Tweet
Tweet

Slide 1

Slide 1 text

A Manager's AI Toolkit Unleash productivity & increase efficiency with AI Bastian Grimm | Peak Ace AG | @basgr 17th May 2024

Slide 2

Slide 2 text

We are on the brink of the most profound technological transition in human history.

Slide 3

Slide 3 text

From healthcare to education and jobs – hardly any aspect of our everyday lives will remain unaffected by AI. AI is changing everything

Slide 4

Slide 4 text

Technological developments in recent years serve as the foundation for today’s massive amount of innovations

Slide 5

Slide 5 text

5 peakace.agency Billions of bits of information have been published on the internet for decades – and they are all available as training data for LLMs. #1

Slide 6

Slide 6 text

Machines are now able to take over various human activities and make precise predictions. Progress in Deep Learning #2

Slide 7

Slide 7 text

Significant advances in cloud, infrastructure and consumer technology are making it easier and cheaper than ever to develop and deploy AI technology. Resource Availability #3

Slide 8

Slide 8 text

Those advancements will not only have an impact on AI development, but also on our work life

Slide 9

Slide 9 text

9 peakace.agency How people are using GenAI Six top-level themes give an immediate sense of what generative AI is currently being used for: Source: https://pa.ag/4brddtI - Technical Assistance & Troubleshooting (23%) - Content Creation & Editing (22%) - Personal & Professional Support (17%) - Learning & Education (15%) - Creativity & Recreation (13%) - Research, Analysis & Decision Making (10%)

Slide 10

Slide 10 text

I appreciate that you‘re all extremely busy…

Slide 11

Slide 11 text

Does any of the following look familiar to you?

Slide 12

Slide 12 text

50,000+

Slide 13

Slide 13 text

70,000+

Slide 14

Slide 14 text

How about a packed calendar, day in and day out?

Slide 15

Slide 15 text

My goal for today’s session: Give you back some of your extremely valuable time!

Slide 16

Slide 16 text

16 peakace.agency What did I bring for you today? Email & Messaging Online Meetings Excel & Sheets LinkedIn & Social Media Research & Summaries Audio, Video & Images Workflow Automation Custom GPTs

Slide 17

Slide 17 text

Before we dive in, let‘s talk about the underlying tech for just five minutes:

Slide 18

Slide 18 text

What are Large Language Models (LLMs)?

Slide 19

Slide 19 text

Large Language Models (LLMs) are AI systems trained on vast data sets (thus “large”) to understand, predict and generate data using transformer-based neural networks. Simply put:

Slide 20

Slide 20 text

What are LLMs good at?

Slide 21

Slide 21 text

21 peakace.agency Information Retrieval and Analysis LLMs can sift through large volumes of text data to extract relevant information, summarise key points, and answer questions, making them valuable for research, data analysis, and decision-making support. Personalised Recommendations LLMs can analyse user preferences and behaviour to provide personalised recommendations, such as articles or products, thus enhancing UX and engagement. Natural Language Processing LLMs excel in understanding language, making them ideal for applications such as chat bots, language translation, sentiment analysis, and text summarisation. What are LLMs good at?

Slide 22

Slide 22 text

What are LLMs NOT good at?

Slide 23

Slide 23 text

23 peakace.agency Understanding Context Beyond Training Data LLMs may not perform well in situations requiring an understanding of context or knowledge beyond their original training data set. Making Ethical or Moral Judgments LLMs lack the ability to make ethical or moral judgments and should not be used in situations where such considerations are crucial. Most LLMs’ decisions are also biased. Limited Understanding and Reasoning LLMs can't form a chain of logical conclusions, instead they’re following probability rules; even if the most common answer to a question is irrational or outright wrong, it will still provide said answer. What are LLMs NOT good at?

Slide 24

Slide 24 text

25 peakace.agency With generative AI, 30% of the hours worked today could be automated by 2030. Source: https://pa.ag/3FbhX8J

Slide 25

Slide 25 text

Enough theory, let’s get down to business:

Slide 26

Slide 26 text

AI for: Emails & Messaging

Slide 27

Slide 27 text

According to Microsoft, office workers spend up to 8.8 hours a week on emails (writing, responding, sorting, email threads, …) 2 hours of emails per day! Source: https://pa.ag/3JKzRlc

Slide 28

Slide 28 text

29 peakace.agency Compose AI: A Chrome extension to make you fast(er) A free Chrome extension to automate your writing Source: https://pa.ag/41TToay

Slide 29

Slide 29 text

30 peakace.agency How does this work? Just type "//" on any website to trigger the Compose AI context window:

Slide 30

Slide 30 text

31 peakace.agency MailMaestro: Bringing Open AI LLMs directly to Outlook Generate custom emails and responses based on short prompts; create one click email thread summaries including action points for the receiver: Source: https://pa.ag/3yguk35

Slide 31

Slide 31 text

32 peakace.agency There‘s tons of different solutions out there Depending on your use-case (team usage, web-based, cold emailing, etc.) and "personal taste" you might want to try any of the following:

Slide 32

Slide 32 text

AI for: (Online) Meetings

Slide 33

Slide 33 text

34 peakace.agency People actually spend A LOT of time in (virtual) meetings… The majority of people (83.13% of respondents) spend a third of the week (up to 12 hours) in meetings Source: https://pa.ag/3QyOdbZ As a manager, for me it's not that uncommon to spend more than half of my work week (20+ hours) in different meetings.

Slide 34

Slide 34 text

The cost of bad meetings is more than $37 billion annually Source: https://pa.ag/44BZGxl

Slide 35

Slide 35 text

Mainly due to a lack of clear objectives, not having a pre-set agenda (or no agenda at all) or due to too many attendees. An astounding 67% of meetings are a failure Source: https://pa.ag/3Qx2vKo

Slide 36

Slide 36 text

37 peakace.agency My current favourite: tl;dv.io With tl;dv, a bot will join your meeting ready to record and transcribe. Often, catching up is faster than joining a live meeting – so this will allow you to reclaim some of your valuable focus hours: Source: https://pa.ag/4bI7WhR What I really like: ▪ Instantly jump to relevant meeting moments (using timestamps & tags) ▪ Skim or copy parts of the transcript (in 25+ languages) ▪ Solve time zone differences once and for all with async meetings

Slide 37

Slide 37 text

38 peakace.agency Otter.ai for automated meeting notes & summaries Another meeting assistant that records audio, writes notes, automatically captures slides, and generates summaries: Source: https://pa.ag/3LhCcWC Word of advice (and caution, again): It's super practical but getting consent to record is important, especially since some services can automatically send meeting summaries by email afterwards! Also, I am not a lawyer.

Slide 38

Slide 38 text

39 peakace.agency Very strong with languages: Sembly supports 42 in total! Pretty neat: Automatically stream meeting notes and tasks to task management tools such as Slack, Trello, etc. Bonus: Sembly is GDPR compliant and SOC 2 Type II certified! Source: https://pa.ag/3JRAn0N

Slide 39

Slide 39 text

AI for: Excel & Sheets

Slide 40

Slide 40 text

41 peakace.agency FormulaBot: Turn your problem into a formula Transform your text instructions into Excel formulas, charts or a full-blown analysis in seconds: It’s like ChatGPT inside your spreadsheet and includes: ▪ Sentiment analysis ▪ Text classification & categorisation ▪ Information retrieval ▪ Freeform-style chat dialogue Source: https://pa.ag/4a7JvJ6

Slide 41

Slide 41 text

42 peakace.agency Numerous.ai: A simple way to use ChatGPT on a big scale Numerous.ai provides a simple and very powerful solution for using ChatGPT inside Google Sheets and Excel – because spreadsheets aren't just for numbers! Source: https://pa.ag/3y4ugDN =NUM.AI("extract the name from the text:"; A1)

Slide 42

Slide 42 text

43 peakace.agency MS Copilot + Excel: Taking data analysis to a whole new level! Super pumped for this, as it’ll enable just about anyone to question, analyse, visualise and refine data effortlessly: Source: https://pa.ag/49YhhBe

Slide 43

Slide 43 text

AI for: Summaries, Research, etc.

Slide 44

Slide 44 text

45 peakace.agency Some of my core ChatGPT use cases right now Do research Create outlines Create summaries Validate [things] Ideation & concepts Speed-up learning Simplify content Write code

Slide 45

Slide 45 text

Prompting Most crucial > (for any of those tasks)

Slide 46

Slide 46 text

It's essentially how humans can talk to AI. It's a way to express what we want and how we want it, usually done with words. Prompting?

Slide 47

Slide 47 text

ChatGPT doesn‘t ask questions – it fills gaps with assumptions. In contrast, any human being would ask questions if they didn't understand something. Also keep in mind:

Slide 48

Slide 48 text

49 peakace.agency DIGITAL STRATEGY AI SOLUTIONS PAID SEARCH & SOCIAL DISPLAY & VIDEO CONTENT & DIGITAL PR ORGANIC SEARCH (SEO) DIGITAL ANALYTICS MARKETING AUTOMATION & CRO DELIVERING DIGITAL EXPERIENCES 130+ Awards 200+ Clients 30 Industries 170+ Peak Acers 25+ Languages FOR CONTEXT:

Slide 49

Slide 49 text

50 peakace.agency Use ChatGPT to summarise, simplify and get key take-aways: In addition, it allows you “to chat” with e.g. PDFs (in my case a complex document describing a patent):

Slide 50

Slide 50 text

51 peakace.agency This just saved me half an hour of reading:

Slide 51

Slide 51 text

52 peakace.agency Have a conversation with your PDF:

Slide 52

Slide 52 text

53 peakace.agency For (complex) data, drag ’n’ drop CSV and explain columns E.g., upload Sistrix data exports & explain columns to get a quick first overview of untapped potentials: Convenience = on-the-fly fixes […] data from the spreadsheet appears to be incorrectly delimited, resulting in a single column containing all data […] it has been successfully reformatted, revealing several columns […]

Slide 53

Slide 53 text

You could do the exact same for e.g., your sales data, your teams’ performance data, etc.

Slide 54

Slide 54 text

55 peakace.agency Did you know? You can link using pre-filled prompts! You can also link directly to pre-filled prompts and execute them – which works for both Custom GPTs and GPT-4 models. Simply add the query string (using “q=xxx“) to the end of your ChatGPT URL. Source: https://pa.ag/crsum 𝗙𝗼𝗿 any C𝘂𝘀𝘁𝗼𝗺𝗚𝗣𝗧 𝗮𝗱𝗱: ?q=your+prompt+goes+here 𝗙𝗼𝗿 the 𝗚𝗣𝗧-𝟰 𝗯𝗮𝘀𝗲𝗺𝗼𝗱𝗲𝗹: ?model=gpt-4&q=your+prompt Use directly in your Chrome browser

Slide 55

Slide 55 text

AI for: LinkedIn & Social Media

Slide 56

Slide 56 text

57 peakace.agency Create more and better LinkedIn content with AI using Taplio Establishing a personal brand on LinkedIn is a key pillar of growth. Stop wasting endless hours writing your next post and get fresh ideas using Taplio’s content inspiration layer. Source: https://pa.ag/4b3fSKh

Slide 57

Slide 57 text

58 peakace.agency Again, there‘s tons of different solutions out there

Slide 58

Slide 58 text

AI for: Audio, Video & Images

Slide 59

Slide 59 text

60 peakace.agency Creating AI prompts is a key skill for any generative AI Whoever formulates the best, most creative and most unique AI prompts will create the best content (and save AI/API credits) – this is especially true for images: Source: https://pa.ag/3UO2fcz There is a marketplace (promptbase.com), which is dedicated to trading prompts – use it for inspiration!

Slide 60

Slide 60 text

61 peakace.agency Need help and/or inspiration? Try DS Prompt Builder: Source: https://pa.ag/3JbNxFM Super-simple drag & drop prompt building based on visual elements to choose style, lightning, etc.

Slide 61

Slide 61 text

62 peakace.agency Midjourney V6 has got REALLY good… just wow! Our all-time favourite AI text-to-image generation tool that's available right now: Source: https://pa.ag/3L1O5iE & https://pa.ag/3C11Zwo Naturally, there's more:

Slide 62

Slide 62 text

63 peakace.agency Also, lots of really handy tools such as cleanup.pictures Remove any unwanted object, defect or even people from your picture in seconds Source: https://pa.ag/4aaoJc1

Slide 63

Slide 63 text

64 peakace.agency Recommended use-cases for generative image AI Even though we‘re still at an early stage, the uses mentioned below are very much possible today and allow for time and cost savings, as well as efficiency gains: Header images e.g., for blog posts or newsletters Social images e.g., for your various feeds such as LinkedIn, Insta, etc. Images for slides e.g., for pitch presentations, slide decks or story boards Image modification e.g., clean-ups, object removal, up-scaling, etc.

Slide 64

Slide 64 text

65 peakace.agency beatoven.ai: Create customisable royalty free music beatoven.ai uses AI music generation to compose unique mood-based music to suit every part of your creative piece, such as video (can be dragged and dropped while creating): Source: https://pa.ag/3T9xQD4

Slide 65

Slide 65 text

66 peakace.agency Suno AI takes it to the next level! Essentially, you can turn a single line text prompt into a full-blown song, including lyrics etc, if you want: Source: https://pa.ag/3WuVWf4

Slide 66

Slide 66 text

67 peakace.agency More AI audio & music tools Especially helpful when you need to clean up audio & video recordings (noisy background, irrelevant voices, or just general quality improvements): Krisp AI: Removes background voices, noises and echoes from all your calls Cleanvoice AI: Removes filler sounds, stuttering and mouth sounds from podcasts or audio recordings Podcastle AI: Studio-quality recording, editing, and seamless exporting – in a single web platform soundraw.io: Royalty-free music, AI generated based on mood, genre and length Descript.com: Write, record, transcribe and edit video and podcasts Soundbite AI: Turn new and existing audio and video content into ready-to-edit social media posts, and summaries

Slide 67

Slide 67 text

AI for: Workflow Automation

Slide 68

Slide 68 text

69 peakace.agency Source: https://pa.ag/3Usq4oI Have you tried Make.com? Boost productivity across every area or team. Use Make to design powerful workflows without having to rely on developer resources.

Slide 69

Slide 69 text

70 peakace.agency More than "just" no-code workflow automation Simply drag and drop apps to automate existing workflows or build new complex processes to save time: Source: https://pa.ag/3Usq4oI

Slide 70

Slide 70 text

ChatGPT Special: Building Custom GPTs

Slide 71

Slide 71 text

Custom GPTs are a way to create tailored, custom versions of ChatGPT that combine instructions, extra knowledge, and any combination of skills. What are Custom GPTs (for ChatGPT)?

Slide 72

Slide 72 text

73 peakace.agency A Custom GPT in its simplest form: Using Peak Ace’s Structured Data GPT to debug and fix errors in JSON-LD mark-up

Slide 73

Slide 73 text

Greatly increase your teams‘ productivity with Custom GPTs (and pre-defined workflows)

Slide 74

Slide 74 text

75 peakace.agency Making GPTs smarter with external data A Custom GPT can also be used to fetch additional information from a third-party data-source via API:

Slide 75

Slide 75 text

Here‘s a quick three-step guide on how to DIY it. So, how can you build this yourself?

Slide 76

Slide 76 text

77 peakace.agency #1 Provide basic info to get started (name, description, …) Login to ChatGPT > choose Explore GPTs > Create (you need ChatGPT Plus) Well defined instructions are key, think prompting.

Slide 77

Slide 77 text

78 peakace.agency #2 Create an ‘Action’ to call a 3rd party API Head to your API provider and grab your credentials. In our case this was the API Dashboard at DataForSEO.com: Get the OpenAPI Schema for DataForSEO: https://pa.ag/3Pa7oZ3 To use with an action, you need to generate a base64- encoded version of your login credentials: btoa(‘APIemail:APIpass’) The annoying part: you need a Schema according to the OpenAPI spec. But no one reads docs anymore – we just leverage ChatGPT to do this:

Slide 78

Slide 78 text

Remember: APIs usually aren‘t free, so make sure you only publish your new Custom GPT for yourself! #3 Test and publish your GPT

Slide 79

Slide 79 text

80 peakace.agency When to use a Custom GPT? Long-term context Custom GPTs are a really powerful tool to ensure instructions remain contextualised over long periods of time. Besides seamless 3rd party data integration, here are my top-3 reasons why building and using Custom GPTs can help a lot: Building workflows Custom GPTs are best suited for composing workflows aimed at people who don’t know how to properly design context with prompt sequences. Sharing instructions For sharing the exact same instructions e.g., cross-team, without having to worry about specifying them (and how) at prompt level.

Slide 80

Slide 80 text

More AI: Beyond ChatGPT (yes, really!)

Slide 81

Slide 81 text

82 peakace.agency Perplexity AI: A chatbot-powered research & conversational search engine Source: https://pa.ag/4a7a1m0

Slide 82

Slide 82 text

83 peakace.agency Claude AI by Anthropic Claude's larger context window (up to 150k words) makes it better for analysis and summaries. Their new model ‘Opus’ is even more powerful: Source: https://pa.ag/3WMmTLB

Slide 83

Slide 83 text

84 peakace.agency Over at Google: say goodbye to Bard and hello to Gemini – Google’s AI chat bot got a new name (and some new features)

Slide 84

Slide 84 text

85 peakace.agency Typingmind.com: Manage ALL your LLMs in one place The leading chat frontend UI to connect with ChatGPT, Gemini, Claude, and more through API – to consolidate your teams’ efforts, centrally manage access and much more: Source: https://pa.ag/4b9UKlN

Slide 85

Slide 85 text

86 peakace.agency Langdock: Model agnostic & GDPR compliant Langdock integrates with LLMs from all providers as well as open-source models for maximum flexibility Source: https://pa.ag/3y1q33s

Slide 86

Slide 86 text

And the future?

Slide 87

Slide 87 text

88 peakace.agency AI recognises vocal differences in the voice Medical researchers have trained an AI to recognise 14 vocal differences in the voice of someone with type 2 diabetes compared to someone without diabetes: Source: https://pa.ag/4bW0mAR The auditory features that the AI focused on included slight changes in pitch and intensity, which human ears cannot distinguish.

Slide 88

Slide 88 text

89 peakace.agency Understanding basic biology (better) with AlphaFold "By unlocking the modelling of protein and ligand structures together with nucleic acids and those containing post- translational modifications, our model provides a more rapid and accurate tool for examining fundamental biology.” Sources: https://pa.ag/4aiQVJW & https://pa.ag/40IuR87 The healthcare system will improve (and it desperately needs to) ▪ Agents will also help healthcare professionals make more informed decisions and be more productive. ▪ Agents will help patients to triage more easily, get advice on how to deal with health problems and decide whether they should seek treatment.

Slide 89

Slide 89 text

90 peakace.agency Humans and robots communicating via EEG interface "NOIR, a general-purpose, intelligent brain-robot interface system that enables humans to command robots to perform everyday activities through brain signals.” Source: https://pa.ag/3SIXi41 Our system demonstrates success in an expansive array of 20 challenging, everyday household activities […] The effectiveness is improved by its synergistic integration of robot learning algorithms, allowing for NOIR to adapt to individual users and predict their intentions […] The robot does the cooking, the necessary instructions come from just thinking about it.

Slide 90

Slide 90 text

91 peakace.agency Drastic improvements to current LLMs... ...as well as significantly more efficient (5x-10x) multi-modality implementations. Sources: https://pa.ag/3st19aN, https://pa.ag/3uggLid, https://pa.ag/3QP6MZ4 & https://pa.ag/3SEjIDu

Slide 91

Slide 91 text

92 peakace.agency AI is becoming more efficient? That may be true - but training efforts and costs will continue to explode from about $100 million today to as much as $10 billion by 2025 or 2026 (according to Dario Amodei, CEO of Anthropic) Source: https://pa.ag/3UMhL8t In two to five years, AI models will become so powerful due to so-called scaling laws that they will no longer need or be able to be controlled by humans, according to Amodei.

Slide 92

Slide 92 text

93 peakace.agency LLM Agents can autonomously hack websites Research indicates strong scaling laws and significantly lower costs compared to current methods. Source: https://pa.ag/3V1SUOw We show that LLM agents can auto- nomously hack websites, performing tasks as […] SQL injections without human feedback. Importantly, the agent does not need to know the vulnerability beforehand. […] we show that GPT-4 is capable of such hacks [and] is capable of autono- mously finding vulnerabilities in websites in the wild.

Slide 93

Slide 93 text

94 peakace.agency Let‘s not even get started on misinformation (at-scale): Sources: https://pa.ag/47Cy7nM & https://pa.ag/3MLCLrC

Slide 94

Slide 94 text

95 peakace.agency Have you seen Microsoft's VASA-1 yet? The model takes a single portrait photo and an audio clip to produce a hyper-realistic video; so essentially the model can turn 1 photo and 1 piece of audio into a fully lifelike human deepfake – beyond scary! Source: https://pa.ag/3UO3XL1 We have no plans to release an online demo, API, product, additional implementation details, or any related offerings until we are certain that the technology will be used responsibly and in accordance with proper regulations

Slide 95

Slide 95 text

…and so much more!

Slide 96

Slide 96 text

97 peakace.agency What should you be prepared for? New players - in all areas (including chips, infrastructure, models, etc.), e.g. Leo AI, Samsung Gauss, etc. Significant progress in the areas of architecture & foundational models (LLMs etc.) Large-scale disinformation campaigns (in which it turns out in retrospect that AI played an important role) Greater impact on jobs (yep, prompting is fast becoming old news) AI will be significantly more interactive in the future (target setting is sufficient) and more natural to use (multimodality, etc.) Model development with a strong focus on human behaviour (see Microsoft LeMa)

Slide 97

Slide 97 text

Thank you! Ciao, Pärnu. =