Inside the system, not on top of it
Many AI products for utilities sit on top of an existing CIS. They are open about it: the CIS stays the system of record. That is the right instinct, and it is also the limit. A layer outside the system has to copy your data, keep a second permission model in step with the first, and work around business rules it cannot see. When it is time to act, it hands the work back to the CIS and to your staff.
AI in SpryCIS is part of the platform. It reads and acts through the same APIs, permissions and workflows as your staff, so there is no second copy of your data to reconcile and no second set of rules to keep in sync.
Agentic operations: the routine work, done
Answering questions is useful. Taking routine work off your team’s plate is better. SpryCIS agentic operations handle tasks that utilities do by hand today, end to end, inside the platform.
Reviewing service order activity from the field is a good example. Every completed service order can carry information that needs a next step, and someone has to read through them to find it.
Field work comes back
Crews complete service orders, and their results and notes land in SpryCIS as they always have.
The agent reviews the activity
It reads each completed order in the context of the account, the premise and the work requested.
Follow-ups are identified
Where the results call for a next step, the agent identifies it instead of waiting for someone to spot it.
Action through governed workflows
Follow-up actions use the standard SpryCIS workflows, under defined permissions, with every step recorded.
More agentic operations are on the way, each aimed at a manual task that slows utilities down today.
The Customer Service AI Assistant
For customer service representatives, the assistant sits on the account page in SpryCIS, where the work already happens.
On the account page
Ask about the account in front of you without leaving it or rebuilding context in another tool.
Your processes as quick actions
Define your own service flows, business rules and standard procedures as guided starting points, so every representative handles common requests the same way.
Reads before it writes
The assistant gathers the relevant account information before it recommends or prepares anything.
Acts through governed workflows
When a task needs a change, it uses the approved SpryCIS workflow and asks for confirmation. A move-in goes through the Move-In wizard, not around it.
Understands utility data
Built-in skills help it interpret what the data means, such as how budget billing activity relates to an account balance.
Every conversation on the record
Chats are tied to the account, so context, tool activity and actions stay traceable.
Access by role
Administrators control which users and roles can use the assistant.
Usage and impact you can see
A dashboard shows adoption, usage and operational impact.
Governed by design
Least privilege
AI actions run under SpryCIS permissions. The assistant works with the representative’s own role, and is never a way around role-based access.
Read before write
Changes go through approved workflows, with clear confirmation before data changes.
Auditable
Account context, tool activity, conversation history and material actions are traceable.
Deterministic core
AI works with governed SpryCIS data and business rules. It does not replace them.
People stay accountable
Staff review consequential outputs and actions.
Your data stays yours
SpryCIS uses Anthropic’s Claude models through Amazon Bedrock in SpryPoint’s AWS environment. Client data and conversations are not used to train the models.
Opt-in
You choose where AI is enabled and tune it to your regulatory, operational and customer service expectations.
Measured by outcomes
We judge AI by service and operational results, not by how much it gets used.
Questions to ask any AI vendor
Whoever you evaluate, these questions separate AI that runs inside your operations from AI that sits beside them.
- Whose permissions does the AI act under, and can it do anything the user cannot?
- Can it complete a transaction, or does it send the user back to the CIS to finish?
- Do its actions go through the same business rules and audit trail as a person’s?
- Does it need its own copy of your customer data, and who keeps that copy current?
- Where does your data go, and is it used to train anyone’s model?
- When it gets something wrong, can you see what it read and what it did?
What it means for your utility
Faster service
Fewer navigation steps to find information and complete routine tasks.
Less training
Guided quick actions reduce onboarding and re-training as processes change.
Consistent handling
Common requests follow your standard procedures, every time.
Fewer escalations
Routine questions get answered without waiting on a specialist.
Work that doesn’t wait
Agentic operations pick up routine follow-ups as soon as the data arrives.
Capacity to grow
Handle more accounts and interactions without equivalent growth in manual effort.
