How it works

RCA posted on every ticket before your engineer picks it up

BuildWright works inside your ticketing tool and diagnoses every escalation before an engineer opens it. Diagnosis time drops by up to 95%.

Book a demo

SUP-128InvestigatingDiagnosed

Item-wise tax breakup does not match the total tax

BuildWright posted a diagnosis

High grounding

Root cause found

File: taxes_and_totals.py
Method: set_item_wise_tax()
Lines: 540–556

Per-line rounding makes the item-wise breakup drift from the total tax on some rate × quantity combinations.

Recommended Review the rounding threshold in set_item_wise_tax().

The flow

Stop paying your best engineers to investigate tickets

BuildWright reclaims 770+ hours of engineering capacity lost to L3 support and escalations.

00
Step zero: Ask BuildWright
End users ask BuildWright inside your product. Simple questions are answered on the spot. A ticket is raised only when needed.
Ask BuildWrightOnline
Why does the tax on each line of my invoice not add up to the total?
Each line is rounded separately, so the line totals can differ from the header total by a few cents. The header total is the correct amount to pay.
Open the invoice and check the header total
Use the header total for payment and reconciliation
Still blocked? Raise a ticket with this answer attached
Ask a questionEnter
01
Ticket arrives
The ticket lands in Zendesk, Salesforce, ServiceNow, Jira or wherever your team works. A rule assigns it to BuildWright.
02
BuildWright investigates
2.1
Trace the code
Read-only access to the code. It names the exact file and method.
2.2
Connect the evidence
Logs and events tied to the customer's symptom, so nobody jumps between six browser tabs.
2.3
Check past resolutions
A problem fixed at one customer is solved faster at the next.
2.4
Post the root cause
The RCA lands on the ticket with evidence, next steps and a confidence grade. Slack alerts your team.
03
Unblock the customer
Engineers check the evidence and ship the fix. On harder cases, they loop BuildWright back in.

Levels of trust

Critical decisions stay in your engineers’ hands

Read-only, always. Your team decides how much it does on its own.

Shadow

Drafts only. Nothing posts. Your team watches how it reasons.

Copilot

A person approves every diagnosis before it posts.

Autopilot

Posts on its own for proven case types. Still flags every ticket that needs a human.

Honest grounding

A tool that says “I don't know”

Every diagnosis is graded on its evidence. When the evidence isn’t there, BuildWright says so and asks for what it needs.

High

The evidence fully supports it. Ready to act on.

Partial

Some evidence supports it. Every gap is named.

Ungrounded

No evidence. BuildWright asks for more.

SUP-214Ungrounded

BuildWright could not support a diagnosis

Not enough evidence to name a root cause

no indexed code repository

Connect the repository for this service, or add the error log from the time of the report, and I will investigate again.

Benchmark

BuildWright correctly traced a real customer-reported bug to lines 540 to 556 of the set_item_wise_tax() method in taxes_and_totals.py, using only the customer’s symptoms on an ERPNext bug.

See the benchmark

Integrations