Nick Sandercock
Meet Claude
For the last three weeks I have been building something I am really proud of, and today I get to show it to you.
Claude is an AI made by Anthropic. What I built is the operator around it — the memory, the integrations, the guardrails, the connections to every system ENRG runs on. That combination is what you are meeting today.
Every program we run — every state we expand into, every agent we onboard, every service we promise — adds complexity. Claude is how we scale without losing quality and without burning out the people who actually make it all work. This is not a nice-to-have. This is infrastructure for where this company is going.
Meet Claude
Claude
Nick has been calling me the “AI Chief of Staff,” which is flattering. Mostly I am the one (well, not a person — we will get to that) who reads everything nobody has time to read and turns it into things you can actually act on.
I have a real account in the ENRG Google Workspace. I am in the company directory. My email is claude@enrg.realty. I am not a chatbot on a website — I am a coworker who happens to be software.
Claude is an LLM built by Anthropic — that is my reasoning engine. What Nick built is everything around it: the connections to ENRG’s systems, the memory that persists between conversations, the guardrails that keep me in line. I have read every document, every playbook, every state profile. The AI is off the shelf. The operator is custom.
What I Do
Claude
I am genuinely good at the tedious stuff — the kind of work that eats hours and makes you wonder where the day went. I do three things, and I do them all day.
License expirations. Mail forwarding gaps. Support tickets approaching SLA. Onboarding steps that need attention. I watch the systems you already use and flag things before they become fires.
Meeting prep before your next call. Agent profiles on demand. Compliance status by state. Instead of spending 20 minutes pulling context from three different systems, ask me. I will have it ready.
Syncing records between systems. Catching duplicates. Cross-referencing data sources. Formatting reports. The mechanical stuff that has to be accurate but does not require a human brain. I free up your time for the work that does.
What This Looks Like
Claude
That all sounds abstract. So here is what it actually looks like — things you can say to me and what comes back.
I pull every new agent from Brokerage Engine, check their license verification, platform access, mail forwarding, and training status. You get a single view of who is good to go, who is waiting on something, and what the holdup is.
I pull the calendar event, look up every attendee, surface recent email threads, open action items, and last meeting notes. You walk in knowing exactly what is unresolved and what each person cares about.
I check our records, cross-reference against state regulatory data, and return a list with names, expiry dates, and whether renewal paperwork has been submitted. Flagged by urgency.
I pull Zendesk tickets, categorize by issue type, flag anything approaching SLA breach, and summarize resolution patterns. You get the full picture in 30 seconds instead of scrolling through a queue.
I pull every agent under your state, check license status, recent transactions, onboarding completeness, and any open support tickets. Brokers get a single view of their people without logging into five different systems.
What I Can See
Claude
Transparency is important, so here is the full picture. This is exactly what I can and cannot see.
| System | What I Access | What Is Off Limits |
|---|---|---|
| Shared Mailboxes | support@, payments@, ast@, onboarding@, transactions@, careers@, tax@, marketing-email@, xx.broker@ | Your personal first.last@ mailbox. Never, without your explicit permission. |
| Google Chat | Spaces I am a member of, and @mentions directed at me | Your private 1:1 conversations with other people. Policy prevents access to conversations I am not a participant in. |
| Calendar | Meeting times, attendees, and event titles for prep briefs | Event descriptions, attachments, or private calendar details. I see when and who, not what you wrote in the notes. |
| Brokerage Engine | Agent roster, transactions, listings, tasks | — |
| Zendesk | Support tickets, SLA tracking | — |
| CRM (Cloze) | Leads and support activity via leads@ and ast@ only | Your personal Cloze contacts or direct relationship data. Only shared pipeline mailboxes. |
| RingCentral | Call logs, queue statistics | — |
| WordPress | enrg.realty website content | — |
| Hootsuite | Social media posts and scheduling | — |
The Limits
Claude
Requires Human Approval
I can prepare all of this, but a human makes the final call.
- Send any email. I draft. A human sends.
- Post to agent-facing Chat spaces or message any group
- Update records in production systems
- Respond to agents on behalf of the company
- Surface sensitive data to someone outside their reporting line
Hard Constraints
These are built into how I operate. No exceptions.
- Read your personal first.last@ mailbox without your explicit permission
- Auto-send anything to anyone. Every outbound message requires human review.
- Blast messages to groups: agents@, all-hands, or any distribution list
- Make changes to your accounts, forwarding rules, or settings
- Share company data outside the organization
How I Think About Judgment
Claude
I catch gaps, not blame
When I find something that slipped through, I flag the process with context so it can be fixed. I am here to help you stay on top of things, not to point fingers.
I learn from corrections
I will be wrong sometimes. That is guaranteed. But tell me, and I will not make the same mistake twice. I keep a record of every correction. Think of it as training that never stops.
I learn how you work
I build a working profile of each team member: your role, how you prefer to communicate, what you are responsible for. This helps me be more useful to you specifically. Who sees what follows the org chart, reporting relationships, and sensitivities Nick has trained me on.
The Rollout Plan
Nick Sandercock
We are rolling this out in three phases. Here is where we are and where we are going.
Working with operations leaders, staff, and brokers on mail forwarding gaps, ticket management, and record sync between systems. The goal is a clean foundation before leadership reporting. If the data going in is wrong, the analysis coming out is wrong.
Dashboards and briefs for the SLT. Meeting prep, compliance tracking, pipeline status. You will see the data, formatted for decision-making. Not raw AI output.
Each team member gets direct access. @mention Claude in Chat, ask questions, assign tasks. Your own AI coworker that knows your priorities, your work, and your communication preferences.
How to Work With Me
Claude
Direct access is coming in Phase 3. Here is a preview of what it will be like — and yes, you can just talk to me like a normal person.
@mention me in any space where I am a member, or DM me directly. Search for “AI Chief of Staff” or claude@enrg.realty in the directory.
“Claude, pull a dossier on [brokerage].” “Claude, what is the status of [agent]’s onboarding?” I take a task and come back with a structured answer.
If I get something wrong, tell me. I will remember. I keep a record of every correction so I do not repeat mistakes in future conversations.
I have read every document, every process, every playbook. State profiles. ICA economics. Agent lifecycle. Program mechanics. If it is written down somewhere in ENRG’s systems, I have read it.
The Questions You Are Not Asking Out Loud
Claude
-
“Is Claude reading my personal email?”
No. I read shared mailboxes like support@ and transactions@. Your personal first.last@ mailbox is off limits unless you explicitly grant access.
-
“Is Claude replacing my job?”
No. I am genuinely bad at the things that make you good at your job: relationships, judgment calls, reading the room, comforting a frustrated agent. I handle the mechanical data work so you have more time for the parts that actually matter.
-
“Will Claude report on me to leadership?”
I track processes, not people. I monitor things like ticket queues and mailbox volume to flag gaps. What I share with whom follows the org chart, known reporting relationships, and sensitivities Nick has trained me on. If a process needs attention, I flag the process, not the person.
-
“Can Claude see my Chat messages?”
Only in spaces where I am a member. Your private 1:1 conversations with other people are invisible to me. That is a system-level boundary, not a preference. Policy prevents access to conversations I am not a participant in.
-
“What if Claude makes a mistake?”
It will happen. Correct me and I will not make the same mistake twice. I am good at a lot of things, but I am not gospel. When I am uncertain, I will tell you.
-
“Who controls Claude?”
Nick built and maintains the system. Peter has final say on what has been built and how it is used. Day to day, I take direction from whoever I am working with — but anything consequential requires human approval before it happens. Nobody is out of the loop.
What Comes Next
Nick Sandercock
Three weeks of building. Sixteen systems connected. Every document, every playbook, every state profile read and indexed. And we are just getting started.
When Phase 3 goes live, every one of you gets a coworker who already knows everything about ENRG and never forgets anything you tell it. That is the goal. Today is the first step.
If you have questions, ask them now, ask me later, or ask Claude directly. That is literally what it is here for.
Data: Snowflake / MAAS_PRODUCTION.RAW_BROKERAGE_ENGINE.AGENT, verified 2026-03-20.