Investors are using AI for research. CFOs are responding with sharper IR and private AI
Investor research is getting automated. According to new Gartner data, 43% of CFOs spent more time preparing for earnings calls in 2025 than in 2024. At least 34% also reported increases in the volume, frequency, and sensitivity of investor communications.
The takeaway is simple: investors have AI. If companies don't level up their own capabilities, they'll lose control of the narrative and fall behind faster-moving funds.
Why this matters right now
AI compresses research cycles that used to take days into minutes. That means investors catch inconsistencies sooner, surface edge cases you didn't consider, and compare you to peers with ruthless precision.
As Dymah Paige, director analyst at Gartner, put it: "It is going to become increasingly difficult for organizations to control their narrative and influence investors with manual tools alone." Private AI that's accurate, secure, and auditable is becoming table stakes.
What investors are already using
- Fundamental and equity research terminals: Automate reading 10-Ks, call transcripts, and news to extract trends, KPIs, and anomalies.
- Technical analysis and trading bots: Spot chart patterns and trigger entry/exit rules without human lag.
- Portfolio "co-pilots": Suggest rebalancing, tax moves, and risk hedges based on goals and constraints.
- Alternative data and sentiment trackers: Parse unstructured signals-social chatter, app reviews, job postings, web traffic-to pressure-test your narrative.
Match investor AI with a private, compliant stack
You don't need to build from scratch. You do need privacy, traceability, and legal comfort. Start with a small, controlled environment and expand as trust grows.
- Stand up a private LLM workspace (virtual private cloud or on-prem) with audit logging.
- Ingest your filings, earnings transcripts, guidance history, FAQs, non-GAAP reconciliations, and peer comps into a vetted retrieval layer.
- Add guardrails: source citations, prompt libraries, PII redaction, and automatic legal/compliance checks.
- Run red-team reviews on sensitive claims and forward-looking statements before anything goes external.
Earnings call prep, upgraded
- Draft and refine: Generate a first-pass script, talking points, and CEO/CFO remarks grounded in citations.
- Anticipate questions: Simulate the toughest buy-side and sell-side Qs by theme: margin durability, unit economics, cohort health, FX risk, and capital allocation.
- Proof pack: Auto-produce appendices with bridges, sensitivity tables, and peer benchmarks to support claims.
- Dry run: Practice with an AI interviewer that probes inconsistencies and flags risky phrasing.
- Post-call: Summarize sentiment by analyst and fund, cluster concerns, and draft follow-ups and 8-K language.
Control the costs that can spiral
Gartner highlights three cost drivers CFOs should get in front of early. Treat them like any other variable cost line: visible, governed, and tied to outcomes.
- Pace of consumption: Track tokens, users, and workloads. Set usage budgets and route heavy jobs to cheaper models when quality allows.
- Vendor dependency: Push for transparent pricing and flexible terms. Negotiate tiered rates, credit rollovers, and exit clauses to avoid lock-in.
- Data management: Good data is expensive. Use federated, adaptive governance to reduce duplication, control lineage, and apply standards by use case-not one-size-fits-all.
Data you'll need to get right
- Source of truth for KPIs, metric definitions, and non-GAAP policies.
- Historical guidance, model assumptions, and variance explanations.
- Peer benchmarks, segment detail, and risk disclosures mapped to controls.
- Access rules so sensitive data never leaks into public prompts or third-party tools.
Metrics that prove value
- IR efficiency: Time to draft script and Q&A, number of manual hours saved.
- Accuracy: Hallucination rate, citation coverage, legal/compliance flags per draft.
- Investor outcomes: Post-call sentiment shift, follow-up question volume, model alignment across top analysts.
- Unit economics: Cost per prepared answer, cost per research task, percent of workloads on lower-cost models.
Practical 30-60-90
- 30 days: Pilot a private IR workspace with last two quarters' materials. Run simulated Q&A and legal review. Baseline costs and quality.
- 60 days: Expand to FP&A, treasury, and legal users. Add governance dashboards and usage caps. Begin vendor renegotiations.
- 90 days: Integrate with BI for live KPI feeds, roll out prompt libraries, and publish an AI-in-IR policy for the board and audit committee.
Risk and compliance guardrails
- Document Reg FD-safe workflows and train spokespeople on permissible disclosures. See the SEC's guidance for context: Regulation FD.
- Keep a full audit trail: prompts, outputs, approvers, timestamps, and sources.
- Quarantine experimental models; production IR uses approved, monitored models only.
Bottom line
Investor research is getting faster and sharper. Meet it with disciplined IR, private AI that's safe and traceable, and cost controls that scale with usage. Do that, and you'll spend less time firefighting and more time shaping how the market sees your business.
Helpful resources
- High-level perspective on AI and governance from Gartner: AI insights.
- Curated AI tools for finance teams: AI tools for Finance.
Your membership also unlocks:
