CYNDRA
How Behring Co put AI to work across its 28-person team with Cyndra
The real estate firm now runs AI workflows in finance and investor relations, built live with the people who use them.
"My team is getting addicted to it."
Colin Behring, Founder and CEO, Behring Co
Your embedded AI department: strategy, governance and workflows built live with your team, so they own it.
At a glance
Client:
Behring Co, a privately owned real estate development and private equity platform in the greater San Francisco region
Team:
28-person team
Engagement:
embedded AI department, ongoing since early 2026
Areas so far:
finance, investor relations, a team-wide AI assistant. Next: sales
Tools the team already used:
QuickBooks, SharePoint, Microsoft 365, Juniper Square
How:
built live on real work, skills the team keeps, human review on high-stakes steps
The challenge
Colin Behring was already sold on AI. Time was the problem.
"Right now I'm the roadblock because I'm getting dragged in other directions."
Colin Behring, Founder and CEO, Behring Co
Do-it-yourself projects kept stalling.
"So you've got like 99% of something and 100% of nothing."
Colin Behring
The same pattern showed up inside each department. In finance, intercompany invoice support was a monthly chore: filter distribution-account transactions in the QuickBooks general ledger, collect about 20 to 30 invoices per entity, file the support in SharePoint and link it for parent-company reimbursement. Across five entities, that was about 10 to 20 hours of manual work a month.
In investor relations, the work was tracking and correspondence across several systems.
"Manually updating like, you know, three or four trackers."
Behring Co investor relations team
Pulling an investor's fund documents to answer an email was, in their words, "our most time consuming task at present."
What we built
A finance skill for intercompany invoices.
QuickBooks general ledger → invoices → SharePoint → parent-company reimbursement
In a live session with their finance lead, we built and saved an /intercompany skill that runs the monthly process and handles missing vendors and invoices with zero or multiple attachments. The finance lead keeps the skill and runs it.
An investor-response skill.
Fund documents and past correspondence → draft replies → senior reviewer
The team's AI assistant could not reliably carry prior investor-email context into a new request, so people repeated themselves. A skills session covered connected data, step-by-step instructions, edge cases, saving and feedback. The investor relations team then built an investor-response skill (version 1) that connects Microsoft 365 and read-only Juniper Square and drafts replies from the fund documents and past correspondence. It never sends. Every draft goes to a senior reviewer before an investor sees it, and the skill is still being refined.
A team-wide AI assistant in Microsoft Teams.
Company files → find a document or pull together a draft
Installed company-wide in May 2026, with access to company files, so anyone can ask it to find a document or pull together a draft.
How we worked
Colin set the model on the call that scoped the engagement.
"Having a fractional AI department is definitely the way to go because it is a distraction for founders."
Colin Behring, Founder and CEO, Behring Co
So we worked inside his team, not around it.
Live builds on real work.
Each session takes one person's actual task, on their actual data, and ends with a working tool.
Skills the team keeps.
Every workflow is saved as a named skill the team can rerun and edit without Cyndra.
Human review on high-stakes steps.
Investor replies are drafted, not sent. Invoice exceptions are flagged for a person.
A weekly call with Colin,
plus live requests from staff in Teams.
Results so far
Finance:
their finance lead estimated the intercompany skill would save at least 8 hours a month against the 10 to 20 hour baseline. A measured figure follows after a full month of use.
Investor relations:
version 1 is in use. The team says it is too early to put a number on time saved, so we have not.
Search and analysis:
the team-wide assistant is Colin's daily tool for finding things and running quick analyses.
"I use him a tremendous amount just to find stuff."
Colin Behring, on the team's AI assistant
"And like I couldn't pay an analyst enough money to do it in 35 seconds."
Colin Behring, on a rent-comparison analysis the assistant produced for him
Estimates are attributed to the people who made them. Cyndra publishes measured numbers only after the client confirms them in writing.
What's next
The next focus is sales automation for the Stak team.
Book an intro call.
Thirty minutes on where an embedded AI department would start in your team.