Intuit and Perplexity Partner on AI Features for QuickBooks
Intuit and Perplexity have announced an AI partnership touching QuickBooks. We look at what it promises and what small businesses should do now.
Intuit has announced a partnership with Perplexity, the AI search company, to build AI integrations across its small business products. QuickBooks sits at the center of the deal. For accountants and small business owners, the news signals another step toward software that answers questions rather than simply recording transactions.
What did the two companies announce?
The announcement, made in late October 2025, pairs Intuit’s financial software with Perplexity’s answer engine. Perplexity is best known for search tools that reply in plain sentences with cited sources. Intuit wants that style of interaction inside its products.
The reported perks are reciprocal: QuickBooks customers get a year of Perplexity Pro at no cost, and Perplexity Pro subscribers get access to QuickBooks Online on similar terms. Eligibility rules and rollout timing were still being filled in at the time of writing. Treat the bundle details as provisional until you see them in your own account.
The promise for small businesses is plain-language help: ask about your numbers, get a sourced answer, and act on it without leaving your workflow.
AI is already inside QuickBooks
This deal is not Intuit’s first move into AI. Intuit Assist, the company’s built-in assistant, already answers questions and drafts messages inside QuickBooks Online. Intuit has also been rolling out agents that chase unpaid invoices, reconcile accounts, and handle routine bookkeeping chores.
The Perplexity partnership extends that push rather than starting it. We work with QuickBooks data every day, and our interest is practical: does a tool save reconciliation time, or create new cleanup work?
Should you connect an assistant to your books?
Before linking any assistant to live accounting data, it pays to be deliberate. AI tools are only as good as the numbers they read, and a misread figure can travel fast through reports, forecasts, and tax filings.
There is also a privacy question. An integration that lets an outside model query your books should say what it can see and how long it retains anything. It should also say whether your data helps train future models. If those answers are hard to find, wait.
Questions worth asking before you opt in
A short checklist covers most of the risk:
- What can the assistant see: the full company file, or a filtered view?
- Can it change records, or only read them?
- Do answers show their sources, so you can verify the numbers?
- Can you switch it off for specific users?
None of these are hostile questions. Vendors expect them, and the answers tell you how much checking a tool will save you versus create.
A sensible way to try it safely
If you plan to test the new integrations, start from a clean base. Reconcile your accounts, tidy uncategorized transactions, and save a backup before anything new connects. Then quiz the assistant on things you already know, such as last month’s total sales or your largest customer. If it nails those, widen the circle. If it stumbles, you have lost a few minutes and nothing else.
Clean data in, trustworthy answers out: that is the whole game.