Turn AI token spend
into R&D tax credits.

Audit-ready R&D Tax Credit study and Capitalized Software ASC 350-40 workpapers for enterprise Tax and Accounting teams.

Book a demo See how it works

Example Co./Projects/Payments ledger service

FY2025

Jira

Payroll

Anthropic

Project

46,200hours

1,318 tickets

=$6.0M

38 people

+$1.9M

494B tokens

=$7.9M

R&D spend

R&D Tax Credit$6.4MQREs§41

Software Capitalization$1.8MCapitalizedASC 350-40

R&D InsightsEarly access3.4×ROIReturn by project

Projects/Payments ledger service

Jira46,200hours
1,318 tickets in PAY and LEDG
Payroll=$6.0M
38 people, about $130 an hour
Anthropic+$1.9M
494B tokens on the payments-agent key
R&D spend=$7.9M
R&D Tax Credit$6.4MQREs
Software Capitalization$1.8MCapitalized
R&D InsightsEarly access3.4×ROI

R&D Tax Credit

Eliminate surveys and interviews
Built from the work as it happened
A more rigorous study in far less time

Software Capitalization

Journal entries on day zero
Time calculated from tickets and payroll
An audit-ready export traced to source

R&D Insights

Where engineering time and AI spend went
Each project next to the priorities you set

Trusted by 150+ companies.

Read customer stories

Every project in one place. Each hour and AI token traced to its project.

ProjectsFY2025 · 109,100 hours · 106 people

ProjectsProjects: 5
EngineeringEngineering time: $13.9M
AI spendAI spend: $6.3M
QualifiedQualified research: $14.9M
CapitalizedCapitalized: $3.9M
ReturnReturn on R&D: 2.9×

Payments ledger service
Platform · 46,200 h · $7.9M
$6.4M
Qualified
Fraud model v3
Risk ML · 25,600 h · $5.9M
$4.9M
Qualified
Mobile onboarding
Growth · 21,400 h · $3.0M
$1.9M
Qualified
Support agent
CX · 9,800 h · $2.6M
$1.7M
Qualified
Admin tooling
IT · 6,100 h · $0.8M
$0
Not qualified

Project

Engineering time

AI spend

Tax

§41 · Form 6765

Accounting

ASC 350-40

FP&A

Return · plan

Payments ledger service
Platform · 38 people
Engineering time
46,200 h$6.0M
AI spend
$1.9M
Tax
$6.4M
Qualified
Accounting
$1.8M
Capitalized
FP&A
3.4×
Plan 4 · Spend 1
Over plan
Fraud model v3
Risk ML · 21 people
Engineering time
25,600 h$3.3M
AI spend
$2.6M
Tax
$4.9M
Qualified
Accounting
$0
Expensed
FP&A
2.1×
Plan 1 · Spend 2
Mobile onboarding
Growth · 26 people
Engineering time
21,400 h$2.7M
AI spend
$0.3M
Tax
$1.9M
Qualified
Accounting
$1.2M
Capitalized
FP&A
4.6×
Plan 2 · Spend 3
Support agent
CX · 12 people
Engineering time
9,800 h$1.2M
AI spend
$1.4M
Tax
$1.7M
Qualified
Accounting
$0.91M
Capitalized
FP&A
1.8×
Plan 3 · Spend 4
Admin tooling
IT · 9 people
Engineering time
6,100 h$0.7M
AI spend
$0.1M
Tax
$0
Not qualified
Accounting
$0
Expensed
FP&A
Internal
Plan 5 · Spend 5

TaxForm 6765

$8.1M

qualifies for the R&D credit, of $11.3M spent

Payments ledger service
Platform · $7.9M spent
$6.4M
Qualified
Support agent
CX · $2.6M spent
$1.7M
Qualified
Admin tooling
IT · $0.8M spent
$0
Maintenance

AccountingWorkpaper C-3

$2.7M

capitalized as software, of $11.3M spent

Payments ledger service
Platform · $7.9M spent
$1.8M
Capitalized
Support agent
CX · $2.6M spent
$0.91M
Capitalized
Admin tooling
IT · $0.8M spent
$0
Expensed

FP&ASpend vs plan

3.0×

return on R&D spend that drives revenue

Payments ledger service
Platform · $7.9M spent
3.4×
Plan 4 · Spend 1
Support agent
CX · $2.6M spent
1.8×
Plan 3 · Spend 4
Admin tooling
IT · $0.8M spent
Internal
No revenue line

Audit-ready study and workpapers.Every team's document from one set of data.

Neo.TaxIllustrative

Federal Research
Credit Study

Qualified research expenses under IRC §41, tax year ended December 31, 2025

Prepared for: Example Co.
Prepared by: Neo.Tax
Issued: Oct 14, 2026

1Executive summary3
2Methodology and data sources6
3Business components11
4QRE computation2,214
5Form 6765 support2,871

Example Co. · Research credit study · TY20251 of 2,943

Form 6765 (Rev. 12-2024)Tax year 2025

Section GBusiness Component Information. Example Co.

1Payments ledger service

2Fraud model v3

3Mobile onboarding

4Support agent

BC	50Qualified Direct Wages	51Qualified Supervision Wages	52Qualified Support R&D	53Total Qualified Wages	54Qualified Supplies Costs	55Qualified Computer Rental or Lease Costs^a	56Qualified Contract Research Costs
1^b	4,626,180	395,260	177,800	5,199,240	1,160	1,086,900	112,000
2	2,597,040	221,780	99,200	2,918,020	580	1,964,200	0
3	1,580,210	134,940	60,360	1,775,510	290	104,700	48,500
4	803,990	68,650	30,715	903,355	145	788,100	0

aLine 55 carries AI API usage in qualified research that meets the computer rental rules.bBC 1 traces to 1,318 tickets in PAY and LEDG and 38 people in payroll.

Name(s) shown on return: Example Co.Generated by Neo.TaxForm 6765 (Rev. 12-2024)

Example Co.

Capitalized software, internal use

Application development stage costs, FY2025

W/P C-3Prepared NT 10/3/25

Project · ASC 350-40 stage	Hours	Labor	AI	Capitalized
Payments ledger serviceIn development	11,510^a	1,496,820^b	287,480^c	1,784,300
Fraud model v3Expensed, research				Expensed
Mobile onboardingPlaced in service	8,790^a	1,108,240^b	107,360^c	1,215,600
Support agentIn development	4,190^a	512,350^b	394,550^c	906,900
Admin toolingExpensed, maintenance				Expensed
Total	24,490^F	3,117,410^F	789,390^F	3,906,800

Journal entryJE-4471 · Oct 3 · Support W/P C-3Ready to post

17150	Capitalized software, internal use	3,906,800.00
60110	Salaries and wages, R&D		3,117,410.00
61420	AI and cloud services, R&D		789,390.00

aAgreed to Jira worklogsbAgreed to the ADP payroll registercAgreed to API usage by keyFFooted

Example Co.

Capitalized labor by person

Payments ledger service, FY2025

W/P C-2Prepared NT 10/3/25

Person · role	Total cost	Cap. %	Capitalized
A. OkaforStaff engineer	412,000^a	41%^b	168,920
K. BrennanEngineer	391,130^a	38%^b	148,629
M. ReyesEngineer	344,500^a	35%^b	120,575
S. PatelEng. manager	368,200^a	22%^b	81,004
J. LindqvistProduct manager	296,800^a	18%^b	53,424
33 others	4,187,370	22%	924,268
Total	6,000,000^F	25%	1,496,820^C-3

aAgreed to the ADP payroll registerbRecalculated from Jira worklogsFFootedC-3Agreed to W/P C-3

Neo.TaxIllustrative

Federal Research
Credit Study

Qualified research expenses under IRC §41, tax year ended December 31, 2025

Prepared for: Example Co.
Prepared by: Neo.Tax
Issued: Oct 14, 2026

1Executive summary3
2Methodology and data sources6
3Business components11
4QRE computation2,214
5Form 6765 support2,871

Example Co. · Research credit study · TY20251 of 2,943

Form 6765 (Rev. 12-2024)Tax year 2025

Section GBusiness Component Information. Example Co.

1Payments ledger service

2Fraud model v3

3Mobile onboarding

4Support agent

BC	50Qualified Direct Wages	51Qualified Supervision Wages	52Qualified Support R&D	53Total Qualified Wages	54Qualified Supplies Costs	55Qualified Computer Rental or Lease Costs^a	56Qualified Contract Research Costs
1^b	4,626,180	395,260	177,800	5,199,240	1,160	1,086,900	112,000
2	2,597,040	221,780	99,200	2,918,020	580	1,964,200	0
3	1,580,210	134,940	60,360	1,775,510	290	104,700	48,500
4	803,990	68,650	30,715	903,355	145	788,100	0

aLine 55 carries AI API usage in qualified research that meets the computer rental rules.bBC 1 traces to 1,318 tickets in PAY and LEDG and 38 people in payroll.

Name(s) shown on return: Example Co.Generated by Neo.TaxForm 6765 (Rev. 12-2024)

Example Co.

Capitalized software, internal use

Application development stage costs, FY2025

W/P C-3Prepared NT 10/3/25

Project · ASC 350-40 stage	Hours	Labor	AI	Capitalized
Payments ledger serviceIn development	11,510^a	1,496,820^b	287,480^c	1,784,300
Fraud model v3Expensed, research				Expensed
Mobile onboardingPlaced in service	8,790^a	1,108,240^b	107,360^c	1,215,600
Support agentIn development	4,190^a	512,350^b	394,550^c	906,900
Admin toolingExpensed, maintenance				Expensed
Total	24,490^F	3,117,410^F	789,390^F	3,906,800

Journal entryJE-4471 · Oct 3 · Support W/P C-3Ready to post

17150	Capitalized software, internal use	3,906,800.00
60110	Salaries and wages, R&D		3,117,410.00
61420	AI and cloud services, R&D		789,390.00

aAgreed to Jira worklogsbAgreed to the ADP payroll registercAgreed to API usage by keyFFooted

Example Co.

Capitalized labor by person

Payments ledger service, FY2025

W/P C-2Prepared NT 10/3/25

Person · role	Total cost	Cap. %	Capitalized
A. OkaforStaff engineer	412,000^a	41%^b	168,920
K. BrennanEngineer	391,130^a	38%^b	148,629
M. ReyesEngineer	344,500^a	35%^b	120,575
S. PatelEng. manager	368,200^a	22%^b	81,004
J. LindqvistProduct manager	296,800^a	18%^b	53,424
33 others	4,187,370	22%	924,268
Total	6,000,000^F	25%	1,496,820^C-3

aAgreed to the ADP payroll registerbRecalculated from Jira worklogsFFootedC-3Agreed to W/P C-3

See what Neo.Tax can do for your team.

Book a demo

Tax leaders get their time back.

Read customer stories

“Filing this first year took at least 50% less time.”

Karin Barratt VP of Tax, Octane

“Neo.Tax significantly reduced the number of interviews we conducted and helped us eliminate a major cost with an external advisor.”

Sean Cannizzaro VP, Head of Tax Operations, Thomson Reuters

“Our CFO loved how unbelievable the ROI was. This is the only AI we've used that actually worked.”

Raymond Salort VP & Director of Taxes, Red Ventures

Built to hold up at audit. Every number traced to the ticket, hour, and token behind it.

PAY-12031

Benchmark 3 ledger partition designs

Daniel Okafor · Opened Mar 4, 2025

LEDG-4472

Reconcile double entries under retries

Marcus Bell · Opened May 19, 2025

LEDG-4523

Roll back failed batch postings

Priya Raman · Opened Aug 7, 2025

Research credit study · TY2025p. 214

Business component 1

Payments ledger service

Technical uncertainty

The team was uncertain how to settle payments consistently under concurrent writes.

Process of experimentation

Benchmarked 3 ledger partition designs under production load.¹
Reconciled double entries when settlement retries fired.²
Rolled back failed batch postings until settlement held at 5k TPS.³

Sources: 1,318 tickets in PAY and LEDG · 38 people on payroll

Research credit study · TY2025p. 214

Business component 1

Payments ledger service

Technical uncertainty

The team was uncertain how to settle payments consistently under concurrent writes.

Process of experimentation

Benchmarked 3 ledger partition designs under production load.¹
Reconciled double entries when settlement retries fired.²
Rolled back failed batch postings until settlement held at 5k TPS.³

Sources: 1,318 tickets in PAY and LEDG · 38 people on payroll

PAY-12031
Benchmark 3 ledger partition designs
Daniel Okafor · Opened Mar 4, 2025
LEDG-4472
Reconcile double entries under retries
Marcus Bell · Opened May 19, 2025
LEDG-4523
Roll back failed batch postings
Priya Raman · Opened Aug 7, 2025