Salesforce
Xero
Outlook
Gmail
Microsoft Teams
Excel
Google Sheets
SharePoint
Google Drive
Jira
Shopify
Leads the company and its customer relationships.
AI-native consulting firm
We deploy our AI agent inside the software your company already uses. It finds the work your team does over and over, and automates it.
Your company’s software, as our agent maps it. Example workspace, fictional data. Select a finding to see its automation.
- Orders
- Payments
- Repetitive work
Inside your software
Your software already knows how your company works.
Every order, invoice and handoff leaves a trail in your tools. Our agent starts there.
Examples of the software companies run, not a list of integrations: Salesforce, HubSpot, NetSuite, SAP, Microsoft Dynamics 365, QuickBooks, Xero, Sage, Outlook, Gmail, Slack, Microsoft Teams, Excel, Google Sheets, SharePoint, Google Drive, Zendesk, ServiceNow, Jira, Asana, Shopify, Stripe, DocuSign, Workday, and in-house tools.
Examples, not a list of integrations. Logos belong to their owners; no affiliation is implied.
How it works
From the software you have
to automations you trust.
Four steps, one workflow
Only what you authorize.
- Salesforce · accounts and opportunities
- NetSuite · customers and sales orders
- Outlook · the orders@ inbox
Example data. Access can be narrowed or revoked at any time.
Order intake, as it runs today.
- An order arrives in the orders@ inbox
- Its lines are copied into a spreadsheet
- Someone keys the order into NetSuite
Mapped from activity in the connected tools. Example data.
Where the hours go.
- Orders retyped into NetSuite · about 40 a week
- Overdue invoices chased by hand · every Monday
- New customers set up in four tools · by hand
Estimated from activity patterns. Example data.
Draft sales orders from PO emails.
- Reviewed and switched on by your team
- 38 orders drafted this week
- 2 price differences sent to a person
Example data. Every run is logged; pause or edit it at any time.
You choose the access
Automation examples
See what the agent finds,
and what it builds.
Three common examples, with fictional data.
The same patterns show up in sales, finance, operations and support.
Order intake Example · 01/03
Repeated work
Order emails are retyped into NetSuite by hand.
Where it happens
- Tools
- Outlook → Excel → NetSuite
- How often
- About 40 orders a week
- Time spent
- About 4 hours a week
- People involved
- 2
Also noticed
Price differences are usually caught after the order is entered.
Draft sales orders from purchase-order emails.
The agent reads each PO, matches the customer and items, and prepares the order. A person approves anything that doesn’t match.
Steps
When a PO arrives in the orders@ inbox in Outlook
Match the customer and items in NetSuite
Create a draft sales order
Send price differences to a person to approve
Post a summary to #orders in Slack
Example with fictional data.
Shows how the agent is designed to work, not a customer result.
Invoice follow-up Example · 02/03
Repeated work
Overdue invoices are checked and chased by hand every week.
Where it happens
- Tools
- Stripe → Excel → Outlook
- How often
- Every Monday, about 30 invoices
- Time spent
- About 3 hours a week
- People involved
- 1
Also noticed
Accounts with a recent payment or an open dispute sometimes get a reminder anyway.
Prepare reminders, checked against payments and disputes.
The agent drafts each reminder from the invoice and account history, and holds back any account with a recent payment or an open issue.
Steps
Every Monday, list overdue invoices in Stripe
Check for recent payments and open disputes
Draft a reminder in Outlook for each clear account
Hold flagged accounts for a person
Log each reminder on the customer record
Example with fictional data.
Shows how the agent is designed to work, not a customer result.
New customer setup Example · 03/03
Repeated work
Every won deal is set up again, by hand, in three other tools.
Where it happens
- Tools
- Salesforce → NetSuite, Asana, SharePoint
- How often
- About 12 new customers a month
- Time spent
- About 45 minutes each
- People involved
- 3
Also noticed
Details are retyped at each step, so small errors carry through.
Set up new customers from the won deal.
When a deal closes, the agent creates the customer, the project and the shared folder from the same details, then hands off to the team.
Steps
When an opportunity is Closed Won in Salesforce
Create the customer in NetSuite
Start the onboarding project in Asana
Create the shared folder in SharePoint
Tell the account team in Slack
Example with fictional data.
Shows how the agent is designed to work, not a customer result.
Three examples.
Your company has its own.
The firm
We deploy the agent. You decide what runs.
Kestrel is an AI-native consulting firm. We set up the agent in your software and work with your team on each automation, so it fits how your company runs.
The team
The people
you’ll work with.
Two founders, based in St. Louis.
Andre Romero-Carey
Co-founder, CTO/CFO
Leads engineering and finance.
Andre · Wealth management internship
Bom · Internship
Bom · Monitor Deloitte internship
Both founders · Education
Our founders’ internships and education. No affiliation or endorsement is implied.
Questions
A few things
you might be wondering.
Kestrel is an AI-native consulting firm based in St. Louis, Missouri. We deploy an AI agent inside the software a company already uses. It finds the repetitive work between tools and drafts automations your team reviews, switches on and oversees.
Not yet. It’s in development, and we’re taking early-access requests. The founders read every one and reply by email.
It’s designed for repetitive work that moves between tools: retyping emailed orders into an ERP, chasing overdue invoices, setting up a new customer in several systems, or copying data between spreadsheets and business software.
Automation tools wait for you to decide what to automate, then have you build it. Our agent finds the repetitive work and drafts the automation, and we deploy it with your team.
It’s being built for the software companies run on: CRM, ERP, accounting, email, chat, documents and spreadsheets, starting with what early-access teams use most. The names on this page are examples, not finished integrations.
Only the tools and data you authorize. You can narrow or revoke its access at any time.
Not without your approval. The agent drafts automations; your team decides which to switch on, sees every run, and can pause any of them.
Early access
Start with
one workflow.
Tell us which software your team uses and where the repetitive work is. Our agent is in development, and the founders read every request.