Case management for a tax practice, built and run.
We have run the operations platform of a US tax firm for over three years. What that means
concretely: a deeply customised org supporting case management across multiple tax types including
1040, 1041, extensions and amended returns, automated client email, text message notification,
calendar integration, opportunity lifecycle management, a missing document workflow and billing
sheet automation. Cadence moved from ad hoc, to fortnightly, to twice weekly eleven minute check-ins
during tax season, because that is what busy season requires.
The problems that engagement was built to solve are the ones this industry keeps describing.
Reminders sent by hand because a batch process cannot send a text message. Records kept up to date
manually with a request that it be automated every three business days. No single place connecting a
client across tax years. No way to see which documents have arrived and which are still
outstanding.
Named, published, on the record
Bill Dooner, Director, Taxfirm.com.
"It was extremely challenging the transition to 'Remote Staff', which Micronetbd helped
us tremendously with it. A lot of tweaking and revisions can be done in real-time while we are on the
call with Micronetbd, many of the easy quick fixes are done in real-time and we are very appreciative
because that's what makes Micronetbd stand apart from other companies."
Bill DoonerDirector, TaxFirm.com
AccuFirst, Inc. and TaxFirm.com have been a Micronetbd client for over three years. This
testimonial is published on micronetbd.org.
The pack composes onto the whole ground, the whole spine and six jobs, and connects the practice
CRM, the document store, the accounting system and the messaging layer. Status: first customer
build. The cost meter underneath it is live in production today.
Shipped Live today, running for a real customer on real data
First customer build The baseline exists; the industry specific part would be built with you
Delivered in your implementation In build, not available yet
The three capability groups
Pack 4
Shipped
Per return economics you can actually read
Objective. Give a practice the cost of AI per return, per preparer and per engagement,
from a meter no model call can bypass.
One wrapper around every model call records the model, the tokens and the dollar cost. The
provider accounts are in the firm's name, billed to the firm at published rates, and the firm
sees the raw invoices. We add nothing to them.
Requirement: the market condition rather than one client. Organisation wide
AI use in professional services reached 40% in 2026 while only 18% track return on investment.
This buyer has already bought AI and cannot prove the return.
Delivered in your implementation
The reminder that sends itself
Objective. Remove the recurring manual send that a person currently performs on a
schedule.
The outstanding items queue drives the reminder and the draft assembles from the firm's own
language, and a person approves before it goes. Note the honest boundary: the platform does not
transmit today, so this capability depends on outbound dispatch landing.
Requirement: AccuFirst and TaxFirm.com, who asked for the reminder run to
be automated every three business days rather than performed by a person going through the
process of sending the emails and the text messages by hand.
First customer build
The missing document workflow
Objective. Track what has been received against what is still needed, per client and
per return, without a spreadsheet.
The document catalogue reads what arrived, the completeness audit says what is missing, and
the queue sorts by how long it has been outstanding.
Requirement: the published Tax and Accounting case study on micronetbd.org,
which describes a practice with no system for tracking documents received against documents
pending.
The stack
The systems a firm this size actually runs.
Salesforce, or whatever holds your client records.
The platform that holds your client and engagement records, mirrored into a database that
belongs to your tenant alone with count parity verified against the source.
Outlook and Microsoft 365, or Google Workspace.
So the email thread that carries the missing document is read alongside the engagement record
rather than separately from it.
QuickBooks and accounting systems.
Connected through a sync layer, so a fee, an engagement and a client resolve to one another.
Document portals and secure file exchange.
So what has arrived and what is still outstanding is a question with an answer.
Text messaging and scheduling tools.
Read today so the reminder queue is complete. The send itself still waits for outbound
dispatch, which is in build for fall 2026.
Anything else with a documented API.
Every write back into a source system is staged, whitelisted and approved by a person, and it
enforces the calling user's own permissions in that system.
We publish the cost per unit of work, because you can do the arithmetic yourself.
$0.0046
Measured cost of one AI document summary
$0.31
Approximate cost to process a 287-page binder
$0.00
Per page cost of optical character recognition, because it runs on our own hardware
$156.55
Total model spend to read, index and summarise a 44,956 document, 331,486 page corpus
Those are the real line items from a live production tenant, not
projections. A practice leader can put their own return volume against them on this page, without
talking to us. That is the point.
The subscription is a published number. The model consumption is a bill your firm receives
directly from your provider, in accounts held in your firm's own name, at that provider's published
rates. We add nothing to it, which means what we make does not move when your usage does. In a
market where 40% of firms have adopted AI and 18% can measure it, being the vendor whose bill is
legible is a position worth holding.
Do not overclaim depth from one relationship. One practice, three years, described specifically
rather than pluralised.
One client, not a plural claim. We have one accounting and tax client. Three years, a
named public testimonial, and a deeply customised platform we run every day. That is depth, and it
is real. It is not breadth and we are not going to write it as though it were.
The most strategic thing that client asked for is not delivered. AI driven client
insight across the practice has been in progress since September 2025 and lost its primary
technical resource in November 2025. Mayeshal is the platform we would deliver it on, and the
reason we are comfortable saying that out loud is that the underlying capability, reading every
document and every email thread and grounding an answer in a citation, is shipped and running in
production today for a customer in another industry.
There is no accounting vertical pack. The baseline platform is what exists, and the
pack would be built with you as the first customer.
The platform does not transmit. Reminders, notices and client letters stop at a
finished draft that a person sends. Outbound dispatch is in build for fall 2026, and when it lands
it will go out through your own system so the send is logged on your record rather than in
ours.
Get started
See it read one of your own matters.
Thirty minutes, on a screen share, on a matter you pick. We connect nothing, install nothing
and ask for no credentials to have that conversation. If it is not obviously useful in the first
ten minutes, we will tell you so and you get your afternoon back.
