<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Precision AI OS</title>
    <description>The latest articles on DEV Community by Precision AI OS (@precisionaios).</description>
    <link>https://dev.to/precisionaios</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4206392%2F0e26416d-59a0-4bd0-bf32-61f13fc51c3e.png</url>
      <title>DEV Community: Precision AI OS</title>
      <link>https://dev.to/precisionaios</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/precisionaios"/>
    <language>en</language>
    <item>
      <title>What an AI accounting agent should do in a CPA firm, and what stays with your people</title>
      <dc:creator>Precision AI OS</dc:creator>
      <pubDate>Mon, 16 Nov 2026 02:02:59 +0000</pubDate>
      <link>https://dev.to/precisionaios/what-an-ai-accounting-agent-should-do-in-a-cpa-firm-and-what-stays-with-your-people-2gdo</link>
      <guid>https://dev.to/precisionaios/what-an-ai-accounting-agent-should-do-in-a-cpa-firm-and-what-stays-with-your-people-2gdo</guid>
      <description>&lt;p&gt;Every tax season the same hours disappear. Partners and seniors spend them chasing clients for missing documents, rebuilding status lists out of email threads, and writing the same follow-up note for the fourth time. None of that is review work, and none of it is client work. It comes straight out of the hours you'd rather spend on the return in front of you, and in March you can't buy them back.&lt;/p&gt;

&lt;p&gt;That's the gap an AI accounting agent is meant to fill. Not the judgment and not the signature. The chasing, the assembling and the sorting.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the agent actually does
&lt;/h2&gt;

&lt;p&gt;An AI accounting agent is software that handles one defined job, such as chasing missing documents or preparing review files, inside systems the firm has authorized. That definition matters more than any feature list. It's one job, in systems you chose, with a clear split between preparing the work and checking it.&lt;/p&gt;

&lt;p&gt;Take document chase. The agent builds a received-versus-requested list for each client and drafts the follow-up email. A staff accountant reads it and releases it. The agent never decides that a client has been chased enough, or that a vague reply counts as a document.&lt;/p&gt;

&lt;p&gt;1040 intake works the same way. The agent compares what came in against last year's return and flags what's missing or changed. The preparer works from that list, and the signing CPA signs. For review prep, the agent puts together the workpaper index, the tie-outs and the open items, and the engagement partner signs off. When an IRS notice arrives, the agent matches it to the client, the year and the return, and the preparer takes it from there.&lt;/p&gt;

&lt;p&gt;In every one of those, the agent does the assembly and a named person owns what goes out.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where the rules still put a person
&lt;/h2&gt;

&lt;p&gt;Software doesn't move the duties that sit on the firm. Circular 230 section 10.22 requires due diligence in preparing and approving returns and other papers filed with the IRS. A tool can make that diligence quicker to carry out. It doesn't carry it out for you, and it doesn't shift the obligation to whoever sold you the tool.&lt;/p&gt;

&lt;p&gt;Then there's disclosure. Section 7216 of the Internal Revenue Code makes it a crime for a preparer to knowingly or recklessly disclose or use tax return information, except where the law allows it. Under 26 CFR 301.7216-3, you need the taxpayer's written consent unless another rule already permits the use. In practice, you settle the consent question before return data ever reaches an agent. Not halfway through the pilot, and not by assuming the vendor's terms cover it.&lt;/p&gt;

&lt;p&gt;Security is the third piece. IRS Publication 4557 says that under the FTC Safeguards Rule, tax preparers have to create and carry out a written security plan. The rule itself, at 16 CFR 314.4, calls for encrypting customer information in transit over external networks and at rest (or approved compensating controls where that isn't feasible), and for logging what authorized users do. So when you look at any agent, ask where the data sits, whether it's encrypted both ways, and whether you can see who did what and when.&lt;/p&gt;

&lt;p&gt;Oversight should be written down too. The NIST AI Risk Management Framework (subcategory MAP 3.5) says processes for human oversight should be defined, assessed and documented. For a firm, defined means you can name the reviewer for each job. Assessed means you check whether the reviews are catching problems. Documented means there's a record you could hand to a peer reviewer.&lt;/p&gt;

&lt;p&gt;That last one is where most firms get loose. Everyone says a person reviews the output. Fewer firms can say who reviewed which draft, and what exactly they approved. If the agent records each draft, who approved it and the version they approved, you get that record without anyone keeping a spreadsheet.&lt;/p&gt;

&lt;h2&gt;
  
  
  A 30-day pilot
&lt;/h2&gt;

&lt;p&gt;Pick one job. Document chase is the usual starting point because the output is easy to judge: either the list was right and the email was usable, or it wasn't.&lt;/p&gt;

&lt;p&gt;Week one is mostly decisions. Choose the job. Name the reviewer and the signer as people, not roles. Write down the systems the agent may read, and nothing past that. Settle 7216 consent. Then set aside a handful of finished files the firm considers right, so you have something concrete to compare against.&lt;/p&gt;

&lt;p&gt;Week two is quiet. Run the agent on old files while staff work as usual, then compare. Did it find the same missing items your people found? Did it miss anything, or flag things that weren't really missing? You're testing your inputs as much as the tool. A sloppy request list produces a sloppy chase no matter what software runs it.&lt;/p&gt;

&lt;p&gt;Weeks three and four use live work, and every draft gets reviewed before it goes out. For each one, note whether it was accepted as written, edited before release, or redone from scratch. Time the review as well. If checking a draft takes as long as writing it, the agent isn't helping yet.&lt;/p&gt;

&lt;p&gt;On day 30 you decide. If most drafts go out with light edits, add a second job. If the agent keeps missing the same kind of document, fix the request list before you add anything. If your people keep redoing the work, change the setup and run another month, or stop. Stopping counts as an answer. Finding out in 30 days that a job doesn't fit beats finding out in April.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why we hold the sends
&lt;/h2&gt;

&lt;p&gt;At Precision AI OS we build an Agent OS for firms like yours. It gives the firm one shared memory and an agent for each area of work, and by default it holds every send and every change in your apps until someone on your team approves it. Each approval is recorded and tied to exactly what was approved. No rule can approve payments, amounts or bank details, email to someone who isn't already on the thread, or deleting files and records. Those are the places where a mistake is expensive and hard to undo, so a person on your team decides.&lt;/p&gt;

&lt;p&gt;That's the standard I'd hold any tool to this season, ours included. The agent should make the preparation faster and the record better. The consent, the judgment and the signature stay with your people.&lt;/p&gt;

&lt;p&gt;More on how we set this up: &lt;a href="https://getprecision.ai/?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=ai-accounting-agent" rel="noopener noreferrer"&gt;https://getprecision.ai/?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=ai-accounting-agent&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>automation</category>
      <category>productivity</category>
    </item>
    <item>
      <title>Best AI tools for CPA firms in 2027</title>
      <dc:creator>Precision AI OS</dc:creator>
      <pubDate>Sun, 25 Oct 2026 10:39:48 +0000</pubDate>
      <link>https://dev.to/precisionaios/best-ai-tools-for-cpa-firms-in-2027-4gpb</link>
      <guid>https://dev.to/precisionaios/best-ai-tools-for-cpa-firms-in-2027-4gpb</guid>
      <description>&lt;p&gt;The best AI tools for CPA firms in 2027 finish one job in the software the firm already runs. A pasted paragraph in a chat window and a W-2 posted into tax software are different jobs. If you run the stack, or you build for firms that do, sort the market that way.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;p&gt;AI for accounting firms in 2027 is a choice of where the work runs. Use the AI inside the system that already holds the client record. Add another tool when the work has to cross the portal, the inbox, the ledger, and the tax software. Check where the data goes, who approves a send or a write-back, whether the step is logged, and which products the connection reaches. A practitioner reviews the draft before it goes to a client or the IRS.&lt;/p&gt;

&lt;h2&gt;
  
  
  AI for accounting firms in 2027: how deep it sits
&lt;/h2&gt;

&lt;p&gt;The AICPA Private Companies Practice Section ran its CPA Firm Top Issues Survey online from April 20 to May 22, 2026. It had 629 respondents. Managing change from technology, and from AI in particular, ranked first for every firm-size group on expected impact over the next five years. Current issues still differ. For firms of 11 to 30 professionals, hiring experienced staff was the top current issue. Plan for that five-year ranking.&lt;/p&gt;

&lt;p&gt;Intuit commissioned a May 2026 survey of 725 US accounting and bookkeeping professionals. Intuit sells software to that audience. They reported about five hours a week moving information across tools, typing it in again, or reconciling it. Forty-one percent said their tools are fully integrated. Forty-eight percent said the setup works and is still fragmented.&lt;/p&gt;

&lt;p&gt;Eighty-eight percent had used AI for at least one client service in the prior 12 months. Thirty percent said AI is embedded as the default in day-to-day work. Fifty-four percent said they use it situationally. On client deliverables, 40% want a support tool, 34% want a draft a person reviews and signs, and 6% want the AI to execute on its own. Sixty percent said that, in the last 12 months, clients asked always or frequently for proof of AI data protection. Three quarters of the sample are in firms of 0 to 99 employees.&lt;/p&gt;

&lt;h2&gt;
  
  
  Four checks before a client file
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Data handling.&lt;/strong&gt; Ask where the file is stored, and whether the vendor trains on it. Also ask whether one client's records can answer a question about another client. IRS Office of Professional Responsibility Issue 2026-19 (June 24, 2026) says privacy can break that way. Data generated for one client can be reused for another client's question. The issue also says to vet a third-party tool, and never to upload sensitive data to an unsecured site.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;A person, before anything leaves.&lt;/strong&gt; The same issue says a practitioner has to review AI-created documents, and language taken from the tool, before they go to a client or the IRS. The practitioner can't rely on AI alone. The issue quotes Circular 230 section 10.22 on due diligence and section 10.37 on written advice. Due diligence means checking the facts, the citations, and the math. Written advice has to rest on reasonable assumptions of fact and law. If the logic is opaque, the issue says, reliance can be unreasonable under section 10.37.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;A log.&lt;/strong&gt; Issue 2026-19 lists documenting the AI use, and the check of that use, as a best practice. A useful record names the client, the tool, the approver, and what was sent or posted. The issue doesn't list those fields. The issue quotes Circular 230 section 10.36. A person with principal authority over the firm's tax practice has to take reasonable steps so the firm has procedures, and so staff follow them. The issue's list for that coverage includes training and a review of outside tools.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The connection.&lt;/strong&gt; Name the tax product and the practice system, edition included. A read login and a write login are different permissions.&lt;/p&gt;

&lt;p&gt;Issue 2026-19 says sections 6713 and 7216(a) carry civil and criminal preparer penalties for unauthorized use or disclosure of tax return information. Section 7216(a) is the criminal penalty. It applies when a preparer knowingly or recklessly uses or discloses that information for a purpose other than preparing the return. Section 6713 is the related civil penalty. The regulation says section 7216(b) sets exceptions and authorizes regulations that permit more uses. The issue cites Treasury Regulation section 301.7216-1(b)(3). That definition includes a taxpayer's name, address, or identifying number furnished for, or in connection with, preparing the return. The regulation says the term isn't limited to those items. Counsel decides whether a step is preparation or a disclosure that needs written consent. This post doesn't make that call.&lt;/p&gt;

&lt;h2&gt;
  
  
  Document intake and extraction
&lt;/h2&gt;

&lt;p&gt;Get the source documents in, read them, and land the figures where a reviewer can tick them.&lt;/p&gt;

&lt;p&gt;Thomson Reuters says 1040SCAN uses optical character recognition to read the page, export the data into tax software, and sort the workpapers into an index. The same page says text-layer matching and AI remove the need to verify the OCR data for 65% of standard documents. That 65% is the vendor's figure. The page doesn't define standard documents, and it doesn't publish an independent test.&lt;/p&gt;

&lt;p&gt;A Thomson Reuters page on SurePrep says the products integrate with CCH Axcess Tax, GoSystem Tax RS, Lacerte, and UltraTax CS, and that only 1040SCAN ORGANIZE works with any tax software. Match the edition on the agreement before anyone promises the export.&lt;/p&gt;

&lt;p&gt;SafeSend describes Next Gen Gather AI as e-sign for engagement letters, questionnaires, document request lists, and sorting. It lists UltraTax CS, GoSystem Tax RS, CCH Axcess, ProSystem fx, and Lacerte. Collecting a file and posting a number into the return are different writes. Test a crooked scan and a combined K-1, and hold a weak field for a named person before export.&lt;/p&gt;

&lt;h2&gt;
  
  
  Client request follow-ups
&lt;/h2&gt;

&lt;p&gt;This job is the chase. What is missing, who was asked, and whether it came back.&lt;/p&gt;

&lt;p&gt;Karbon's AI page says it can draft an email from a task, summarize a thread, and build a client brief from email, notes, work, and billing. It can flag missed time and suggest an assignment. Karbon says this AI is included for current customers at no added cost, and that it will say so before any charge is added. An administrator can turn it off. Karbon says firm data stays in Karbon, isn't shared outside Karbon, and isn't used to train models.&lt;/p&gt;

&lt;p&gt;The same page marks Kai, its coworker, as early access. The introduction says Kai is coming. Meeting intelligence is marked coming soon. Use what an administrator can turn on today.&lt;/p&gt;

&lt;p&gt;SafeSend lists automatic reminders and a view of what the client has completed. Find the control that stops a reminder, and name who may send one. A follow-up that reads last year's organizer may be a use of tax return information. Get counsel's answer on sections 6713 and 7216 before that send exists.&lt;/p&gt;

&lt;h2&gt;
  
  
  Workpapers and review prep
&lt;/h2&gt;

&lt;p&gt;The reviewer needs a source, a number, and the exception.&lt;/p&gt;

&lt;p&gt;Basis says it prepares a return from client intake to review-ready workpapers, runs a month-end close, and cites evidence on audit testing. Its small-firm page says K-1 aggregation includes the footnotes and the citations, that each workpaper figure is cited to its source, and that journal entries are staged for review before they post.&lt;/p&gt;

&lt;p&gt;Basis also says it doesn't train models on client data, that client data stays in US regions, and that it is SOC 2 Type II certified. Ask for the report, and for the ledger list, in the contract. Those pages don't name every general ledger.&lt;/p&gt;

&lt;p&gt;QuickBooks Online's help article, updated August 26, 2026, says the product suggests a match or a category and shows how strong the data behind the suggestion is. A person posts. On Essentials, Plus, and Advanced, Accounting AI can group high-confidence rows so one action posts them. The article says these features are meant to work the same way on the bank page in Intuit Accountant Suite. The reviewer can edit or exclude a line. A write into the ledger, or into the return, is its own permission.&lt;/p&gt;

&lt;h2&gt;
  
  
  Research and drafting
&lt;/h2&gt;

&lt;p&gt;The output is a memo or a client explanation with a source a CPA can open.&lt;/p&gt;

&lt;p&gt;Blue J says its tax answers draw on primary authoritative content, Tax Notes, and IBFD. Open the cited page.&lt;/p&gt;

&lt;p&gt;Microsoft's privacy page for Microsoft Copilot says prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation models. Copilot shows organizational data only when that person already has permission to view it. The page names SharePoint permissions as one control.&lt;/p&gt;

&lt;p&gt;When someone uses Copilot in an app the page names, such as Word, Excel, or PowerPoint, Microsoft stores the prompt and the response. An admin can set retention in Microsoft Purview for Copilot chat. Web search sends a query to Bing. Agents can sit under their own terms. Some third-party models sit under separate terms, and Microsoft says some of those models are outside the EU Data Boundary. For a tax answer, have a person open the primary source.&lt;/p&gt;

&lt;p&gt;Issue 2026-19 says to check the facts, the citations, and the math. Its note on section 10.35, not the quoted rule itself, says the practitioner has to understand the tool, including its limits and the chance of error. Training belongs in the section 10.36 procedures. The issue says to treat the text the tool writes as a draft.&lt;/p&gt;

&lt;h2&gt;
  
  
  Practice management
&lt;/h2&gt;

&lt;p&gt;Assignment, status, and the handoff live here. The tax conclusion stays in the tax software.&lt;/p&gt;

&lt;p&gt;Karbon fits when the job, the email, and the time entry already sit in it. Treat an assignment line as a suggestion. Have a person correct the missed time the product flags. Karbon says later agents will handle repetitive work such as client onboarding and tax preparation. Today an administrator can turn the AI off. The page says per-agent controls come later, as its agent management system develops. Read the agreement after the marketing page. The two can differ.&lt;/p&gt;

&lt;h2&gt;
  
  
  Governance sits under the other jobs
&lt;/h2&gt;

&lt;p&gt;Write down which tools are allowed, which data they may see, who may approve a send, and how long the log is kept. Issue 2026-19 says the duties of competence, diligence, and confidentiality stay in place, and that the final decision stays with the practitioner. The issue names California, Colorado, Illinois, and Utah as examples of states with AI governance legislation. This post doesn't summarize those statutes.&lt;/p&gt;

&lt;p&gt;One page is enough. Add a tool to it before the first client file.&lt;/p&gt;

&lt;h2&gt;
  
  
  A reference architecture
&lt;/h2&gt;

&lt;p&gt;This is a pattern for connecting the jobs. It records no measured deployment.&lt;/p&gt;

&lt;p&gt;The steps run in this order:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Sources:&lt;/strong&gt; tax software, the client portal, the inbox and the ledger.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Per-client memory:&lt;/strong&gt; what the sources hold for each client, kept separate by client.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;One job:&lt;/strong&gt; a single defined task that works from that client's memory.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Approval gate:&lt;/strong&gt; the job's output stops here before anything leaves the firm.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Named reviewer:&lt;/strong&gt; a specific person approves or rejects it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Send or write-back:&lt;/strong&gt; only approved work goes to the client or back into the system.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Audit log:&lt;/strong&gt; the job, the gate and the reviewer each write to the log.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Shared memory holds that client's organizer, open items, prior-year notes, and workpaper index. Partition it by client. Issue 2026-19 gives the reason: data generated for one client can otherwise be reused for another client.&lt;/p&gt;

&lt;p&gt;Put the gate in front of a client message and in front of a write to tax software or the ledger. The reviewer sees the draft, the sources, and the exceptions. Store the client, the tool, the input, the output, the approver, and the time where the firm already keeps records. Make that log exportable.&lt;/p&gt;

&lt;h2&gt;
  
  
  Start with one job
&lt;/h2&gt;

&lt;p&gt;Pick document collection for one return type. Count open items per client, and days to a complete file, before the tool is on. Count them again with the reviewer's time included. Name the approver. Read the data terms for the plan the firm pays for. If the AI inside the practice system, the tax software, or the ledger already covers the job, use that.&lt;/p&gt;

&lt;h2&gt;
  
  
  From Precision
&lt;/h2&gt;

&lt;p&gt;This post is from Precision AI OS. Steven de la Fe, CPA, is a co-founder. Precision AI OS is based in Miami, Florida, and works with businesses of every size. We build the agent layer for your business that works across your entire technology stack. Our agents work across the tools firms already run, such as TaxDome, Karbon, Canopy, QuickBooks, UltraTax CS, CCH Axcess Tax and Microsoft 365. They use public APIs where they exist and work in portals that have no API. The whole team shares one memory, and sends are held for approval by default. Managed is taking discovery calls now. Free, Team, and Enterprise are coming. A team &lt;a href="https://getprecision.ai/resources/ai-implementation-partner-cpa-firms?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=best-ai-tools-for-cpa-firms-in-2027-4gpb" rel="noopener noreferrer"&gt;implements AI agents for a CPA firm&lt;/a&gt; by connecting the tax software, practice management, books, and email the firm already runs, and the firm keeps those tools. The &lt;a href="https://getprecision.ai/industries/cpa-firms?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=best-ai-tools-for-cpa-firms-in-2027-4gpb" rel="noopener noreferrer"&gt;CPA firms page&lt;/a&gt; describes how those agents work across the systems a firm already uses. Learn more at &lt;a href="https://getprecision.ai" rel="noopener noreferrer"&gt;getprecision.ai&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Sources
&lt;/h2&gt;

&lt;p&gt;Checked October 25, 2026. Each row is a page opened that day. Product lines are what that vendor says on the linked page.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Source&lt;/th&gt;
&lt;th&gt;URL&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;AICPA and CIMA, CPA Firm Top Issues Survey release, June 23, 2026&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.aicpa-cima.com/news/article/aicpa-survey-cites-change-management-for-technology-and-ai-as-top-long-term-issue-facing-accounting-firms" rel="noopener noreferrer"&gt;https://www.aicpa-cima.com/news/article/aicpa-survey-cites-change-management-for-technology-and-ai-as-top-long-term-issue-facing-accounting-firms&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Intuit, 2026 Accountant Technology Survey, Firm of the Future&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.firmofthefuture.com/news/accountant-tech-survey-2026/" rel="noopener noreferrer"&gt;https://www.firmofthefuture.com/news/accountant-tech-survey-2026/&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IRS Office of Professional Responsibility, Issue 2026-19, June 24, 2026&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.irs.gov/pub/opr-taxpros/issue-number-2026-19-introductory-guidelines-for-responsible-ai-use-in-federal-tax-practice.pdf" rel="noopener noreferrer"&gt;https://www.irs.gov/pub/opr-taxpros/issue-number-2026-19-introductory-guidelines-for-responsible-ai-use-in-federal-tax-practice.pdf&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IRS bulletin, Alerts from the Office of Professional Responsibility, June 24, 2026&lt;/td&gt;
&lt;td&gt;&lt;a href="https://content.govdelivery.com/accounts/USIRS/bulletins/41d6e70" rel="noopener noreferrer"&gt;https://content.govdelivery.com/accounts/USIRS/bulletins/41d6e70&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IRS, Section 7216 information center, page reviewed or updated September 7, 2026&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.irs.gov/tax-professionals/section-7216-information-center" rel="noopener noreferrer"&gt;https://www.irs.gov/tax-professionals/section-7216-information-center&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;26 CFR Â§ 301.7216-1, Legal Information Institute text. Subsection (b)(3)(i) matches the quotation in Issue 2026-19.&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.law.cornell.edu/cfr/text/26/301.7216-1" rel="noopener noreferrer"&gt;https://www.law.cornell.edu/cfr/text/26/301.7216-1&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Thomson Reuters, 1040SCAN&lt;/td&gt;
&lt;td&gt;&lt;a href="https://tax.thomsonreuters.com/en/products/1040scan" rel="noopener noreferrer"&gt;https://tax.thomsonreuters.com/en/products/1040scan&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Thomson Reuters, SurePrep&lt;/td&gt;
&lt;td&gt;&lt;a href="https://tax.thomsonreuters.com/en/sureprep" rel="noopener noreferrer"&gt;https://tax.thomsonreuters.com/en/sureprep&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SafeSend&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.safesend.com/" rel="noopener noreferrer"&gt;https://www.safesend.com/&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Karbon AI&lt;/td&gt;
&lt;td&gt;&lt;a href="https://karbonhq.com/feature/ai/" rel="noopener noreferrer"&gt;https://karbonhq.com/feature/ai/&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Basis&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.getbasis.ai/" rel="noopener noreferrer"&gt;https://www.getbasis.ai/&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Basis for small firms&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.getbasis.ai/solutions/small-firms" rel="noopener noreferrer"&gt;https://www.getbasis.ai/solutions/small-firms&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Intuit, QuickBooks Online bank-transaction suggestions, updated August 26, 2026&lt;/td&gt;
&lt;td&gt;&lt;a href="https://quickbooks.intuit.com/learn-support/en-us/help-article/bank-transactions/ai-suggestions-help-match-categorize-bank/L8FHOh4AD_US_en_US" rel="noopener noreferrer"&gt;https://quickbooks.intuit.com/learn-support/en-us/help-article/bank-transactions/ai-suggestions-help-match-categorize-bank/L8FHOh4AD_US_en_US&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Blue J&lt;/td&gt;
&lt;td&gt;&lt;a href="https://www.bluej.com/" rel="noopener noreferrer"&gt;https://www.bluej.com/&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Microsoft, Data, Privacy, and Security for Microsoft Copilot&lt;/td&gt;
&lt;td&gt;&lt;a href="https://learn.microsoft.com/en-us/microsoft-365/copilot/microsoft-365-copilot-privacy" rel="noopener noreferrer"&gt;https://learn.microsoft.com/en-us/microsoft-365/copilot/microsoft-365-copilot-privacy&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Precision AI OS, pricing&lt;/td&gt;
&lt;td&gt;&lt;a href="https://getprecision.ai/pricing" rel="noopener noreferrer"&gt;https://getprecision.ai/pricing&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Precision AI OS, about&lt;/td&gt;
&lt;td&gt;&lt;a href="https://getprecision.ai/about" rel="noopener noreferrer"&gt;https://getprecision.ai/about&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Precision AI OS, AI implementation partner for CPA firms&lt;/td&gt;
&lt;td&gt;&lt;a href="https://getprecision.ai/resources/ai-implementation-partner-cpa-firms?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=best-ai-tools-for-cpa-firms-in-2027-4gpb" rel="noopener noreferrer"&gt;https://getprecision.ai/resources/ai-implementation-partner-cpa-firms?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=best-ai-tools-for-cpa-firms-in-2027-4gpb&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Precision AI OS, AI for CPA and accounting firms&lt;/td&gt;
&lt;td&gt;&lt;a href="https://getprecision.ai/industries/cpa-firms?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=best-ai-tools-for-cpa-firms-in-2027-4gpb" rel="noopener noreferrer"&gt;https://getprecision.ai/industries/cpa-firms?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=best-ai-tools-for-cpa-firms-in-2027-4gpb&lt;/a&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

</description>
      <category>ai</category>
      <category>accounting</category>
      <category>automation</category>
      <category>productivity</category>
    </item>
  </channel>
</rss>
