Asset and portfolio management
Every asset watched. Not sampled.
PSV builds the layer that reads your budgets, rent rolls, loan documents, capex plans, and prior reporting across the whole portfolio, on a cadence you set. So “which assets are drifting?” stops being a week of work and becomes something you already know.
Nothing is sent, filed, or approved on your behalf. Every figure carries the document it came from, and a person signs.
Portfolio, continuously
Illustrative
One tile, one asset. Most of them are fine, which is exactly why nobody has time to check them. The few that move are the reason you check them all.
Pacific Software Ventures builds AI for commercial real estate asset managers and portfolio managers at owner-operators, REITs, funds, and family offices. PSV maps how the firm reports, builds the Company Brain that holds budgets, rent rolls, loan documents, capex plans, and prior reporting as context AI can read, and deploys agents that review the whole portfolio on a set cadence rather than sampling a few assets before a board meeting. The work covers budget-to-actual variance, lender covenant and maturity exposure, capex plan tracking, lease events, and investor and board reporting. PSV offers three engagements: the AI Audit maps how the firm works, Company Brain and AI Deployment builds the context layer and the agents on top of it, and AI Training puts the capability in the team through PSV-led enterprise training and the CRE AI Institute. Deployments run in the firm’s own Claude or Microsoft Copilot workspace or in its own cloud environment, agents are scoped to the systems and folders the firm approves, every figure traces to its source document, and a person approves anything that reaches an investor, a lender, or a board.
You have already tried this
Four moves, all of them reasonable.
None of these were mistakes. Each one solved a real problem. It is worth being precise about which problem, because the one that keeps hurting is not on the list.
- 01
You standardized the reporting template.
What it changed
Every asset now reports in the same format, and the board sees one document.
What it did not
The template is still filled in by hand, one asset at a time, by the person who had the bandwidth that month.
- 02
You bought the portfolio dashboard.
What it changed
The numbers are in one place and the charts refresh without anyone rebuilding them.
What it did not
It reports what happened. It does not read a loan document, and it will not tell you which assets deserve your attention this month.
- 03
You gave the team ChatGPT.
What it changed
Writing got faster, and a few people got genuinely good at it.
What it did not
It writes well about the one asset someone pasted into it. It has never seen the rest of the portfolio, or last year's variance, or why you approved that capex.
- 04
You hired another analyst.
What it changed
Coverage went up by one person's capacity, which is real.
What it did not
The cadence did not move. You still sample before the board meeting, and the institutional memory still walks out when they do.
The shared failure
Every one of them improved the artifact. None of them changed the coverage.
A template, a dashboard, a paragraph, a pair of hands. Each is a better way to produce the thing you were already producing. Not one of them is something that reads the portfolio, so a person still picks which assets get looked at, and the assets nobody picked are the ones that surprise you.
The same portfolio, twice.
Illustrative
Assets a quarter actually gets reviewed
Assets that actually moved
One of them is in both rows. That one gets caught, and it is the one you were already worried about. Everything else in the second row is found later, in a variance you have to explain rather than a variance you called.
Your reporting is not late. It is sampled.
A quarter is not a monitoring interval. It is how long an asset is allowed to drift before anybody is required to look at it.
What we build
One layer underneath the portfolio, reading all of it.
PSV builds the Company Brain for the hold period: your budgets, rent rolls, loan documents, capex approvals, and prior reporting held as context that AI can actually read, with every figure traced to the document it came from. Agents run the whole portfolio against it on the cadence you set. Most assets stay quiet. The ones that move raise their hand.
This morning’s pass
Illustrative
Every tile was read on the same pass. Five had something to say. The other sixty-seven are the point: nobody spent a morning confirming they were fine, and nobody had to guess which ones to check.
What came back
Reforecast
Operating reforecast has moved outside the approved range two months running, and the explanation on file is still last quarter's.
Leasing
Concessions are rising while renewals hold. Two signals pointing opposite ways on the same asset.
Debt
A covenant test date lands before the next reporting cycle, so the first person to see it should not be the lender.
Capex
The project is tracking behind the schedule it was approved on, and the approval memo says why that matters.
Rollover
Two significant expirations sit inside the hold horizon with no renewal path recorded anywhere.
Each one opens to the document it came from. Nothing arrives as an assertion you have to take on faith, and nothing leaves the building without a person.
A person can analyze one asset deeply. A system can watch all of them. You still decide what it means.
Additive, not a replacement
Nothing gets ripped out, and nothing gets migrated.
This is not another platform for the portfolio to live in. We build on the AI accounts your firm already pays for, and the Brain makes the work already running there aware of your assets.
- ClaudeWhere the long reads happen: a loan document, a reforecast, a board memo that has to hold up.
- ChatGPTWhere the team already drafts. The good prompts stop living in one person's head.
- Microsoft CopilotAlready deployed across M365, already through IT, already inside the tenant your auditors signed off on.
What stays exactly where it is.
- 01
Your accounting and property systems stay the system of record. Nothing is migrated out of them.
- 02
Your reporting templates stay yours. The output arrives in the format your investors already read.
- 03
Your approval chain stays intact. The system drafts and flags; the sign-off does not move.
- 04
Your spreadsheets stay. Models people trust are context to read, not legacy to replace.
If your firm has standardized on one AI surface, we build there. If it runs more than one, the Brain serves all of them, so the portfolio context does not fork with the tooling.
How the work is bought
Three engagements, all downstream of the Brain.
Most asset-management teams start with the audit, because the honest answer to what should be automated is usually not the thing that hurt most recently.
- 01
AI Audit
Explore AI AuditMap how the portfolio actually gets reported.
We sit with the people who close the month and build the board package, and map the real path from property data to an approved narrative, including the parts that live in somebody's head. What we find becomes the Brain's first layer.
- 02
Company Brain + AI DeploymentFlagship
Explore the Company BrainGive AI the portfolio, then put it to work on all of it.
Budgets, rent rolls, loan documents, capex approvals, and prior reporting become context the AI can read, with every figure traced to its source. Then the agents that run the whole portfolio against it on your cadence.
- 03
AI Training
Explore AI TrainingPut the capability in the team that owns the assets.
PSV-led enterprise training and the CRE AI Institute, run on your own portfolio rather than a generic demo, so your asset managers practice on the reporting they are actually accountable for.
Not sure which one your firm needs? That is the question the assessment answers, and the answer is sometimes none of them yet.
Get your free AI assessmentWhat running looks like
Not a faster quarter. A portfolio that is never unread.
Once it is deployed, these are not things the system does when asked. They are things that are true between the asking.
- 01
Coverage
Every asset is read on the same pass.
Not the ones somebody had bandwidth for, and not the ones that were interesting last quarter. The asset that has been quiet for six periods gets the same read as the one already on the watch list, which is the only way the quiet one can ever surprise you less.
- 02
Variance
Movement arrives explained, and traced.
Budget-to-actual drift comes back with the line it came from, the prior period it is being measured against, and a drafted question for the property team where the explanation is missing. The review meeting starts at the disagreement instead of the arithmetic.
- 03
Exposure
Obligations are re-checked, not remembered.
Covenant tests, maturities, reserve conditions, and approved capex commitments live in one place and get looked at again on schedule. The failure mode this removes is the ordinary one: everybody knew about it, and it was in a document nobody reopened.
- 04
Reporting
The quarter is assembled, not built.
The investor and board package draws on a state that already exists, with the narrative drafted from the same sources the numbers came from. The reporting window turns from a construction project into a review, which is what your team is actually paid for.
And what it does not do
It does not send, file, approve, or commit. Every output is a draft with its sources attached, and the name that goes on it is still yours.
Proof
We will not quote you a number for work we have not done for you.
Plenty of vendors will tell you how many hours this saves. We publish a figure only where it traces to something we handed a specific client, and we will not invent one to fill a section. Here is the class of outcome instead, and then the receipts.
Coverage stops depending on capacity.
More of the portfolio gets read, more often, without the reading being somebody's whole week. The constraint moves off how many assets a person can hold in their head.
Reporting stops being a deadline sport.
The package draws on a state that is already current, so the scramble comes off the calendar and the time goes into the part that needs judgment.
The firm keeps what it learns.
Why a variance was accepted, why a capex was approved, why an asset went on the watch list. Held by the firm rather than by whoever happened to be in the room.
Read the case studies, numbers attached
Prefer to hear it from a client? Ask for a reference.
Environment and trust
Four claims, all of them checkable.
You are going to forward this page to somebody in IT. These are the four sentences they will want, written so they can confirm or challenge each one.
- 01
It runs where your data already lives.
Your firm's own Claude or Microsoft Copilot workspace, or your own cloud environment. We do not stand up a new home for portfolio data and we do not ask you to move it.
- 02
It sees only what you approve.
Agents are scoped to named systems, sites, and folders. Access is granted deliberately, listed plainly, and reviewable by the people who own the data.
- 03
Every figure carries its source.
Outputs cite the document they came from. Where a number cannot be traced, it is flagged rather than smoothed over, because a confident wrong number is the expensive failure in this job.
- 04
People decide what becomes firm knowledge.
Nothing is promoted into the Brain because a model was confident about it. A person approves, and your data is not used to train public models.
Before you loop in your CIO
The questions asset managers ask.
Straight answers, including the ones we will not give a number for.
No, and we would not recommend it. Those keep doing the thing they are good at, which is showing you what already happened once somebody loaded it. What we add is the layer underneath that reads the documents behind the numbers, the budgets, the loan agreements, the capex approvals, the prior reporting, and reviews every asset against them on a schedule. A dashboard answers what happened. This answers which assets you should be looking at, and why.
You define it once, during the audit, and then it stops being tribal knowledge. Thresholds, covenant terms, watch-list rules, and the exceptions your firm has always made in practice get written down and become part of the Brain. It is not a generic anomaly detector deciding on your behalf what should worry you.
It drafts it. Every figure carries the document it came from, and the narrative is written from those same sources rather than from a model's impression of the quarter. Then a person reads it, changes what needs changing, and signs. Nothing reaches an investor, a lender, or a board without that step.
Into your environment, not ours. Deployments run in your firm's own Claude or Microsoft Copilot workspace, or in your own cloud environment, scoped to the systems and folders you approve. Your portfolio data is not used to train public models, and access is reviewable by the people who own the data.
Breadth is relative to the team, not to the asset count. Twelve assets covered by two people who also handle leasing questions, lender reporting, and the annual budget is the same sampling problem in a smaller frame. The build gets scoped to the portfolio. The first question we would ask is what your reporting calendar actually costs you in attention, not in hours.
The first phase is a read of how your firm genuinely reports, including the parts nobody documented. Then shadow mode against a period you have already closed, so you can check the system against work you already know is right before it touches anything live. We would rather you catch it in shadow than trust it in production.
No. It reads, drafts, and flags. Sending, approving, filing, and committing stay with people, and the trail shows who did each one and when.
It helps, because it makes the disagreement specific. When a variance comes back with the source line attached and a drafted question for the property team, the conversation is about the explanation rather than about whose spreadsheet is right.
We will not put a salary comparison on a marketing page, because the honest answer depends on your portfolio, your reporting obligations, and how much of the work is genuinely repeatable. The structural difference is worth stating though: another hire adds one person's capacity and still samples, and the institutional memory leaves when they do. Pricing comes after the assessment, against real scope.
Let’s talk
Find out what your hold period is actually costing you in attention.
Take the free AI assessment. Ten minutes, an honest read on where your reporting stands, and which engagement fits, if any.
We take on a limited number of new partners each quarter.
Prefer email? zed@pacificsoftwareventures.com
Guardrails built inAES-256 encryptedYour data stays yoursOngoing optimization included