Every day, accountants and auditors bring Trulli their toughest questions – from fact-checks to deep dives into complex regulations. Trulli isn’t just a chatbot. It’s an AI-powered assistant, built for accounting and audit.
Unlike standalone tools, Trulli is embedded in the Trullion platform, moving with you through every workflow. Each month, accountants ask thousands of queries across categories like policy research, document summarization, data analysis, compliance validation, and refining work.
What makes Trulli different from ChatGPT?
There are plenty of general-purpose AI systems out there. Trulli is different because it’s built specifically for accountants:
- Closed-loop AI. Trulli only references documents you upload into Trullion. It never pulls from the open web, reducing risk and ensuring answers are grounded in your actual data.
- Source and traceability. Every Trulli response links back to its source document, making it easy to review, validate, and audit outputs.
- Proprietary pre/post processing. Trulli runs on a pipeline engineered for accounting tasks. It uses computer vision and OCR to turn unstructured documents into structured data, applies standards-aware rules to capture key details like lease terms and payment schedules. Trulli then returns answers – including summaries, data analysis, and policy comparisons – always linked back to source documents.
General-purpose models like ChatGPT can’t:
- Parse accounting documents into structured, auditable fields
- Apply standards in a systematic way
- Produce outputs tied back to evidence with a built-in audit trail
These capabilities make a world of a difference — enabling Trulli to provide answers that are not only accurate, but also reviewable and ready for audit.
The top 5 ways accountants use Trulli
We’ve seen five use cases that deliver the most value. Here’s how Trulli handles them:
1. Pinpointing details
Example question. “There is a charge for $1800. What is this for?”
Answer. Based on the account statement, the charge of $1800 on [DATE] is for [PAYMENT]. This appears to be a direct debit payment for business rates to the local authority [NAME].
Why it matters. No more digging through endless documents. Trulli can search across a single PDF or hundreds of files at once, and return the exact detail you need with a clickable link back to the source.
2. Applying accounting treatments
Example question. “Our client is changing from straight-line to double-declining balance depreciation. What footnote disclosures are required?”
Answer. Changing from straight-line to double-declining balance depreciation requires specific footnote disclosures:
- Description of the change: Clearly explain how and why the depreciation method was changed. For example, the company might explain that the double-declining balance method better reflects the pattern in which the assets’ future economic benefits are expected to be consumed.
- Effect on net income: Disclose the impact of this change on the current period’s net income. This should be a quantifiable dollar amount, clearly presented in the income statement or related footnotes.
- Cumulative effect on prior periods: A change in depreciation method is generally treated as a change in accounting estimate under US GAAP (ASC 360-10-35-9). This means the change is applied prospectively – only to current and future periods – and there’s no need to adjust prior periods. You generally would not need to disclose a cumulative effect on prior periods. However, it’s crucial to consult the relevant accounting standards and potentially seek expert advice to confirm the appropriate treatment in your specific situation.
Why it matters. Instead of flipping through standards or asking colleagues, Trulli gives you the answer in plain language – and you can dig deeper or tailor as needed. Watch the demo.
3. Summarizing and explaining changes
Example question. “Why did cash flow from operations increase by $1.2M this quarter?”
Answer. Cash flow from operations increased by $1.2M (from $3.8M in Q2 2024 to $5.0M in Q2 2025) primarily due to earlier collections from key customers and a $650k reduction in accounts payable settlements this quarter. Additional contributors include higher net income ($420k) from increased subscription revenue and a $150k decrease in prepaid expenses. These gains were partially offset by a $120k increase in inventory purchases.
Why it matters. Analyzing variances usually means hours of manual work. Trulli pinpoints the drivers behind changes and connects explanations back to source data.
4. Validating accuracy and compliance
Example question. “The lease modification journal entry has been posted – does it comply with IFRS 16 modification guidance?”
Answer. Yes, the journal entry complies with IFRS 16. The modification reduced the lease term by 12 months, triggering a remeasurement of the lease liability using the revised discount rate of 4.3%. The corresponding adjustment to the right-of-use asset was recorded in full, as required by IFRS 16.44–46. [View journal entry → GL report, line 245]
Why it matters. Compliance checks are high-stakes and repetitive. Trulli verifies entries against source documents, applies the correct rule, and flags any exceptions. Watch the demo.
5. Refining language
Example question. “Can you edit this audit question? ‘From the accounts payable listing, are there any payments made after year-end that relate to the prior reporting period?’
Answer. Here are three ways to make the question more clear:
- 1. Cutoff focus: “To ensure proper accounts payable cutoff, were any payments made after year-end for goods or services received in the prior reporting period? If so, please provide details.”
- 2. Data-specific: “Review the accounts payable listing and identify any payments made after [Year-End Date] that pertain to invoices dated on or before [Year-End Date]. Provide supporting documentation for any payments identified.”
- 3. Accruals focus: “Were there any payments made after year-end that should have been accrued as accounts payable in the prior reporting period? If so, provide details and justification for why they were not initially accrued.”
Why it matters. Editing and refining often takes longer than drafting. Trulli instantly improves clarity and ensures compliance, saving hours of review time.
Prompting best practices
Prompts in Trulli should read like clear workpaper instructions: what to do, under which rules, on which files, and in what format. Load your request with as much context as possible, just as you would when onboarding a new team member.
Order matters for Trulli – upload your documents and datasets first, then give instructions – so answers stay accurate and citeable.
A helpful framework comes from Ben Hylak’s (ex-Apple VisionOS designer) prompt template:
- Instruction: What do you want Trulli to do?
- Context: What are the standards, period, or background details it should apply?
- Input data: What are the exact terms, numbers, or files that Trulli should analyze?
- Output: How should the answer be structured (summary, table, draft note, etc.)?
For example:
- Instruction: “Summarize the key terms of this lease contract.”
- Context: “Under IFRS 16, for the Q2 2025 audit.”
- Input data: “See attached lease agreement.”
- Output indicator: “Provide a bulleted list of terms and a draft disclosure paragraph.”
We’ll publish a separate deep dive blog post with more Trulli prompt examples, troubleshooting, and best practices – stay tuned.
Why these questions matter
These five query types are the fastest ways to unlock value with Trulli. Whether you’re finding facts, interpreting results, checking compliance, or refining documentation, good prompts lead to faster, more accurate, source-backed answers.
Trulli isn’t just a chatbot. It’s an AI agent built for accounting and audit – able to read, interpret, compare, and validate the documents you already work with.
Watch our four short videos to see Trulli in action, then upload your own files and start asking. Try Trulli on your next task.
