I built a 20-page take-private discussion deck, model included, in under an hour, and it’s a deck I’d be comfortable putting in front of my old VP. Here’s a preview of the result, and then I’ll walk through how it came together.
I started my career as an investment banking analyst and spent four years building decks like the one in this post, mostly after midnight. Each one took days: pulling the data, updating the model, rebuilding the analysis in Excel and then transferring it into slides. The work spread across several tools, and much of it was incredibly tedious and mind-numbing.
With the launch of SuperAnalyst, I put our new agent to the test across a real banking workflow to see if it could produce something not just quickly, but something that wouldn’t get me laughed out of the room.
The setup: our firm is advising on a take-private of H&M, and my MD wants special committee discussion materials like the ones we did on a previous deal.
(This is a hypothetical exercise for illustration only. It does not reflect any real or planned transaction and is not investment advice.)
The Full Workflow
The Starting Deck
In a real scenario, I’d have previous discussion materials to use as a jumping off point. For this experiment, I used a publicly available deck: the discussion materials a bank presented to the special committee of Taro Pharmaceuticals on a take-private. Executive summary, public market perspectives, projections, football field, DCF, everything you’d expect.
Mine happened to be a PDF, but a PowerPoint or screenshots of a few slides work just as well. Whatever the format, SuperAnalyst creates slides that look like yours.
OK, our timer starts now.
Step 1: Save it as a Template (1 minute)
I uploaded the deck to SuperAnalyst and asked it to rebuild it as a take-private template.
Strictly speaking, this step isn’t needed, since I could have just attached the deck in step 3. But in this case, I saved it as a template because I want to be able to reuse it. The next time I need take-private committee materials for a different company, I can start from the template instead of digging up the original deck again.
Step 2: Rebuild the Excel Analysis (26 minutes)
Now that it has my template, I asked SuperAnalyst to rebuild the analysis, pointing it at a few reference files: the Canalyst model for H&M and a comps file of apparel retailers, both of which already live inside of AlphaSense.
Canalyst models are analyst-built models of public companies with detailed historicals and forecasts. Because this already lives in AlphaSense, SuperAnalyst started with a full H&M model instead of a blank workbook.
This is my exact prompt, so you can try it out yourself:
Add a new tab named “Take-Private Valuation” to the H&M Canalyst workbook, preserving all existing tabs. Build a formula-linked analysis of a hypothetical Persson-family buyout of minority shareholders, using SEK millions except per-share figures. Include:
- Financial outlook: revenue, EBITDA, EBIT, margins, capex, and cash flow through FY2030, linked to the model.
- DCF: construct unlevered free cash flow, use editable WACC and terminal-growth assumptions, and show implied value per share with a sensitivity table.
- Trading comparables: use the attached apparel CSV to select relevant peers and derive valuation ranges from forward EV/EBITDA and P/E.
- Offer scenarios: calculate prices at 20%, 30%, and 40% premiums to a clearly dated reference price, showing total equity value, enterprise value, implied multiples, and cash required to acquire minority shares.
- Summary: compare DCF, peer-implied values, and offer prices in a valuation chart. Clearly separate sourced inputs from assumptions, reconcile valuation dates and currencies, treat leases consistently, and distinguish Canalyst forecasts from consensus. Use an editable ownership assumption for the minority buyout calculation. Use Taro pdf to understand how tables and analyses should look like.
SuperAnalyst worked for 26 minutes, which is less time than I used to spend lining up logos on a comps page. It added a full take-private tab to the Canalyst model: outlook, DCF with sensitivities, comps and offer scenarios. All numbers are formulas or link back to the source model, so you can check them and change assumptions.
One thing I liked: It ran the DCF on both the Canalyst forecasts and a consensus case and called out the gap, SEK 207/share on Canalyst numbers vs SEK 113/share on consensus, with the offer prices in between. It caught the gap before my hypothetical VP did, which in my experience was not the usual order of events.
Which of these cases to believe is still our judgment call. What changed is that I got to that question in 26 minutes instead of two days.
Consensus vs Canalyst and the DCF build
Peer-implied values and offer scenarios
Step 3: Build the Deck (23 minutes)
In a real scenario, I may want to take this analysis into Excel and make some of my own edits to the assumptions there. In this case, I’m happy with everything in the file and move on to the next step.
In that same thread, I asked SuperAnalyst to build the H&M deck using the template from Step 1:
Build me a take-private deck in the template I attached earlier. Use the analysis from the Excel model you created. Search and fill out the missing data as necessary.
For this step, SuperAnalyst worked for 23 minutes.
The result follows the original deck page by page: executive summary, share price, and trading multiple charts from AlphaSense market data, financial tables from the Canalyst model, and the football field from the valuation in Step 2.
That’s a 20-page deck: 9 native charts and 14 slides with native tables, all editable with SuperAnalyst and in PowerPoint, plus 47 cited sources at the back. It’s the kind of draft I’d have been happy to hand to my VP, and it took 23 minutes instead of a weekend.
Original vs SuperAnalyst: valuation summary
Original vs SuperAnalyst: public market perspectives
With this deck, SuperAnalyst got me 95% of the way there, and now I can swap in the real fixes. Making those changes was easy too, a couple of chart labels needed nudging and I wanted to tighten the executive summary wording, which I did with our PowerPoint add-in. I can now make those last edits using AI inside the deck itself, without bouncing back to a prompt.
Step 4: Save as a Custom Skill (1 minute)
This one’s optional too, but a good practice (and it only adds another minute to my time). My final step was to save this entire workflow as a Skill so that SuperAnalyst understands that whenever I ask it to do this type of work in the future, it needs to first build me the model with a clear explanation for where the numbers came from, and then build the deck.
All I have to do is instruct SuperAnalyst: “Create a skill based on this workflow,” and it will ask me any clarifying questions it needs. Then, whenever I want to rebuild this analysis for any new name, I just have to run the skill, and it has all my materials and preferences saved.
My Key Takeaways
Here is what I’d do again, and what I’d tell any analyst or associate who wants to try it.
- Show it what good looks like. Don’t describe the deck you want; attach one. A precedent deck, a few slides from an old pitch, even a screenshot. SuperAnalyst picks up the structure, how dense each page is, the chart types and the formatting, and builds yours the same way. The better the example, the better the output.
- Split it into steps on purpose. It can do it all in one prompt, but doing the model first meant I understood the numbers before they went on slides. Same with the deck: start with the core pages, then attach slides you like and ask it to add them.
- An analyst still owns the numbers. What goes away is the mechanical work, and the 2 a.m. fights with footnote formatting.
This wasn’t possible a year ago. Back then, our slides took a research report and poured it into one of a handful of fixed templates: a title, five bullets, maybe a bar chart. Ask for a football field and you’d get a well-formatted paragraph about valuation. Now, with flagship models, the right tools in SuperAnalyst, and AlphaSense data behind every number, any slide is mechanically possible, for any company.
In the end, the timer stopped at 51 minutes: 1 for the template, 26 for the model, 23 for the deck (and one more minute if you save as a Skill). And during most of those minutes, I could focus on other things while SuperAnalyst worked in the background. It’s still a draft, and I’d verify every number before sending it to my VP. But the part that used to eat my weekends is now the short part, and I get to spend the time on the question that actually needs an analyst, which is whether the numbers are right.
Watch me walk through the entire workflow here.
Want to try it on your own work? Log in to AlphaSense and start exploring how SuperAnalyst can accelerate your workflows or sign up for a free trial.
