Platform · AI
AI with a job description.
AI earns its place when it has a clear job. In Plant an App, it takes on steps in your processes, answers and acts in conversations with residents and staff, and helps your team build, working with the data and the workflows you give it, and leaving the decisions that matter to people.
Stall
AI without a job stays a pilot.
The demo impresses. Then it meets the real work, and the pilot never makes it to production.
- 1
It sits beside the system.
A chat window in one tab, the case in another, and staff copy answers from one to the other.
- 2
It can talk, but not act.
It explains how to book an inspection, then sends the resident to a form to do it themselves.
- 3
No one knows what it sees.
It’s unclear which data reaches the model, and who answers for what it says.
- 4
It never reaches production.
Without answers, the pilot stays a pilot, and the work stays the same.
AI doesn’t need more freedom.It needs a job.
Assign
Three jobs, one platform.
AI isn’t a separate system beside the application. It does its best work where the work already happens: in the process, in the conversation, and in the builder. It uses the same data, workflows, and users as the rest of it.
- In the process
A step in the work.
AI reads, checks, and sorts inside a workflow, between your rules and the people who decide.
For examplePre-checking the documents attached to a permit application. - In the conversation
An assistant people talk to.
It answers from your own content and acts through your workflows, for residents and for staff.
For exampleA resident books a final inspection in a chat. - In the builder
A partner for your builders.
Builders describe what they need in plain words, and AI drafts it in the builder.
For exampleA SQL query from a plain-language question.
Automate
AI takes on a step. People keep the decision.
Much of the work in a process is reading: a stack of attachments, a contract, a free-text description. AI does that step inside the workflow and hands the result to the rules and the people that come next.
AI stepPre-check attached documents
Building permit application · Residential addition
- Site planAttached
- Proof of ownershipMissing
- Contractor licenseExpired
- Structural calculationsAttached
Back to the applicant with what’s missing, before a reviewer opens the file
Check what’s attached.
Before a reviewer opens the file, AI reads each document against what the application requires and flags what’s missing or out of date.
Turn free text into fields.
Pull the jurisdiction, the expiration date, and the key clauses out of a contract, as fields your rules, reports, and reviewers can use.
Summarize and transcribe.
A long case file summarized before review, or a recorded hearing turned into text.
AI returns fields, not loose text.So the next step is a rule, a review, or an approval.
Converse
Assistants that answer, and act.
An assistant is only as useful as what it knows and what it can do. In Plant an App, it answers from your own content, knows who it’s talking to, and acts through your workflows, so a question can end with the work done.
Can I book the final inspection for my permit?
Runs workflowCheck application status
Your kitchen addition passed its last review. The earliest final inspection is Tuesday at 10:00. Shall I book it?
Yes, please.
Runs workflowBook an inspection (Tuesday, 10:00)
Booked. The confirmation is on its way to your email.
Knows your content.
It answers from your websites, PDFs, spreadsheets, and records, and keeps up as they change.
Knows who’s asking.
A signed-in resident or a user from another system, with what the assistant needs to know about them, so the answer fits the person.
Acts, or hands over.
It checks a status, books an inspection, or opens a request through the workflows you’ve given it, and hands over to staff when a person is needed.
Where it works
- Your website
- A page of its own
- Microsoft Teams
- An API
There’s no second automation to govern: the assistant works through the same workflows as the rest of the application.
Build
Your builders describe it. AI drafts it.
The same AI works alongside the people who build your applications. They describe what they need in plain words, and AI drafts it inside the builder, for them to review before anything runs.
SQL from a question.
Describe the data you need, and the assistant writes the query from your actual tables and columns.
A first draft of a form.
Describe the form in plain language, and AI proposes the fields to start from.
Answers while you build.
An assistant that knows the platform answers questions inside the builder, from its documentation and tutorials.
What AI drafts, your team owns.It’s in the designer, versioned, and changed like anything else.
Choose
Your models. Your keys. Your infrastructure.
The AI behind every job is yours to choose, from the provider your organization already approved, on your own account.
- OpenAI
- Azure OpenAI
- Amazon Bedrock
- Google Vertex AI
Your own accounts.
Connect the provider you already trust with your own keys, kept encrypted and never shown again.
A better model, without a rebuild.
Put the AI step behind a Workflow Contract, and moving to another model is a change in one place.
Data stays where you run it.
Run Plant an App on your own infrastructure, and your data leaves it only to reach the model you’ve chosen.
The workflow outlivesthe model behind it.
Supervise
Checked before it starts. Reviewed as it works.
Like anyone new on the job, an assistant is tested before it starts and reviewed while it works, so you can see what it was given, what it did, and what it answered.
Tested before it starts.
Write the questions and the answers you expect, and test the assistant before residents or staff talk to it.
Kept to its job.
A filter keeps each assistant on topic and turns away questions that don’t belong to its job.
Reviewed as it works.
Every conversation is saved and rated, and every AI call records who made it, with which model, and how many tokens.
AI does the job.People make the call.
Related capabilities
Bring the work no one has time for.We’ll give AI a job in it.
Talk to us about the documents your staff read, the questions your team answers every day, and the queries your builders write, and see where AI takes on the work.