05/05·Claude operator retainer

Claude operator retainer.

A productive agent is running. But Claude keeps evolving, your data changes, new edge cases show up. Per month: one eval run against your gold set, one prompt update, one model check, one incident slot when something breaks. So nobody has to be hired internally for this.

Claude stays in production, every month.

RET-OPS-M

·

EUR 2,400 net / month (1.5 to 3k)

·

monthly, 8h reserved

·

remote, asynchronous, 30-minute sync per month

0X/05·Methodology anchor

Where in the Six-Step-Plan map.

[00]

method.step0Title

method.step0Sub

[01]

method.step1Title

method.step1Sub

[02]

method.step2Title

method.step2Sub

[03]

method.step3Title

method.step3Sub

[04]

method.step4Title

method.step4Sub

[05] ●

method.step5Title

method.step5Sub

Step 5 ongoing: trust ladder maintenance every month.

Volle Methode ansehen →

02/05·Detail

// promise

Per month one defined delivery bundle: one eval run, one prompt-pack update, one model-migration check, one incident response slot.

// process

Day 1 to 5

Eval run, drift detection.

Day 6 to 15

Prompt or skill update if needed.

Day 16 to 25

Model migration check, recommendation.

Day 26 to 30

Report, sync session. Incident slot available at any time.

// deliverables

+One eval run against the gold-set (drift report).
+One prompt-pack or skill update (version bump documented).
+One model migration check (Sonnet/Opus versions relevant?).
+One incident slot up to 4 hours (response within 1 working day).
+Monthly report (Janis brand, REP format), 2 to 4 pages.

exit criterion

No exit; cancellable monthly after a minimum term of 3 months. Success indicator: eval score remains stable or improves.

// tech footprint

Claude (Sonnet 4.6 plus), eval framework, skill repository, MCP server, n8n optional. Existing client stack.

03/05·Reference profiles

Wer das typischerweise kauft.

Existing pilot client after rollout

branche: Any sector, prerequisite is a productive Claude workflow.
größe: 50 to 300 FTE.
anlass: Pilot is running but model updates, eval drift, prompt iteration and skill maintenance overwhelm internal IT.
outcome (geschätzt): Eval score stable across model migrations, no throughput drop on Anthropic releases. Response time on production issues under 24 hours.

Existing client without pilot, with own skills

branche: Mid-market manufacturing or consultancy that has already introduced Claude internally.
größe: 100 to 500 FTE.
anlass: Own skills/MCPs built, maintenance silently consumes internal IT time.
outcome (geschätzt): Internal IT recovers an estimated 4 to 8 hours per month; skill drift is detected before it reaches users.

04/05·Testimonial

template, zur Freigabe durch Kunde

[TEMPLATE]

„The retainer is our insurance against model updates and prompt drift. We know that every month an eval run takes place and Janis Tirtey intervenes before our users notice any degradation."
CTO, B2B software vendor, approx. 200 employees.

05/05·What you do not get

Does not fit.

×Client wants to use the retainer as a cheap hours bucket for consulting.
×No productive workflow exists (pilot first).
×More than three workflows to maintain (second retainer or dedicated model).