Contextual answers(On the roadmap)
Ask about the job, customer or invoice in front of you, and get an answer from ShieldBook's own records, within your permissions.
AI in ShieldBook
AI that works inside your business, not beside it.
Most AI in business software is a chat window that can't see the work. ShieldBook's direction is assistance and agents that understand the real record, act through the same permissions and rules as your people, and leave a name on everything they do.
ShieldBook's AI is direction, not a released feature. The foundation it stands on, permissions enforced on the server and a recorded history of every action, is built today.
01Why context matters
Not a chatbot attached to a CRM.
AI is only as useful as what it can see and what it's allowed to do. ShieldBook already holds the whole job on one record, with rules on every action. That's the difference between an assistant that guesses and one that knows.
AspectA chat window beside the businessAI inside ShieldBook
What it knowsWhatever you paste into the chat, or a copy of your data that's already out of date.The live record: the customer, the property, the equipment's history, the open estimate, the balance.
What it can doWrite text you then re-type into the system that actually runs the business.Use the same actions your people use, so a draft is a real draft on the real job.
Who it answers toWhoever has the login. Its actions look like a person's, or aren't recorded at all.Its own identity and permissions, a person approving what matters, and a name on every action.
02What it will help with
The work around the work, handled with you.
Everything in this chapter is planned direction. Each one builds on records and actions ShieldBook already has.
Admin, drafted(On the roadmap)
Job summaries from a technician's notes, follow-ups after an estimate, invoice drafts from sold work. Prepared for a person to check.
Customer communication(On the roadmap)
Replies and follow-ups that know the history, drafted in your voice, sent only when someone approves.
What needs attention(On the roadmap)
Approved work not sold, visits to book, balances to chase, members coming due: surfaced before anyone asks.
Automation on events(On the roadmap)
Confirmations, reminders and outreach that start from what happens on the record, not from a timer polling for changes.
Agents with a job(On the roadmap)
An office assistant, a dispatch agent, a lead agent: each with a narrow role, its own permissions and a person it answers to.
In development On the roadmapEverything else is built and working today.
03How an agent will act
Agents recommend. ShieldBook's rules govern.
An agent in ShieldBook is never a borrowed staff login and never a shortcut around the rules. It proposes through the same actions a person would use, the server checks its permissions, and anything consequential waits for a person.
Its own identity
An agent is its own actor with its own permissions. When it acts for a person, the history records both.
The same rules
Agents use the same actions, permissions and business rules as people. There are no separate AI versions of anything.
Every action recorded
What an agent did, when, and who it acted for, in the same activity history as everyone else.
Inside your business
An agent works inside one company's records and nowhere else.
Your choice of AI
One controlled interface for any capable AI. ShieldBook isn't tied to a single AI provider.
04Human oversight
People approve what matters.
Some actions change what a customer is promised, charged or told. When agents arrive, those will wait for a person, enforced by the server: an instruction in a prompt is guidance, never permission.
An approval boundary can be narrowed for a specific, proven workflow. It is never removed by default.
- Sending a message to a customerNeeds a person
- Refunds and voidsNeeds a person
- Prices and discountsNeeds a person
- Permissions and rolesNeeds a person
- Scheduling commitments to a customerNeeds a person
05Current and future
Each step earned by the one before it.
ShieldBook isn't building speculative AI ahead of the work. The operating system comes first, proven on real days in the field. AI arrives in steps, starting with reading and ending, for narrow jobs, with agents that run on their own within strict limits.
Build the operating system
Customers, jobs, estimates, invoices and payments on one record, with rules on the server.
In place
Prove it on real work
Run a working service business on it, every day, before anything is automated.
Happening now
Reliable actions and structured records
Every business action built once, permissioned and recorded whoever invokes it, and a structured history that grows with every job.
Happening now
AI that reads(On the roadmap)
Answers from ShieldBook's own records, within the asker's permissions.
Ahead
AI that acts, with approval(On the roadmap)
Drafts and proposals through the same actions people use, with a person approving what matters.
Ahead
Agents that respond to events(On the roadmap)
A call-out, an approved estimate, a visit coming due: agents react, inside their permissions.
Ahead
Constrained autonomous work(On the roadmap)
Narrow processes that run on their own, with budgets, limits and stop conditions.
Ahead
In place today
The foundation AI will stand on
The rules an agent will follow are the rules ShieldBook already enforces for people. That's why AI can come later without a redesign.
- Five roles and server-enforced permissions
- An append-only history of every action, with a name on it
- Business actions built once, so every invoker follows the same rules
- Structured records: Concern, Cause, Correction, Recommendation on every job
Keep exploring
- Platform How ShieldBook carries one record through the whole service lifecycle: inquiry, customer, property, job, estimate, approval, scheduling, field work, invoice, payment, follow-up and repeat business.
- Marketing and growth Where ShieldBook is going beyond operations: lead capture, nurture, campaigns, content, messaging and closed-revenue attribution on the same record.
- What's built and what's next Every ShieldBook capability with its status: built and working today, in development, or on the roadmap.
Talk with us about where AI fits.
We'll show you the foundation in ShieldBook today and walk through where assistance and agents are going.
ShieldBook is being developed and proven inside a working service business as it prepares for broader use. A demo is a walkthrough and an honest conversation about fit.