AI Can. AI Can't.
A practical cheatsheet for accounting teams navigating AI: what to hand off, what to hold onto, and where to be careful.
The co-pilot rule
AI handles the instruments: research, drafting, synthesis. You handle the judgment, the sign-off, and the responsibility. Both matter. Neither is optional.
Standards & Research
Delegate to AI | Keep with You |
|---|---|
| Look up ASC codification guidance across multiple topics | Determine how the standard applies to your specific facts |
| Summarize recent FASB updates and exposure drafts | Assess impact on your company's accounting policies |
| Cross-reference two standards for interaction effects | Resolve conflicts when reasonable accountants disagree |
| Explain a complex standard in plain English for stakeholders | Confirm no nuance was lost in the translation |
Contract Review
Delegate to AI | Keep with You |
|---|---|
| Identify performance obligations across a contract portfolio | Validate completeness and distinctness of each obligation |
| Flag variable consideration, termination, and renewal clauses | Estimate the amount to include in the transaction price |
| Summarize 50+ pages of contract terms in minutes | Resolve ambiguous terms where intent of parties matters |
| Process hundreds of similar contracts consistently at scale | Review the anomalies AI surfaces; those are the real work |
Memo Writing
Delegate to AI | Keep with You |
|---|---|
| First draft of a technical accounting position memo | Professional judgment behind the position itself |
| Structure the argument: facts, standard, application, conclusion | Verify the argument holds against your actual facts |
| Generate alternative accounting treatment analyses | Select the appropriate treatment and defend it |
| Draft journal entry rationale and supporting narrative | Sign off and catch errors the AI doesn't know it made |
Disclosure & Audit
Delegate to AI | Keep with You |
|---|---|
| Compare your disclosures against 10-Ks from peer companies | Determine whether your disclosure is actually adequate |
| Generate first-draft footnote and disclosure language | Ensure language reflects your specific facts and audit history |
| Research how peers handle a novel disclosure requirement | Own the conversation with your auditors about your approach |
| Flag gaps between your current disclosures and best practice | Make the materiality call on what to change and when |
The Gray Zone · Handle with Care
These tasks sit in the middle. AI can help, but its output needs closer scrutiny than the "delegate" column above.
Novel transactions with no clear precedent
AI is good at applying known patterns. When a transaction is genuinely new (no comparable precedent, ambiguous standard application), AI output will sound confident but may be wrong. Treat it as a starting point, not an answer.
Materiality determinations
AI doesn't know your auditor's preferences, your audit committee's risk tolerance, or what the market will focus on. It can surface the relevant factors. The call is yours.
Catching its own errors
Language models can't reliably flag when they're wrong. Hallucinated citations, misapplied standards, and misread contract terms can all look plausible. Human review is not optional; it's the whole point.
Your company's institutional context
Prior accounting positions, informal auditor conversations, policy history: AI doesn't have this unless you tell it explicitly. Complex transactions often turn on exactly this context.
TAbot is built for exactly this workflow.
TAbot is an AI platform for technical accounting teams, built to accelerate research, contract review, and memo drafting while keeping your team in control of every judgment call.
Ready to see how TAbot can transform your accounting workflow? Let's discuss your specific needs and how we can help.