Explore

Slice the corpus on any axis.

Filter by firm, year, function, metric or slide category — then see the result as a narrative arc, a treemap of slide types, or a catalog of slides.

clear
51,416 slides across 1229 decks match.
Treemap Arc Catalog
HR must enable the workforce experience across three employee experience dimensions - individual, team, and operating environment
MorganStanley · 2025 · framework_other
HR must enable the workforce experience across three employee experience dimensions - individual, team, and operating environment
39% — employee turnover intent
 Unprecedented disruption will impact the workforce, meaning HR functions of tomorrow must be ready by focusing on the following areas:
MorganStanley · 2025 · strategic_options
Unprecedented disruption will impact the workforce, meaning HR functions of tomorrow must be ready by focusing on the following areas:
 Transforming HR requires a bold leap forward on the innovation S-curve
MorganStanley · 2025 · framework_other
Transforming HR requires a bold leap forward on the innovation S-curve
 HR business partner role will require fundamental changes to meet future needs
MorganStanley · 2025 · comparison_table
HR business partner role will require fundamental changes to meet future needs
 HR business partner role will require fundamental changes to meet future needs
MorganStanley · 2025 · key_messages
HR business partner role will require fundamental changes to meet future needs
 Top five concerns for March 2025: trend data
misc · 2025 · industry_trends
Top five concerns for March 2025: trend data
 Inflation/prices – concern since 1974
misc · 2025 · industry_trends
Inflation/prices – concern since 1974
54% — Public concern level
 Lack of faith in politicians/politics
misc · 2025 · industry_trends
Lack of faith in politicians/politics
26% — Percentage of respondents
 What’s driving global growth in IT services?
LEK · 2017 · industry_trends
What’s driving global growth in IT services?
 Most companies expect to increase their spend on outsourced IT
LEK · 2017 · industry_trends
Most companies expect to increase their spend on outsourced IT
50% — Percentage of respondents
 Outsourced IT has huge upside — $1 trillion in annual spend by 2018
LEK · 2017 · market_sizing
Outsourced IT has huge upside — $1 trillion in annual spend by 2018
$1 trillion — Annual IT spend
 Fragmented IT spend: more on specialized service providers, less on legacy providers
LEK · 2017 · financial_analysis
Fragmented IT spend: more on specialized service providers, less on legacy providers
92 — Average spend per vendor
 Specialists preferred by 2-to-1 margin over generalist vendors
LEK · 2017 · market_landscape
Specialists preferred by 2-to-1 margin over generalist vendors
39% — Percentage of respondents
 Generalist vendor response: aggressive moves to meet new market realities
LEK · 2017 · case_study
Generalist vendor response: aggressive moves to meet new market realities
9% — Operating profit
 Implications for vendors and investors: big opportunities for those who know where to look
LEK · 2017 · recommendation
Implications for vendors and investors: big opportunities for those who know where to look
 Scorecard: Reviewing our predictions from 2024
AirStreetCapital · section_divider
Scorecard: Reviewing our predictions from 2024
 Think before you speak: o1 “thinking” ignites the reasoning race
AirStreetCapital · industry_trends
Think before you speak: o1 “thinking” ignites the reasoning race
 Deeply sought: could frontier reasoning ever be found in the open?
AirStreetCapital · case_study
Deeply sought: could frontier reasoning ever be found in the open?
52.5 — AIME 2024 pass@1
 Are you not entertained? DeepSeek V3 brings you to R1
AirStreetCapital · case_study
Are you not entertained? DeepSeek V3 brings you to R1
79.8 — AIME accuracy
 More thinking, more tool use, less cost: DeepSeek V3.1 and V3.2-Exp
AirStreetCapital · industry_trends
More thinking, more tool use, less cost: DeepSeek V3.1 and V3.2-Exp
 Parallel reasoning: beyond depth to branch-and-merge inference
AirStreetCapital · other
Parallel reasoning: beyond depth to branch-and-merge inference
83.4% — Accuracy
 The reasoning timeline: from o1 “thinking” to R1, GPT-5 and parallel compute routing
AirStreetCapital · timeline
The reasoning timeline: from o1 “thinking” to R1, GPT-5 and parallel compute routing
 12 months pass, and OpenAI models remain at the frontier of intelligence
AirStreetCapital · industry_trends
12 months pass, and OpenAI models remain at the frontier of intelligence
68 — Intelligence Index Score
 The illusion of reasoning gains
AirStreetCapital · diagnosis
The illusion of reasoning gains
 How reasoning breaks: minor variations
AirStreetCapital · diagnosis
How reasoning breaks: minor variations
7x — error rate
 How reasoning breaks: small shifts cause big failures
AirStreetCapital · diagnosis
How reasoning breaks: small shifts cause big failures
65% — accuracy
 CoTs still work, even when models aren't honest...
AirStreetCapital · key_takeaways
CoTs still work, even when models aren't honest...
99% — Detector Sensitivity
 ...but there are safety concerns, like the “AI Hawthorne effect”
AirStreetCapital · diagnosis
...but there are safety concerns, like the “AI Hawthorne effect”
 Safety by design meets its skeptics
AirStreetCapital · industry_trends
Safety by design meets its skeptics
8.3% — Attack Success Rate
 Trading capability for transparency
AirStreetCapital · industry_trends
Trading capability for transparency
95% — Fraction of Samples
 Monitor your CoTs: just because you can read 'em, doesn't mean you should trust 'em
AirStreetCapital · diagnosis
Monitor your CoTs: just because you can read 'em, doesn't mean you should trust 'em
 But can LLMs reason without generating tokens?
AirStreetCapital · industry_trends
But can LLMs reason without generating tokens?
 Researchers begin to prioritise quality and diversity of post training data over volume
AirStreetCapital · industry_trends
Researchers begin to prioritise quality and diversity of post training data over volume
3x — Average Score
 The evolution of AI reward signals towards environments with verifiable rewards
AirStreetCapital · other
The evolution of AI reward signals towards environments with verifiable rewards
 Even so, RL environments and evaluations are fraught with challenges
AirStreetCapital · problem_statement
Even so, RL environments and evaluations are fraught with challenges
 RL from verifiable rewards: promising, but the evidence cuts both ways
AirStreetCapital · key_messages
RL from verifiable rewards: promising, but the evidence cuts both ways
 Ushering in an era of AI-augmented mathematics
AirStreetCapital · industry_trends
Ushering in an era of AI-augmented mathematics
57.6% — Pass@32
 Bigger models, same budget: RL with LoRA adapters
AirStreetCapital · industry_trends
Bigger models, same budget: RL with LoRA adapters
 Beyond reasoning is...continual learning?
AirStreetCapital · industry_trends
Beyond reasoning is...continual learning?
 Too many cooks?
AirStreetCapital · industry_trends
Too many cooks?
89.3 — Accuracy (%)
 The Muon Optimizer: expanding the compute-time Pareto frontier beyond AdamW
AirStreetCapital · case_study
The Muon Optimizer: expanding the compute-time Pareto frontier beyond AdamW
10-15% — Training Time
 Cutting your losses: significant memory reduction for LLM training
AirStreetCapital · case_study
Cutting your losses: significant memory reduction for LLM training
24x — Memory consumption
 How much do LLMs memorize?
AirStreetCapital · other
How much do LLMs memorize?
~3.6 — bits per parameter
 Learning from superintelligence: AlphaZero teaches chess grandmasters new concepts
AirStreetCapital · case_study
Learning from superintelligence: AlphaZero teaches chess grandmasters new concepts
0.85 — improvement in puzzles solved
 Kimi K2: stable trillion-scale MoE for agentic intelligence in the open
AirStreetCapital · case_study
Kimi K2: stable trillion-scale MoE for agentic intelligence in the open
 Open source vs. proprietary: where are we now?
AirStreetCapital · market_landscape
Open source vs. proprietary: where are we now?
67 — Artificial Analysis Intelligence Index
 OpenAI pivots from the “wrong side of history” to aligning with “America-first AI”
AirStreetCapital · case_study
OpenAI pivots from the “wrong side of history” to aligning with “America-first AI”
20k — GitHub Stars
 The New Silk Road: China's open models overtake the previously Meta-led West
AirStreetCapital · industry_trends
The New Silk Road: China's open models overtake the previously Meta-led West
63 — Model Adoption
Next page →