The Problem
Tax resolution is a deadline-driven business where the cost of a missed step isn't a rebooking — it's a federal lien, a levy, or a case that deteriorates while you're waiting on a client to dig up their 1040 from 2019. Most firms have figured out the technical work. What they haven't figured out is the intake process: the document chasing, the status calls, the client who calls three times a week because they're scared and don't know what's happening. That friction doesn't just slow your staff down — it delays the collection of information you need to file, which delays resolution, which delays getting paid.
- !Enrolled agents spending billable hours sending follow-up emails for missing POAs, transcripts, and prior-year returns
- !No centralized view of where each case stands in the intake pipeline — staff answers client calls without knowing what's been submitted
- !Clients in collections anxiety calling multiple times per week for updates your team doesn't have time to give
- !IRS correspondence deadlines tracked on spreadsheets or sticky notes, with no automated escalation when a date is approaching
- !New client intake relying on manual intake forms, PDFs, or phone calls that create inconsistent case files from day one
Where AI Fits In
AI in a tax resolution firm starts where the pain is loudest: intake and client communication. Automated document request workflows, AI-assisted triage of IRS notices, and a client-facing chat layer that answers status questions without pulling your staff off substantive work — these are the systems that create breathing room. Once intake is stable, the same infrastructure supports deadline tracking, engagement letter workflows, and ongoing client updates through resolution.
Most Common Starting Point
Most tax resolution firms start with AI-assisted client intake — structured onboarding flows that collect the financial disclosure information, prior-year returns, and IRS correspondence needed to open a case, with automated follow-up until the file is complete.
Structured Intake & Document Collection System
A guided onboarding flow that collects financial disclosure data, IRS correspondence, and prior-year returns — with automated follow-up sequences until the file is case-ready.
Client Communication Layer
An AI-backed messaging interface that answers common status questions, explains next steps in plain language, and escalates to staff only when a question requires professional judgment.
IRS Notice Triage Tool
Classifies uploaded IRS notices by type, extracts response deadlines, and routes them to the correct workflow — whether that's an OIC, an installment agreement, or a penalty abatement.
Deadline Tracking & Escalation Engine
Monitors active case deadlines across your portfolio and fires escalation alerts to assigned staff when critical IRS response windows are approaching.
Other Areas to Explore
Every tax resolution firm business is different. Beyond the most common use case, here are other areas where AI automation often delivers results:
Before You Automate Anything, Answer These Questions Honestly
AI will not fix a practice that doesn't know what it's doing yet. That sounds harsh, but it's the truth — and any vendor who tells you otherwise is selling you a deployment, not a result. Before you start looking at tools, run through these questions with whoever manages your operations.
Do you have a defined intake checklist? Not a general sense of what you need — an actual list, by case type, of every document and piece of information required before work begins. If your EAs are each running their own mental checklist, automation will just encode the inconsistency.
Are your resolution workflows documented? The AI needs to know what happens after intake. If your OIC workflow lives entirely in the head of your senior enrolled agent, you're not ready to automate handoffs.
Do you use a case management system consistently? Doesn't have to be fancy. Could be a well-maintained spreadsheet. But if case status information is scattered across email threads, sticky notes, and individual staff memory, the AI has nothing reliable to pull from when a client asks for an update.
- Disqualifier: You're still taking on any tax problem that walks in the door without a defined service menu. Automation requires repeatable process — and general tax chaos isn't a process.
- Disqualifier: Your staff turnover is high and roles aren't clearly defined. AI systems need a human on the other end who knows what to do when an escalation lands.
- Disqualifier: You haven't standardized your engagement letters, fee agreements, or retainer structures. If every engagement is negotiated from scratch, there's nothing to automate.
If you hit one of those disqualifiers, the move isn't to skip automation — it's to fix the underlying process first. A month of operational cleanup will make your eventual AI build ten times more effective. The firms that get the most out of this work are the ones who come in with documented workflows and a clear sense of where their staff time is actually going.
What the Chaos of Manual Intake Is Actually Costing You Per Case
The hidden cost of not automating intake isn't visible on a P&L line. It accumulates in ways that are easy to rationalize: a few extra calls here, a delayed document there, a case that sat in limbo for two weeks because nobody followed up on the financial disclosure statement. Add those up across an active caseload and you have a serious drag on both capacity and revenue.
The most expensive pattern is document delay. In tax resolution, cases can't move to substantive work — can't pull transcripts, can't calculate an RCP, can't draft an OIC — until the file is complete. Every day a case sits waiting on a missing 433-A is a day you're not billing and a day closer to a deadline you can't push back. (Source: Taxpayer Advocate Service, 2023 Annual Report to Congress) The IRS processed over 3.2 million balance-due cases in fiscal year 2022 — which means the volume of clients sitting in exactly this limbo, waiting on their own representation to move, is not small.
The second cost is your enrolled agents' attention. An EA billing at professional rates who spends 30 minutes a day on document-chase emails and status calls is not a minor inefficiency — it's a structural problem that gets worse as your caseload grows.
- Client churn risk: Anxious clients in IRS collections who don't hear from you regularly assume nothing is happening. They call. They get frustrated. Some of them call the bar association or file a complaint. Proactive communication isn't just good service — it's risk management.
- Capacity ceiling: Without automated intake, your growth is capped by how many cases your staff can manually shepherd through onboarding. Adding cases means adding headcount, not adding systems.
- Referral revenue lost: A client whose case resolved smoothly and who felt informed throughout refers people. A client who felt ignored during the process — even if the outcome was good — usually doesn't.
The firms that have systematized intake don't just run faster. They run with less staff friction, fewer missed deadlines, and clients who are calmer because they know what's happening. That's not a soft benefit — it's a practice that's easier to work in and easier to scale.
What AI Vendors Are Actually Selling Tax Resolution Firms (And What to Watch For)
The automation vendor space has discovered tax resolution, and not all of what they're pitching is built for how this work actually runs. Here's what to be skeptical of.
Generic "legal and financial" platforms that claim to handle tax resolution. Tax resolution has specific workflow requirements — IRS transcript pulls, CDP hearing timelines, OIC calculation logic, CAF number management. A platform built for general professional services intake will handle maybe 60% of your needs and create workarounds for the rest. Workarounds become the new manual process.
Chatbots positioned as client communication solutions. A bot that can answer "what is an Offer in Compromise" is not the same as a system that can tell your client where their specific case stands today. The former is a FAQ page. The latter requires integration with your actual case data. Make vendors show you the integration before you believe the demo.
According to the IRS Data Book, the agency received over 49,000 OIC applications in fiscal year 2022 — and acceptance rates remain well below 50%. (Source: IRS Data Book, 2022) Any vendor claiming their AI will improve your OIC acceptance rates through automation should be pressed hard on the mechanism. Acceptance depends on RCP calculations and negotiation strategy, not intake efficiency.
- Red flag: The vendor has never worked with an EA or seen an actual IRS notice. Ask them what a CDP hearing deadline is. Ask them what a 433-B is. If they don't know, they don't understand your compliance exposure.
- Red flag: The implementation plan is entirely self-serve. Tax resolution workflows have enough edge cases that you need someone who will actually configure the system to your case types, not hand you a login and a tutorial.
- Red flag: No clear answer on where PII goes. Your clients' financial disclosure data, Social Security numbers, and IRS correspondence are highly sensitive. Any vendor that can't explain their data handling, encryption, and access controls clearly is not a vendor you should trust with this data.
Oaken builds on a stack that includes Presidio for PII detection and redaction — because in this business, a data handling mistake isn't just embarrassing, it's a liability that follows you.
How It Works
We deliver working systems fast — no multi-month assessments, no slide decks. A typical engagement runs 3-5 weeks from kickoff to live system.
Weeks 1-2
Audit current intake process and case management workflow. Map document collection gaps, identify the highest-volume client communication triggers, and configure the intake system against your actual case types — OIC, installment agreement, CDP, penalty abatement.
Weeks 3-4
Deploy the client intake flow and document request automation. Connect to your existing case management or CRM. Test IRS notice classification against a sample of real correspondence from your files.
Week 5
Activate the client communication layer and deadline escalation engine. Train staff on handoff protocols — what the AI handles, what gets escalated, and how to override. Run parallel with existing process before full cutover.
The Math
Staff hours recovered from intake coordination and client status calls
Before
EAs chasing documents and answering 'what's the status?' calls instead of working cases
After
Cases open complete, clients informed automatically, staff focused on IRS strategy
Common Questions
Can AI actually handle IRS correspondence, or does it still need an enrolled agent to review everything?
AI handles classification and routing — identifying notice type, extracting deadlines, flagging urgency, and dropping the notice into the right workflow. The substantive response strategy still requires your EA. What you're eliminating is the manual sorting, the 'what is this notice even asking for?' step, and the risk that a CP503 sits in an email inbox for three days before someone looks at it.
We use [specific case management software] — will AI tools integrate with it?
It depends on the platform and whether it has a usable API or data export. Most major tax resolution and tax practice management platforms have some integration path. Before any build starts, Oaken audits your current tech stack and identifies where data lives, what can be connected, and where we need to build a bridge. We don't assume clean integration — we verify it.
Our clients are already nervous about IRS issues. Won't adding AI to their experience make it worse?
The opposite, done correctly. Anxious clients want to hear from you more, not less. An automated status update that tells a client their POA has been filed and the next step is transcript retrieval does more for their anxiety than a weekly manual call your staff doesn't have time to make. The goal isn't to replace human communication — it's to make sure clients aren't sitting in silence between touchpoints.
What happens to client data — Social Security numbers, financial disclosures, IRS notices?
This is the right question to ask any vendor. Oaken uses Presidio, an open-source PII detection and redaction framework from Microsoft, as part of the data handling pipeline. Client data is encrypted in transit and at rest, access is role-based, and sensitive information is never used to train models. We can walk through the architecture in detail before any engagement begins — because your clients trusted you with this data, not a vendor they've never heard of.
We're a small firm — two EAs and an admin. Is this still worth doing?
Possibly more so than at a larger firm. At two EAs, every hour lost to intake coordination and status calls is a direct hit to capacity. The build scope would be narrower — focused on intake automation and client communication rather than a full enterprise workflow — but the leverage per hour recovered is higher. The right conversation starts with where your staff time actually goes right now.