Four agents,
accepted on
their numbers
Scroll to move through them, or skip ahead.
Agents on your own data
Accepted on measured numbers
AI that takes the repetitive work off your team.
Agents built on your own data, accepted against contractual accuracy and latency targets. Not demos.
Book a free call30 minutes · free · no preparation needed
Your specialists spend their day retrieving, not
The problem
-
01
Incident triage eats engineering hours
Dashboards, logs, wikis and runbooks, opened by hand, every alert.
-
02
Every question becomes a ticket
Analysts queue behind engineers for answers a query could give.
-
03
Routine requests flood your specialists
High-volume work buries the cases that actually need a human.
Most pilots
never ship
The demo works, real data doesn't
Messy documents, scans and edge cases break what looked perfect on stage.
Nobody measures accuracy
With no test set, "it feels right" is the only evidence anyone can offer.
Privacy blocks the rollout
Legal says no once they ask where the data actually goes.
Nobody owns it after launch
Models, prompts and sources drift, and quality quietly decays.
We write the numbers into the
What we do
AI agents on your own data
Retrieval, reasoning and action over your logs, documents, tickets and databases, with cited sources and a human approval step where it matters.
Automation and data plumbing
The pipelines, integrations and workflow automation that agents need underneath them, and that pay for themselves on their own.
- 01Paperwork, read by softwareContracts, invoices and scans into checked fields
- 02Tickets and incidents, sortedRoutine cases closed, the rest handed over
- 03Answers from your own documentsCited to the page, or a correct decline
- 04Pipelines that run themselvesIntegrations and automations, timed and audited
- 05Numbers for Monday morningDashboards, and SQL you read before it runs
- 06Software that reads picturesPhotos and footage checked, a person where unsure
- 07Tools, sites and screensInternal tools, pages that sell, screens before code
Each one is shown the way it arrives, on the services page. Most clients start with one agent on one painful workflow, then expand once the numbers hold.
Accuracy goes
in the contract
Every agent we ship is accepted against a written test set and a service level. If it doesn't hit the numbers, it isn't done.
01p95 latency
Time to a complete, grounded answer, measured in production, not in a notebook.
02Accuracy
Scored against a labelled set of real cases your own experts agreed on.
03Grounding
Share of claims traceable to a real source document, log line or record.
04Escalation rate
How often the agent correctly hands a case to a human instead of guessing.
Latency distribution
p95 target · under 15s
p95 means 95 of every 100 answers land left of that line. The shape here is illustrative. The threshold is the number we sign.
The artefact
This is what
you get signed
Every build ends with an acceptance report against your own labelled test set. If a row says FAIL, the agent is not accepted and we keep working. Below is the format that document takes, filled with representative figures from an incident-triage engagement.
Acceptance report
Incident triage agent
Specimen
| Metric | Target | Measured | Result |
|---|---|---|---|
| p95 latency | < 15s | 11.4s | Pass |
| Accuracy | ≥ 0.75 | 0.78 | Pass |
| Grounding | ≥ 0.80 | 0.82 | Pass |
| Escalation rate | ≤ 0.15 | 0.11 | Pass |
Targets are agreed during the first week, before a line of code is written.
01 / 04
Four agents,
Regulated finance
One wrong digit is unacceptable
Risk analysis over contracts, reports and transactions, pulling out counterparties, dates, amounts and banking details from a mix of clean text and scans, across inconsistent legal formats and synonyms.
- Hybrid design: deterministic rules for structured fields
- Fine-tuned transformer for organisations and document IDs
- Banking details validated against official directories
- Low-confidence documents routed to human review
~99%
precision on banking details, where errors are unacceptable
>95%
recall across the tracked entity types
-90%
target reduction in document processing time
API+UI
integrated into the existing back office
Customer support
10,000 tickets a day, sorted
The challenge was never volume alone, it was sorting it. Closing the routine cases automatically is only safe if the agent also knows, reliably, which cases it must not touch.
- Answers grounded in your help centre and past tickets
- Confidence thresholds that decide close vs. escalate
- Clean handover, the specialist sees what the agent found
- Every escalation logged and scored, so the split improves
- 60%Routine, closed by the agent
- 30%Complex, routed to a specialist
- 10%Edge cases, held for human review
Data engineering
Root cause in under a minute
A data team running 200+ ETL pipelines. Every failure cost an engineer 30 to 45 minutes of manual triage across Grafana, ClickHouse logs, Confluence and runbooks, before any fixing started.
- Agent triggered by the monitoring alert itself
- Retrieval across logs, metrics, runbooks and past incidents
- Structured RCA: cause, components, actions, sources
- Slack entry point for follow-up questions
30-60s
to a full RCA summary, down from 30 to 45 minutes
<15s
p95 latency to a complete grounded answer
0.78
accuracy against the labelled incident set
0.80
grounding, claims traced to a real source
Analytics
Analysts stopped queuing
A BI platform with 50+ dashboards. Every "can you add a chart for this segment?" became a two-day ticket for data engineering, and there were more of them every week.
- Natural-language to SQL over the live warehouse schema
- Handles filters, time windows and period-over-period growth
- Generated SQL shown to the analyst before it runs
- Guardrails on cost, row limits and table access
Minutes
to answer a new question, down from a two-day ticket
50+
dashboards the agent can reason across
0
engineering tickets for routine chart requests
Self
analysts serve their own questions
First call to
production
in weeks
01
Audit
1 weekWe pick the workflow, build the test set with your experts, and agree the SLA and the price of the build.
02
Build
4 to 12 weeksWeekly demos on your real data. Accuracy tracked against the test set from day one.
03
Accept
1 weekGo-live only when the agreed numbers are met. Your team trained, everything documented.
04
Run
ongoingMonitoring, drift checks, retraining and new use cases on a monthly retainer.
You can stop after any stage. Nothing is locked in.
Trust
Your data stays yours
GDPR, signed DPA, EU hosting or a fully self-hosted deployment.
You own the code
Full IP transfer on final payment. No lock-in, no black boxes.
Accuracy is measured
Labelled test sets and a written acceptance report, every time.
Humans stay in the loop
Every action that carries risk waits for a person to approve it.
What it costs.
It starts with a free call. The first week after that is a flat €700 and ends with a test, a plan and a price. Everything beyond is quoted against the job we agreed, so the figures below are starting points, not estimates you have to guess at.
Start here
A call
We find out what is wasting your team's time and whether an agent is the right fix. No preparation, no obligation.
FreeThe first week
1 weekThe test, a plan and a price. It comes off your build if you go ahead.
€700An agent
6 to 10 weeksOne job, live and passing the test.
from €12,500Paperwork
6 to 10 weeksPulling details out of documents.
from €10,500An answer assistant
5 to 7 weeksAnswers from your own documents.
from €8,000The plumbing
4 to 6 weeksPipelines, integrations, automations.
from €4,500Keeping it running
monthlyWe watch it and keep it accurate.
€1,000Hosting and model costs are billed at what they cost us. Nothing is locked in: you can stop after any stage.
Let's
measure it
It starts with a free 30-minute call. If there is a job worth doing, the first week is €700 and comes off the build.
- We pick the job that is wasting the most time
- We build the test with your own people
- You get a price, credited in full if you go ahead
Or email contact@iteamcollage.com
30 min · free