Evidence with a history
Preserve original sources, distinguish proposed information from established facts, and record why something changed.
THE NEXT ENGINE
Individual returns deserve their own architecture. The same preparation philosophy will meet a new domain: households, people, income, deductions, credits and the relationships between them.
Architecture direction · S-corporation preparation is the current working engine.
THE FIRST PROVING GROUND
1120-S
Books, owners, assets, basis, corporate reporting and a professional return package.
Explore the S-Corp EngineShared principles.
Distinct domains.
A DIFFERENT STARTING POINT
The S-corp engine taught us useful things about preparation integrity. It also contains assumptions that belong to a corporation: books, shareholder allocations, corporate tax accounts and entity-specific review questions.
A household should not have to become a corporation in order to use those lessons. Individual preparation starts with people and their relationships, then connects income, expenses, payments, activities and prior history to the right person and tax year.
The direction is a separate individual engine. Evidence, review and reproducibility carry forward; the domain model, calculations and form applicability must be designed and researched for individual returns.
WHAT CARRIES FORWARD
The strongest ideas are commitments to the professional, not assumptions about a particular form.
Preserve original sources, distinguish proposed information from established facts, and record why something changed.
Keep missing, zero, not applicable, conflicting and unsupported information meaningfully different.
Use reviewed facts, matching-year rules, defined money handling and independently tested calculations.
Keep direct workpapers, actual forms, diagnostics and saved history connected to the same preparation.
WHAT CHANGES
A return is a collection of related tax items, not one enormous form-shaped object.
The proposed model gives people and items stable identities. Review, source history and changes can then follow the item they actually concern.
Taxpayer, spouse, dependents and other relevant people need identities, relationships and effective periods. A filing-status decision should be supported by facts, not redefine the people underneath it.
Wages belong to a person and payer. Investment items retain their character. Withholding and other payments keep their own source and year so they can be reviewed without being counted twice.
Eligibility, expense facts, income interactions and limitations need explicit dependencies. A credit is a determination built from evidence, not simply another amount field.
Business interests, rentals and carryforwards need their own ownership, character and origin. A single universal “loss balance” would lose the distinctions the next year needs.
A LESSON WORTH CARRYING FORWARD
A preparer may establish a fact from a workpaper, a client discussion, a reconstruction or a professional determination. The new engine should make that manual path complete from the beginning.
Optional imports can propose information into the same review process. A failed parser should leave a supported manual preparation usable. A replacement document should not silently overwrite a fact the preparer established later.
Original evidence, extracted claims, preparer assertions and the selected fact should remain distinguishable. That is an architectural priority for the next engine, informed by the real source-authority problems encountered in the first.
The evidence philosophyBUSINESS TO OWNER
The long-term goal is a structured shareholder package that can move from a saved S-corp preparation into the individual engine. Reported K-1 information, owner-specific support and the identity of the originating preparation would cross a documented boundary.
The individual preparer would review the match and acceptance, supply personal facts, and address the applicable individual limitations. Corporate QBI information is not the household’s final deduction; a corporate reporting amount is not automatically the final personal result.
A corrected source package would create an explicit replacement for review. It would not silently rewrite an accepted individual return. Each engine should also remain usable with manually established information and material from other sources.
Recipient-specific reporting and supporting information.
Identity, period, version, changes and unresolved facts.
Household facts, personal limitations and 1040 output.
Proposed future interchange. A live engine-to-engine integration is not available today.
HOW TO BEGIN
The proposed first proof is deliberately narrow: a simple individual return with ordinary wages, bank interest, a standard deduction and withholding. The goal is a complete visible path from a blank case through reviewed facts, calculations, workpapers and actual matching-year forms.
That foundation should demonstrate corrections, explicit missing facts, saved history and output review before intake automation or broad tax coverage is added. A manually prepared return is a real architectural test.
Later expansion can introduce joint returns, dependents, investments, deductions, retirement items, K-1s and business activities in deliberate stages. Each adds new relationships and new regression obligations. State interactions need their own treatment; storing jurisdiction information does not establish state-return support.
There is no promised launch date or claim of a working 1040 engine today. The direction is clear: S corporations first. Individuals next. One preparation philosophy.
Follow the project directionWe welcome practical conversations about household models, source authority, professional review and the business-to-owner connection.