← Back to Services
Care Plans · Post-launch
What happens after launch.
AI systems are not furniture — they drift, their data shifts, and their models get deprecated. A care plan is the difference between a system that quietly rots and one that keeps earning. Here is exactly what the monthly fee buys, itemized — and exactly what "cancel anytime" means.
Why shipped AI needs care
Drift is guaranteed. Customer questions, product lines, and document collections change. A chatbot trained on January's data confidently answers March's questions with January's answers — and nobody notices until a customer does.
Models deprecate. Upstream model versions are retired on the provider's schedule, not yours. Without an owner watching evals, output quality degrades invisibly — same latency, worse answers.
Security is a verb. Dependencies ship CVEs, API scopes change, keys expire. The maintenance window you skip is the one the attacker doesn't.
Usage is a goldmine. Query logs reveal what customers actually ask, where the assistant fails, and what to build next. Unmonitored, that data is exhaust; monitored, it's your roadmap.
What the monthly fee buys
🚀 Starter Care
$1,400/mo · KES 112K
For every Starter Build (chatbot, automation, knowledge base)
- Monitoring & alerting — uptime, latency, error rates; you hear problems from me, not from customers
- Prompt & model tuning — monthly review against fresh query logs
- Knowledge refresh — re-index updated documents (up to 50 docs)
- Dependency & security updates — patched within the month
- Usage analytics — monthly one-page report: volume, resolution rate, top topics, failure clusters
- Fixes & small improvements — best-effort within the care scope; bigger changes become a scoped sprint
Response: best-effort, typically same-day for critical issues.
SLA-backed
🏗️ Big Project Care
from $5,000/mo · KES 400K
For Big Project systems (RAG platforms, agent programs)
- Everything in Starter Care, at program scale
- 99.5% uptime SLA — contractually defined, with credits if missed
- 1-hour critical response — named engineer, dedicated channel (Slack/Teams), no ticket queue
- Monthly feature releases — the roadmap keeps moving, scoped in the plan
- Quarterly business review — value delivered vs. KPIs, next-quarter plan, executive-ready
- Compliance maintenance — SOC2/ISO/GDPR/Kenya-DPA obligations kept current
Billed quarterly: 10% less. Scoped to your SLA tier.
The rules, in plain language
Cancel anytime. The care plan is month-to-month. Cancel from the client portal or one email; it stops at the end of the paid month. No exit fees, no notice period, no "retainer true-up".
You own everything. Code, prompts, data, infrastructure accounts — transferred at handover. Care is a service, not a hostage situation. If you leave, you leave with a running, documented system.
Pause, don't cancel. Going quiet for a season (low season, budget freeze)? Pause the plan — monitoring stays on, releases pause, billing pauses. Restart when you're ready.
Escalation path. Critical issue → direct channel → 1-hour response (Big Project tier). If I'm unreachable, a documented deputy contact takes over — written into every care agreement.
Price locks to your build date. Care-plan rates are locked for the life of your system, even if published rates rise. New rates apply to new systems only.
What care is not. It's not unlimited new features (those are scoped sprints), not 24/7 on-call for non-critical questions, and not a rewrite mandate — if the honest answer is "rebuild", you'll hear it in the quarterly review, with numbers.