long does the playbook take to deploy for a founder?
+
Most founders complete the full playbook deployment in a single afternoon. Step 1 — connect QuickBooks or Xero to the Watcher — takes about 15 minutes. Step 2 — write out the Editor vendor dictionary — takes another 30–60 minutes depending on how many vendors you have. Step 3 — wire the Narrator tone — takes 15 minutes if you have 3–5 example emails ready. The whole stack is live inside 3 hours; the first weekly briefing arrives the following Monday.
What does a founder persona get out of the playbook in the first 90 days?
+
Three things in 90 days: (1) a steady weekly briefing so the founder stops losing context every Friday afternoon, (2) a clean chart of accounts with categorized transactions so the founder stops hand-coding quarter-end, and (3) a draft investor update every month so the founder stops dreading the monthly investor email. The aggregate time savings is 5–10 hours per week of founder time, which at a fully-loaded cost of $200/hr is $40K–$80K per founder-quarter.
When should a founder stop using the playbook and hire a real CFO?
+
The recommendation is to layer the playbook under a fractional or full-time CFO at $5M–$10M ARR, not replace a CFO wholesale earlier. The playbook covers routine monitoring, categorization, and forecasting — but a human CFO owns the strategy: board management, fundraising narrative, compensation design, lender relationships, audit preparation, and the dozen soft-skill jobs the playbook cannot do. Most companies today use the playbook to extend the founder-as-CFO stage longer than they otherwise could.
Is the Agentic AI CFO Playbook suitable for an agency owner?
+
Yes — agency cash flow is uniquely volatile because revenue arrives in lumps (project milestones, monthly retainers, ad-hoc engagements) while costs run continuously (payroll, software, office). The Referee is the highest-value agent for an agency because AR anomaly detection catches late-paying clients in week 1 rather than month-end. The Editor project-level categorization keeps the agency P&L clean by client, by project, and by deliverable.
How does the playbook help an agency collect AR faster?
+
The Referee watches invoice aging daily and fires an alert on any invoice that crosses your 30 / 60 / 90-day threshold. Combined with the Editor auto-approval of incoming customer payments, the agency owner knows exactly which clients are slipping inside the week rather than the month. The playbook acceptance rate on these alerts is typically 70–80% because most late-paying clients need a nudge email rather than a phone call.
What setup does an agency owner need?
+
A QuickBooks Online account with class tracking enabled (so the Editor can categorize at the client / project / deliverable level), a Stripe account or similar payment processor that posts payouts to your bank feed, and a 1-hour setup window. The playbook then runs autonomously: weekly cash briefing, weekly AR anomaly digest, monthly client-revenue briefing. Most agency owners report that the playbook pays for itself in the first month through faster AR collection.
Is the playbook suitable for a Series A in-house CFO?
+
Yes — the playbook at a Series A company is layered under an in-house CFO, not a replacement. The Watcher and Editor handle the routine monitoring and categorization the CFO would otherwise spend 2–3 days per week on, freeing the CFO for board management, fundraising preparation, and forward-looking strategy. The Strategist reimplements the company 12-month forecast workflow inside an agent stack, with the CFO retaining full override control on scenario assumptions.
How does a Series A CFO use the playbook before a board meeting?
+
The Narrator drafts a draft board pack on the first business day of each month. The CFO reads the draft, edits any sections that need context the playbook does not have (a new enterprise contract, a delayed hire, a strategic decision), and finalizes the pack 48 hours before the board meeting. The full CFO time-commitment to the board pack drops from 15–20 hours per quarter to 4–6 hours per quarter.
What's the difference between the playbook and a Series A company existing FP&A stack?
+
Existing FP&A stacks at Series A companies (Mosaic, Pigment, LiveFlow, Anaplan) are software platforms where a human builds models, maintains dimensions, and triggers forecasts manually. The playbook is an agent stack where the Watcher, Editor, Strategist, Narrator, and Referee do the equivalent work autonomously; the human CFO becomes the editor and exception-handler rather than the model builder. For very large Series A and Series B+ companies the agent stack layers on top of — rather than replaces — the existing FP&A platform.
Is the playbook suitable for a post-fundraise operator?
+
Yes — post-fundraise operators run the most time-sensitive finance ops of any stage because the runway they just bought has a known end date. The Watcher threshold alerts become more aggressive (every 15 minutes instead of every hour) because missing a cash dip in week 50 of an 18-month runway is more costly than missing the same dip in month 30 of a 36-month runway. The Editor vendor-spend categorization becomes more granular because the post-fundraise company is making big strategic bets.
How does a post-fundraise operator use the playbook with a tight runway?
+
The weekly briefing becomes the most-read email of the week because every fundraise-trigger decision is downstream of the briefing content. Most post-fundraise operators wire the Referee high-severity escalation loop directly into a Slack channel with on-call rotation so a cash event does not sit in an inbox for 4 hours without an acknowledgment. The playbook discipline of running an explicitly tracked runway window lets the operator make defensible decisions to their board in real time rather than retrospectively.
What's the upgrade path from the playbook to a full FP&A stack?
+
The playbook is designed as the entry tier. Post-fundraise operators who outgrow the agent stack usually layer an FP&A platform (Mosaic, Pigment, Cube, LiveFlow) on top — the agent stack continues to handle Watcher / Editor / Narrator / Referee duties in days when the FP&A platform is overkill, and the agent stack outputs become inputs to the FP&A deeper modeling. The playbook does not become obsolete as the company grows; it becomes the routine monitoring tier underneath the deeper modeling tier.
What are the use cases for the Agentic AI CFO Playbook?
+
Four canonical use cases are documented in the playbook: founder-as-CFO at a pre-seed or seed-stage startup (Watcher + Narrator are the highest-value agents, deployed in week one), agency owner at a $1M–$10M agency with volatile AR (Referee + Editor are highest-value), Series A in-house CFO at a $10M–$30M ARR SaaS (Strategist + Narrator layered under the existing FP&A stack), and post-fundraise operator at a Series B+ startup with a tight 18-month runway (all five agents active).
Which agents should a founder deploy first?
+
Founders deploy Watcher + Narrator first. Watcher gives the founder a daily cash view that does not require them to remember to log into QuickBooks every morning; Narrator gives the founder a weekly briefing they can actually read in 3 minutes. Both deploy inside 2 hours and produce a measurable quality-of-life improvement inside week 1. The Editor, Strategist, and Referee come online in week 2–3 as the founder has time to think through categorization rules and scenario overrides.
Which agents should an agency owner deploy first?
+
Agency owners deploy Referee + Editor first. Referee AR-anomaly detection is the highest-ROI piece because late-paying clients cost the agency $20K–$50K per year in working-capital drag per million of revenue; catching them inside the week rather than month-end is worth more than every other agent combined. Editor handles project-level categorization so the agency P&L tracks profitability by client rather than just by month.
Which agents should a post-fundraise operator deploy first?
+
Post-fundraise operators deploy all five agents at once. The Watcher cadence is set to every 15 minutes instead of the default hourly because the cost of a missed cash event is a missed fundraising trigger. The Editor vendor dictionary is calibrated to segment-level categorization. The Strategist runs 8–12 override scenarios at any time. The Narrator delivers weekly investor updates and monthly board packs on a locked cadence. The Referee runs the escalation loop with a 1-hour acknowledgment timeout. The full stack is the post-fundraise operating mode.
How does the playbook relate to Cash Flow Intelligence?
+
The playbook Strategist agent reimplements the Cash Flow Intelligence tool at /tools/cash-flow-intelligence in an autonomous framework. Cash Flow Intelligence is a single-shot tool for a 12-month forecast with three scenarios; the Strategist is the always-on version — it auto-regenerates the forecast on trigger rather than only when a human runs the tool. Most teams use both: Cash Flow Intelligence for the monthly formal review, the Strategist for everything that happens between monthly reviews.
How does the playbook relate to Financial Analysis?
+
The playbook board pack uses the same peer-compared ratio benchmarks that /tools/financial-analysis produces. The Financial Analysis tool is the single-shot deep-dive on a quarterly cadence; the playbook Narrator uses the same benchmark dataset monthly and references the same peer-tier distribution in every board pack. Most series A and B CFOs run the Financial Analysis tool quarterly and reference the prior quarter report inside the playbook monthly board pack to show trend lines.
How does the playbook relate to Outsourced CFO Services?
+
The playbook is the routine-monitoring tier; the /outsourced-cfo-services tier is the human strategic CFO tier that layers on top. The playbook handles the Watcher / Editor / Strategist / Narrator / Referee duties autonomously; the outsourced CFO owns the strategic, board-facing, fundraising-facing, and lender-facing work the playbook cannot do (compensation design, audit preparation, lender relationship management, fundraising narrative).
What problem does the Agentic AI CFO playbook solve?
+
The playbook solves the routine-monitoring tax on every small-team finance function. Today most founders, agencies, and post-fundraise operators spend 5–15 hours per week on cash monitoring, transaction categorization, forecasting, board-pack drafting, and anomaly detection — work that is mission-critical but not strategic. The agent stack replaces 80–90% of that work at near-zero marginal cost, with humans retaining judgment over strategy.
Why use agentic AI for finance instead of traditional FP&A software?
+
Traditional FP&A software is a model builder; agentic AI is a model operator. Traditional FP&A (Mosaic, Pigment, Cube, LiveFlow, Anaplan) requires a human to build a model, maintain the dimensions, and trigger forecasts manually. Agentic AI for finance runs the equivalent work autonomously — the Watcher polls cash on a cadence, the Editor categorizes transactions, the Strategist regenerates the forecast, the Narrator drafts briefings, the Referee catches anomalies. The two architectures are complementary.
How much does the Agentic AI CFO Playbook cost?
+
Zero to start. The Agentic AI CFO playbook is free — no signup, no credit card, no account. The free playbook ships with the full prompt templates, vendor dictionary scaffolding, escalation loop configuration, and free weekly briefing cadence. The infrastructure cost to run the agent stack is $0–$49/mo depending on which LLM and orchestration stack you wire it to. The paid CFO Tech Stack tiers add Google Slides export, premium tone library, and one-click Slack integration at $49–$199/mo.
How do I get started with the playbook?
+
Three steps: (1) download the free playbook template from this page; (2) connect QuickBooks, Xero, or NetSuite to your Watcher source; (3) wire the first two agents (Watcher + Narrator are the recommended starting pair for most teams) and let them run for a week. The first weekly briefing arrives the following Monday. Most operators complete the first deployment in under 3 hours.
What is the free AI setup assistant?
+
The free AI setup assistant is a chat-with-CFO-Tech-Stack interface that walks you through the playbook deployment one step at a time. It is built on the same agent stack the playbook describes, with the prompts locked to deployment guidance rather than finance monitoring. Most operators use the setup assistant for their first playbook deployment and then drop into the prompts directly as they get comfortable.
What JSON-LD schemas does the playbook page publish?
+
The playbook page publishes four schemas in addition to the default Organization + WebPage schemas rendered by the page wrapper: HowTo (the 4-step deploy-the-playbook procedure with monetary-cost=0), SoftwareApplication (treats the playbook as a software artifact with FinanceApplication category, Web operating system, provider organization, free Offer, and 6-element featureList), FAQPage (60 mainEntity entries covering the 5 agents, 4 personas, 4 use cases, and the general what / why / how cluster), and Article (headline + author + publisher + datePublished=2026-07-15 + mainEntityOfPage).
What are the AEO citation tags on the playbook page?
+
AEO citation tags are the four HTML <meta> tags that AI search engines like Perplexity, ChatGPT, and Google AI Overviews preferentially pull for citation. The playbook page publishes: citation_title, citation_author (CFOTechStack), citation_publication_date (2026-07-15), and citation_publisher (CFOTechStack.ai). Publishers like Sifted, BOSS Publishing, and Emburse use the same tags in their newsroom CMS setups, so the schema is the standard AEO-friendly metadata.
What is agentic AI in finance?
+
Agentic AI in finance is the use of autonomous AI agents — software entities that take actions and make decisions without a human in the loop — to handle routine CFO work: cash monitoring, transaction categorization, forecasting, board reporting, and anomaly detection. Unlike a chatbot or a forecast tool, an agent takes a defined action on its own when a triggering event occurs: the Watcher fires an alert when cash crosses a threshold, the Editor categorizes a transaction without approval, the Strategist regenerates a forecast on a trigger, the Narrator drafts and sends a weekly briefing, the Referee escalates a high-severity anomaly.