The 60-hour junior analyst week is ending — not because analysts are being asked to work less, but because most of what filled those hours was never analysis. Copying rent-roll figures into a model, retyping lease terms from a scanned PDF, rebuilding the same comps sheet for the fourth deal this month: that work absorbed the evenings and the weekends, and a general-purpose AI tool now does the first pass of nearly all of it in minutes. For a 4-to-20-person commercial real estate firm, this is not a threat to the analyst and it is not a reason to buy software. It is a chance to end a grind that was quietly costing the firm money, talent, and accuracy — and to redeploy the reclaimed hours into the judgment work that actually wins deals. This piece explains what really fills the long week, what AI removes and what it leaves untouched, and how a small firm ends the grind deliberately instead of waiting for it to end badly.
What Actually Fills the 60-Hour Week
Start by naming honestly where the hours go, because the answer is not “deep analysis.” Industry analyses of commercial real estate underwriting workflows put data ingestion and model population — pulling numbers out of rent rolls, operating statements, and offering memorandums and typing them into a spreadsheet — at roughly 40 to 60 percent of a junior analyst’s day. The long week is mostly transcription.
Picture the typical week at a small brokerage or investment shop. A new deal arrives as a stack of PDFs: a rent roll formatted one way, a T-12 formatted another, a lease abstract that has to be read line by line. The analyst opens the model template and starts copying. Every unit, every expense line, every escalation clause moves by hand from a document a machine could not read into a spreadsheet a machine could. That is the bulk of the work, and it is the part that runs into the evening.
The second layer is repetition. The same comps set gets rebuilt for the third deal in the same submarket. The same market write-up gets drafted from scratch when last quarter’s version needed only a refresh. The same investor email gets composed one careful sentence at a time. None of this is hard, and none of it improves with the fifth repetition — it simply takes hours that have to come from somewhere.
The judgment layer — is this a good deal, is this sponsor credible, does this rent roll hide a concentration risk — is the smallest slice of the week and the only slice the firm is actually paying for. The 60-hour week is what it costs to get a few hours of real thinking out from under a mountain of transcription.
The Long Week Is a Cost, Not a Badge
In a lot of firms the long analyst week is worn as evidence of diligence. It is worth being blunt: it is not diligence, it is unpriced overhead, and the owner is the one absorbing it.
Three costs hide inside the grind. The first is turnover. A bright junior who took the job to learn how deals get done, and instead spends most of a 60-hour week retyping lease terms, does not stay long — and every departure resets the firm’s training clock and burns a recruiting cycle. The second is error. Manual transcription across dozens of documents, late at night, is exactly the condition under which a transposed number slips into a model and quietly changes a decision. The third is opportunity cost: the hours poured into rebuilding a comps sheet are hours not spent touring a building, calling a sponsor back, or second-looking an underwriting that felt slightly off.
There is also a development cost that compounds. Analysts learn investment judgment by doing analysis, not by doing data entry. A junior who spends most of the week transcribing learns the craft slowly, which keeps them dependent on the principal’s review longer and keeps the principal in the weeds. The grind does not just cost hours now; it slows the day the analyst becomes genuinely useful.
Deloitte’s 2026 Commercial Real Estate Outlook found the share of firms reporting a genuinely transformative impact from AI rose to 7 percent in 2025 from 1 percent the year before, with incremental-improvement reports climbing to 24 percent from 7 percent — and it named the AI skills gap, not the technology, as the biggest barrier to getting there. The firms pulling ahead are not the ones with the most software. They are the ones whose people learned to hand the transcription to a machine. This is the same timing logic that makes waiting a small firm’s advantage rather than its handicap, covered in our look at why CRE is late to AI.
What AI Removes, and What It Leaves Behind
Be precise about what is actually changing, because the honest version is more useful than either the vendor pitch or the career-panic headline. AI removes the transcription and the repetition. It does not remove the analyst.
A current general-purpose model — ChatGPT, Claude, or Gemini — reads a messy lease PDF, a rent roll, or a T-12 and returns a structured first pass in the time it takes to get coffee. What used to be a week of underwriting compresses hard: banks deploying AI underwriting on commercial loans report 50 to 75 percent reductions in time-to-decision, and initial deal screening that ran for hours now runs in well under one. Industry reporting suggests AI has already absorbed something like 30 to 40 percent of the tasks that filled the 2024 analyst week. The transcription layer is the part that evaporates.
What stays is everything the transcription was in the way of. Whether the deal is actually good. Whether the sponsor’s track record holds up. Whether a rent roll that models cleanly hides tenant concentration or a lease that rolls at the wrong moment. How to structure the offer, what to say on the call, which relationship to protect. McKinsey has estimated generative AI could unlock on the order of $110 to $180 billion of value across real estate — and none of that value is the model deciding to buy. It is people making better decisions faster because the machine cleared the desk.
The right mental model is editor, not author. The analyst stops being the person who assembles the raw material by hand and becomes the person who checks, corrects, and judges the machine’s first draft — a shift from data processor to decision-maker that is the real subject of our piece on rethinking the analyst role in a small investment firm. The output still needs a human who knows what “wrong” looks like on a rent roll. AI makes that human faster; it does not make them optional.
Fluency First, Software Second
Here is where most firms get the sequence backwards, and the mistake is expensive. Feeling behind, they shop for a tool — an AI underwriting platform, a document-extraction subscription — before their own people can reliably prompt a model to summarize a lease. That is the move that produced a decade of underused proptech subscriptions, and it is the wrong first step.
The first step is fluency, and it is cheaper and faster than a purchase. Teach the analyst — and ideally the whole desk — to use the general models the firm can already access for a few dollars a seat to do real work: draft a first-pass lease summary, pull a rent roll into a clean table, produce a market write-up, tighten a client email. This is a low-thousands investment, and focused training for a small team generally lands in the market range of roughly $2,000 to $15,000, not a capital project. It captures value in the first week and, just as important, it shows the firm which parts of the job a prompt already solves and which parts genuinely need more.
Fluency first also protects the firm on confidentiality, which is not a footnote when the documents are deal terms and tenant financials. Learning to use these tools well includes learning which account tiers keep your data out of training, what belongs in a prompt and what does not, and how to redact before you paste. That judgment is part of the training, not something a purchased platform hands you. Knowing the vocabulary well enough to ask a vendor the right questions matters too, which is why a shared baseline like our plain-English glossary of CRE AI terms is worth an afternoon.
Only after the desk is fluent does a software or automation question become answerable. Once your people have done the work by hand with a model, they can see the one workflow that is still slow — usually something document-heavy and specific to how your firm operates, like reconciling CAM across a portfolio — and scope it properly. A custom automation for that kind of task runs in the market range of roughly $25,000 to $150,000 depending on complexity, and it is worth building only when fluency has already shown you exactly what it needs to do. The full 90-day version of this sequence, from first prompt to a fluent desk, is laid out in the CRE AI training playbook.
What the Reclaimed Hours Are Actually For
Ending the grind creates a question the firm has to answer on purpose: what happens to the hours. Get this wrong and the gain evaporates into more deals half-underwritten. Get it right and it is the whole point.
The reclaimed time is not a cost to bank by cutting the analyst. For a lean firm, more analyst capacity at the same headcount is the prize, not fewer people — the near-term reality across the industry is fewer analysts each doing more, not zero analysts. The move is to redeploy the hours up the value chain: from transcription toward the judgment and relationship work that was always the actual job.
Concretely, the analyst who is no longer copying rent rolls until 9pm can second-look the deals that felt off, build the comps narrative instead of just the comps grid, sit in on sponsor calls, and take on the market read the principal never had time to delegate. That is how a junior turns into a genuine underwriter faster, which is the retention argument answering itself — the job finally resembles the one they were hired to want.
For the principal, the reclaimed hours press directly on the firm’s real constraint, which in a small shop is never the software budget and always the number of good decisions the team can make in a week. A firm that clears the transcription off its analysts’ desks can look at more deals, respond faster, and think harder about the ones that matter — which is exactly how a disciplined small firm out-operates a larger, slower competitor, the argument at the center of the small-firm CRE manifesto.
Frequently Asked Questions
What tasks actually fill a junior CRE analyst’s 60-hour week?
Mostly transcription and repetition, not analysis. Industry workflow analyses put data ingestion and model population — copying figures from rent rolls, operating statements, and offering memorandums into a spreadsheet — at roughly 40 to 60 percent of a junior analyst’s day. On top of that sits repeated work like rebuilding comps sets, redrafting market write-ups from scratch, and composing routine investor emails. The genuine judgment work — is this a good deal, is this sponsor credible — is the smallest slice of the week, and the long hours are mostly what it costs to dig it out from under the manual work.
Can AI really cut analyst hours, or does it just shift the work around?
It removes the work, not just moves it. A current general-purpose model reads a messy lease PDF or rent roll and returns a structured first pass in minutes, and banks deploying AI underwriting report 50 to 75 percent reductions in time-to-decision. Industry reporting suggests AI has already absorbed something like 30 to 40 percent of the tasks that filled the 2024 analyst week. The analyst still checks and corrects the output, but the hours spent creating it by hand largely disappear rather than reappearing somewhere else.
Does ending the grunt-work week mean I need fewer analysts?
Usually not — it means the analysts you have do more of the work you actually hired them for. The near-term pattern across the industry is fewer analysts each handling more, not teams cut to zero. For a small firm, the value is added capacity at the same headcount: the same people underwriting more deals, second-looking the ones that matter, and taking on judgment work the principal never had time to delegate. Cutting the headcount to bank the saving usually just trades a time problem for a capacity ceiling.
Will AI replace my junior analyst?
No, but it changes the job. AI removes the transcription and repetition; it does not make the market read, the sponsor evaluation, the deal structuring, or the relationship management. The analyst shifts from assembling the raw material by hand to editing and judging the machine’s first draft, which still requires someone who knows what a wrong number looks like on a rent roll. The role moves up the value chain rather than disappearing.
What should a junior analyst do with the reclaimed hours?
Move up the value chain. The analyst who is no longer copying figures until late can second-look questionable deals, build the comps narrative rather than just the grid, join sponsor calls, and take on market analysis the principal previously kept. This is also how a junior develops investment judgment faster, since they learn by doing analysis rather than data entry. The reclaimed time is best treated as added capacity for judgment work, not as slack to cut.
Do I need special software, or can general AI tools do this?
Start with general tools. ChatGPT, Claude, and Gemini already draft lease summaries, structure rent rolls, produce market write-ups, and clean up client email for a few dollars a seat, and getting your people fluent with them is the cheaper, faster, reversible first move. Buy or build dedicated software only after fluency has shown you the one workflow that is still slow and specific to your firm. Buying a platform before your people can prompt a model is the sequence that produced a decade of underused proptech.
Is my confidential deal data safe if analysts use AI tools?
It can be, with the right practices, and learning them is part of getting fluent. That means using account tiers that keep your inputs out of model training, deciding what belongs in a prompt and what does not, and redacting sensitive identifiers before pasting. Deal terms and tenant financials are sensitive, so confidentiality handling should be built into how the team is trained rather than assumed. Treated as a skill rather than an afterthought, AI use is compatible with the discretion the business requires.
How long does it take a small firm to see the analyst week shrink?
The first gains show up in days, not quarters. Once an analyst can reliably prompt a model to summarize a lease or structure a rent roll, the transcription time drops on the next live deal. A fuller shift — where the desk defaults to AI for first-pass work and the principal trusts the process — is more a matter of a focused training effort over a few weeks than a long software rollout, precisely because a small firm has no committee or legacy system slowing adoption.
What does this cost — training versus building an automation?
They are different sizes of decision. Getting a small team fluent with general models is a low-thousands investment, with focused training generally landing in the market range of roughly $2,000 to $15,000. A custom automation for a specific, document-heavy workflow is a larger commitment, typically in the market range of $25,000 to $150,000 depending on complexity, and it is worth building only after fluency has revealed exactly what it needs to do. Fluency first is the cheaper move and the one that tells you whether the bigger spend is even warranted.
Where do I start if my analyst is drowning right now?
Pick one recurring, document-heavy task — lease summaries or rent-roll cleanup are good candidates — and have the analyst run the next real instance through a general model instead of doing it by hand, with someone checking the output. That single change captures time immediately and shows the team what these tools can already do. From there, build a short, deliberate training effort so the whole desk works the same way, and only then ask whether any remaining bottleneck justifies dedicated software.
Where to Start
The 60-hour junior analyst week is ending whether or not your firm plans for it, and the firms that plan come out ahead. The mistake is to treat it as a software-shopping problem or a headcount decision; it is neither. It is a fluency decision — teach your people to hand the transcription to a machine, then redeploy the reclaimed hours into the judgment work that wins deals. The first step costs almost nothing and pays off on the next live deal. A free AI-readiness assessment gives you a starting map: a short working session that looks at where your analysts’ hours actually go, which of them a general model can already absorb, and where a lean firm’s speed advantage is largest — then returns an honest recommendation on where to begin. Book a free AI-readiness assessment and end the grind on purpose.
Arthur Wandzel
