Atomiq logoAtomiq
Case studies

Work that survived contact with production. Every number below is measured against a baseline we agreed with the client before the build started.

Three engagements in full — what the problem actually was, what we built, and what changed afterwards.

B2B SaaS

68%

of support tickets resolved without a human

A B2B SaaS platform, ~400 staffAI Automation

Support agent with a real eval harness

A support team of thirty-one was absorbing a ticket volume growing faster than headcount could follow. An earlier chatbot pilot had been quietly switched off after it started confidently inventing billing policy. We rebuilt the thing around measurement first: a graded set of twelve hundred real tickets, then a retrieval-backed agent scored against that set every night.

Results

of tickets resolved with no human touch
68%

of tickets resolved with no human touch

median first response
12 hrs → 40 min

median first response

answers escalated as incorrect, down from 9%
0.4%

answers escalated as incorrect, down from 9%

from scoping to production cutover
7 weeks

from scoping to production cutover

Read the full story

Financial services

9 days → 4 hrs

turnaround on credit memo drafting

A commercial lender, mid-marketAI Automation

Document workflow for an underwriting team

Credit memos took nine working days, almost all of it analysts retyping figures out of PDFs into a template. We built structured extraction over the source documents and a review interface that puts an analyst on every field before it lands — because in this sector an unreviewed number is not a saving, it is a liability.

Results

median memo turnaround
9 days → 4 hrs

median memo turnaround

of extracted fields accepted without correction
94%

of extracted fields accepted without correction

memo throughput per analyst
3x

memo throughput per analyst

of fields traceable to a source document
100%

of fields traceable to a source document

Read the full story

Healthcare

£1.7M

annualised operating cost removed

A four-site private clinic groupAI Automation, then AI Enablement

Intake triage across four clinics

Four sites, four different intake habits, and a shared referral inbox that ran two to three days behind. We built classification and routing on top of the record system already in place, under clinical governance review throughout, then trained the intake teams to run it themselves.

Results

annualised operating cost removed
£1.7M

annualised operating cost removed

referral acknowledgement
2–3 days → 20 min

referral acknowledgement

on one written triage standard
4 sites

on one written triage standard

urgent referrals missed since cutover
0

urgent referrals missed since cutover

Read the full story
Method

How to read these numbers. A results page is only worth reading if you know how the results were arrived at.

Baselined before the build

Every figure is measured against a number we agreed with the client at scoping. Where no baseline existed, our first job was to establish one.

Measured after go-live, not at demo

Results are taken from at least a full quarter of production running, so seasonal effects and the novelty period are both in the sample.

Anonymised where the client asks

Most of this work sits under NDA. Sectors and shapes are accurate; names appear only where we have written permission.

If you would like to speak to a reference before starting, ask on the first call. We will introduce you to someone who has run one of these systems for at least six months.

Get started

Let’s find out what AI is actually worth to you.

A 30-minute call, no deck. Bring the workflow that frustrates you most and we will tell you whether it is worth automating.

Typically replies within one business day.