The Junior Analyst Who Never Sleeps: How Claude Sonnet 4.5 Became the Office MVP
No Coffee Breaks, No Small Talk, and Somehow Still Better at Excel Than Brad from Finance
It was one of those Tuesdays that pretends to be calm. Then a client dropped a 41-page PDF, a lopsided CSV, and a request for a six-slide summary before lunch. I opened Claude Sonnet 4.5, described the outcome, and watched it build a clean Excel with working formulas, pull charts that matched the math, draft a Word brief, then spin the whole thing into a tidy PowerPoint. I still did the judgment and the edits. The heavy lift moved off my plate, which is the point.
This past Monday, September 29th, Anthropic released their latest model, Claude Sonnet 4.5. If you think Sonnet 4.5 is mainly a coding story, you’re missing the bigger shift. The model handles real office files. Not pretend mockups, actual .xlsx, .pptx, .docx, and .pdf you can hand to a colleague without apologizing. Ask for a revenue model with named ranges and a scenario tab, it creates it. Ask for a QBR deck from a research memo; it lays out slides with titles, bullets, and charts that are tied to the same spreadsheet it produced. Hand it a messy PDF, and it extracts tables into an Excel sheet you can audit. The experience feels like a sharp junior analyst who works very fast and does not complain about formatting.
The spreadsheet skills are the part that surprised me. You can be specific, sheet names, target cell ranges, formula types, chart styles. It writes SUMIFS and INDEX-MATCH without getting cute, labels assumptions, and documents the steps so you can follow the logic. That matters when you share the workbook with a CFO who will click into cells and check references. I ask it to insert a “Read Me” tab with sources, caveats, and a short list of things it would want if it had more data. It does, and it saves me an email thread.
PowerPoint is similar. Give it a desired arc, for example, “Problem, Evidence, Options, Recommendation, Next steps,” and it will assemble a six to eight-slide draft that is more than clip art and lorem ipsum. I usually swap a couple of layouts, tweak chart axes, and tighten language. You still own the story. You just start from something workable instead of a blank slide at 10 p.m.
Word and PDF round out the loop. For Word, Sonnet 4.5 is strong at executive summaries, board memos, and redline-aware rewrites. For PDFs, it can read long documents, pull the tables, cite what it used inside the draft it produces, and keep section references straight. I have used it on contracts, on industry reports, on a 70-page diligence packet that would have taken a day to triage. Hours, not days.
None of this happens in a vacuum. Microsoft is wiring Anthropic’s models, including Sonnet 4.5, into Copilot Studio and across parts of its Copilot experiences for Microsoft 365. If your team lives in Excel, Word, and PowerPoint already, this means the same model that can build your workbook in a chat can also sit inside the tools you use every day. The path from one productive session to a repeatable internal agent gets shorter.
The coding talk is real too. Sonnet 4.5 scores well on tough software benchmarks, and it holds context over long sessions. Oddly, that stamina shows up in office work. The same traits that help it debug an application help it keep a spreadsheet model, a memo, and a deck aligned over a morning’s work. Ask it to revise the Excel assumptions, then refresh the charts, then update the slide captions and the Word summary to keep the narrative consistent. It can do that.
And then there is the writing. Creative writing tests have it at or near the top, and you can feel that range in business drafts. It can shift tone, adjust for an executive audience, and keep a through-line when you ask for tighter language or a different angle. I use it to produce a first pass, then I annotate for voice and audience. It follows direction without losing the point. Now I will say that my personal experience with using Claude Sonnet 4.5 (and Opus 4.1) for writing is really good, but I’m not as positive as others I see reviewing the tools. It is very good, but I still find ChatGPT-5 Thinking even better (that’s the only one that I use regularly that beats it).
How do you get real value out of it if you are a busy manager, director, or partner who spends most days in documents, not code?
Start by writing specs, not vague prompts. Ask for concrete outputs, file types, sheet names, formulas to include, chart types, slide count, and target reader. Tell it your house style. Then review like a hawk. Ask it to expose assumptions and list weak spots. When it shows its work, your edits get faster.
Instrument the repeatables. If you build the same board packet or QBR every month, give Sonnet 4.5 a standard input file and a checklist. Keep a short paragraph that explains your voice, a data dictionary, and a directory where it can find last month’s files for comparison. The model respects patterns, which saves you from re-explaining preferences.
Connect the numbers and the words. Have it generate the Excel first, then ask for a memo or slide narrative that explains the variances and the why. Because both come from the same session, the language lines up with the math, and you spend less time fixing mismatches.
I do not think this replaces judgment. It shortens the distance from messy input to a credible first draft across the tools office people touch all day. That is the move. And since Sonnet 4.5 is landing inside the Microsoft world so many of us already live in, I expect the next wave of “small but huge” wins to come from people who quietly build two or three repeatable agents that prepare their weekly reports before the coffee cools.
My verdict after steady daily use: this is an amazing model for business users. It makes the work feel lighter without making it feel lazy. That line matters.
I’m going to make it my main model for the next two weeks (unless Gemini 3 comes out!) and really take it for a spin. I’ll let you know how it goes!
Moving Forward with Confidence
The path to responsible AI adoption doesn’t have to be complicated. After presenting to nearly 1,000 firms on AI, I’ve seen that success comes down to having the right framework, choosing the right tools, and ensuring your team knows how to use them effectively.
The landscape is changing quickly - new capabilities emerge monthly, and the gap between firms that have mastered AI and those still hesitating continues to widen. But with proper policies, the right technology stack, and effective training, firms are discovering that AI can be both safe and transformative for their practice.
Resources to help you get started:
In addition to publishing thought AI leadership on a regular basis, I also work directly with firms to identify the best AI tools for their specific needs, develop customized implementation strategies, and, critically, train their teams to extract maximum value from these technologies. It’s not enough to have the tools; your people need to know how to leverage them effectively.
For ongoing insights on AI best practices, real-world use cases, and emerging capabilities across industries, consider subscribing to my newsletter. While I often focus on legal applications, the broader AI landscape offers lessons that benefit everyone. And if you’d like to discuss your firm’s specific situation, I’m always happy to connect.
Contact: steve@intelligencebyintent.com
Share this article with colleagues who are navigating these same questions.