A first look at the skills
Ten single-skill outputs run on a randomly selected value stock and what they output.
How it works
Before pushing into the full institutional-style initiation report, I ran each major skill in isolation against a real company. The point was to see what each skill actually produces — how far away was the output from something we expect an industry professional to produce.
The featured stock was a random value stock (Telefónica), bur also incorporated a small basket of different stocks: Telefónica (Euro deep-value telecom anchor), Evergreen Marine (Asian deep cyclical), Berkshire Hathaway (US value icon), Porsche SE (holding company discount), and Microsoft (growth stock). Most of the single-skill demos focus on Telefónica — the deepest test case — with a few using other names where there was focus on more than one name.
Each skill output is markdown (simple plain language text). Some skills (earnings-analysis, initiating-coverage) target Word / PowerPoint / Excel as the the output; the markdown captures the analytical content without the formatting wrap.
The ten outputs
- 01
Telefónica + 6 European telecom peerscomps-analysisPeer multiples table (P/E, P/B, EV/EBITDA, FCF yld, div yld, ROE, leverage) with min/25th/median/75th/max statistical bands; TEF position highlighted.
- 02
Vodafone-anchored deep-value telecom screenidea-generation (/screen)84-name universe filtered to 18 hits on EV/EBITDA, P/B, dividend yield criteria. Top-3 ranked with thesis bullets, key risks, and data quality flags.
- 03
European Telecom Servicessector-overviewMarket structure, competitive landscape, valuation context, key debates, investment implications. Plus the 2025-26 capital-allocation reset wave context (TEF + Vodafone dividend cuts).
- 04
Telefónica strategic positioningcompetitive-analysisFive-forces, moat assessment, segment-by-segment competitive landscape, key risks.
- 05
Telefónica long thesisthesis-trackerThesis statement, five pillars with status, key risks, catalyst calendar, scorecard, update log, conviction level.
- 06
5-name basket, 90-day forward lookcatalyst-calendarCalendar table with date / event / company / type / impact / positioning / notes. Weekly preview format and 30-day forward look.
- 07
Coverage basket digestmorning-notePre-market basket digest — top idea, sector reads, what changed overnight, watch items. Native format is short markdown / email.
- 08
Microsoft Q3 FY26 (illustrative)earnings-analysisBeat/miss summary table, drivers, management commentary, updated estimates, valuation impact, thesis check. The skill targets 8-12 page DOCX; this is the analytical content in markdown.
- 09
Telefónica Q1 2026earnings-previewConsensus expectations vs our estimate, top 5 watch items, bull/base/bear scenarios with probability weights, trade construction, recent estimate revisions.
- 10
Telefónica post-Q1 hypotheticalmodel-updatePlug actuals into model with delta vs prior estimate, forward estimate revisions table, valuation impact, recommendation action. Distribution-ready summary paragraph.
What this looks like in practice
Run individually, the skills produce credible-looking analytical content with the structural hallmarks of institutional research: statistical bands, scenarios with probability weights, beat-miss tables, dated update logs. Each one is on its own a useful scaffold for the kind of analysis a junior analyst might draft.
What you don’t see at this level is whether the specific facts in each output are right. That’s the question chapters 3 and 4 take up — running the full initiating-coverage pipeline, then fact-checking the result.
Practitioner take
From a sell-side perspective, or a shorter-term, earnings momentum focused buyside firm view, some of these tools can do things which are incredibly burdensome - if you trusted the output. Thinking back to my sell-side days, being able to have quick access to comp tables, calendars/catalysts, sales-trading morning notes and quarterly earnings analysis/model updates would have been a godsend as you headed to speak on the trading floor five minutes after the earnings have been released.
For longer-term investing, having the ability to track thesis and have them automatically updated would be beneficial for a portfolio. Likewise, a quick pass on identifying the key issues in a new sector may be a helpful backdrop for starting work. My recurring concern is whether these products have consensual outcomes built into them, whether you can trust the outcome and making sure there is a clear delineation between increasing the ability to parse more data and actual decision making.
Chapter 2 of 7