I go down the AI rabbit hole each week so you don’t have to. Follow me on Linked In for more on this subject

A word from our sponsor!

Our summary last week of the big Google I/O developer conference and in particular the analysis from Search Engine Land confirmed what Propellic founder Brennen Bliss had told us was coming when we recorded our podcast (soon to be released) episode with him.

With Google potentially integrating AI-driven answers directly into search results, Brennen predicts significant changes in how SEO is approached. He states, "If it's a top of funnel query, you're going to get an AI-generated answer." This shift would challenge traditional SEO strategies focused on driving traffic through organic search results, as AI could directly provide answers to users, thereby bypassing some traditional search results.

He explains, "Google is going to bring most likely with their SGE product, the AI-driven answer directly into the search user journey." This indicates a future where AI not only enhances but also partly controls the flow of information in search queries.

Are you prepared for this change?

Check Propellic out for yourself here and grab yourself a call with Brennen.

New fund launched by travel ai veteran to supercharge early stage travel tech

Great news this week as Frank Reeves, founder of ai powered hotel tech (from well before the GenAI boom) Avvio which exited to SHR in 2022, announced he is launching a fund to specifically cater exclusively for early stage travel tech startups.

The fund is called Voyria and Frank explains “our sector expertise in travel combined with a truly differentiated approach to collaborative partnerships will cultivate ecosystems that empower growth and innovation within Travel Tech.”

The funding of course is always good to be able to push ahead with a roadmap, but I suspect the been there, done that advice of someone who has walked the path is where the real value lies here.

Bizarre happenings post the “big reveal” week

The soap opera cleverly disguised as one of the world’s leading tech innovators, OpenAI, didn’t fail to disappoint the week after their “Spring Release” announcements. Firstly all the (remaining) big players on the ethics and risk side of the business left. In this post bluntly titled “Open AI no longer takes risk seriously” Lawfare Media talks about the departures of Ilya Sutskever, OpenAI’s co-founder and chief scientist, and Jan Leike, a pioneering AI researcher “resigned from the company (Leike explicitly in protest), and OpenAI disbanded Superalignment.”

What is Superalignment? According to article it was “a team dedicated to making sure its products did not destroy humanity.” Seems like a good thing to have? “While they were certainly the highest-profile departures, Sutskever and Leike are not the only OpenAI employees with safety-focused job descriptions to have left in recent months. Daniel Kokotajlo and Cullen O’Keefe, members of the company’s AI governance team, also quit recently. Like Leike, Kokotajlo has publicly stated that he has “lost trust in OpenAI leadership and their ability to responsibly handle AGI.” But I’m sure everything will be OK.

That wasn’t even the biggest news! What really caught the headlines was the allegation from Scarlett Johansson who “Says OpenAI Ripped Off Her Voice Without Consent”. TL:DR here is that Sam Altman is a fanboy of some sci-fi movie called “Her” where the Siri type character is voiced by Johansson. The head of the most innovative company in the world was unable to come up with anything innovative and asked Johansson if they could use her voice. She said NO! It seems Altman forgot to relay this info to anyone building the demo for the Spring Release. When it clicked, two days prior to the event, he reached out again. She again said NO. So they just proceeded anyway.

Meanwhile over at Meta they’ve just ushered in an ai council completely compiled of white men to make sure nothing like this ever happens over there…..

And to round out the show you couldn’t make up if you tried, Tech Crunch also reported that “Expedia says two execs dismissed after ‘violation of company policy’. And those execs include Rathi Murthy who we reported on here last week as the star of the show for the launch of Romie - Expedia’s new chat assistant.

It is beyond amazing that Google who took a dump on all the world’s content creators has come out of the week in front. Or maybe it is Microsoft who with very little fanfare at all just gave Windows “the biggest update to Windows in its forty-year history” with the new version built completely with Co-Pilot at the centre. Watch the demo to see what this means in practicality (less than 1min).

Let’s talk about RAG

Whilst retrieval augmented generation or RAG as it is more commonly known is just a well accepted part of the ai framework for those used to building with ai tools, I have found it isn’t a concept immediately known or understood in the boardroom yet - but it absolutely must be.

Phocuswire’s Cathy Walsh and Mike Coletta went about rectifying that this week for their readers. RAG “optimizes LLM responses via reference to an external knowledge base (e.g., a travel company database). Basically it plays the critical role in stopping the LLM from hallucinating (or as some prefer to call it, lying).

This is critical because 95% of those I consult to, want their data both protected but also want to be able to leverage their hard won IP into ai usage. And ultimately it is this which will help each company stand apart from the next, depending on how well they’ve managed to capture and store that data over their existence. Remember when everyone was just saying “data data data big data”. That’s now becoming important even if most didn’t have a clue what to do with it before now.

One thing I do want to highlight (and I’m more than happy to be wrong on this) was the article did suggest “OpenAI's GPTs are the easiest way to experiment with RAG implementations, as they can augment the ChatGPT LLM with proprietary information through the simple pasting of text content or uploading of files.” That’s not advice I give. You can see for example from this community forum post on the OpenAI site it is VERY unclear as to whether this data is used to train models. Whilst there was once an opt out toggle, they removed it and then never spoke of it again……..

Very early on Sam Altman did say categorically (and is contained in their documentation) is that they will not train on data used via the API. “These OpenAI Business Terms govern use of our APIs, ChatGPT Enterprise, ChatGPT Team, and other services for businesses and developers.

3.2 Our Obligations for Customer Content. We will process and store Customer Content in accordance with our Enterprise privacy commitments. We will only use Customer Content as necessary to provide you with the Services, comply with applicable law, and enforce OpenAI Policies. We will not use Customer Content to develop or improve the Services.”

I only recommend this type of experimentation via the API.

If you think someone (or everyone) you know or work with could grow from being more informed on the topic of ai + travel (or could use the training above) then please forward this email to them and they can click the button below:

Speaking of needing a job…….

Lonely Planet announced this week it has opening for an Engineering Manager who has “Familiarity with data engineering, data science, machine learning, and/or developing LLM (Language Model) products.” Some sparkling candidates available this week!

ai content = SLOP!

A great piece in The Guardian this week about the tsunami of ai created content spilling onto the web. “Slop is what you get when you shove artificial intelligence-generated material up on the web for anyone to view….it functions mostly to create the appearance of human-made content, benefit from advertising revenue and steer search engine attention towards other sites. Just like spam, almost no one wants to view slop, but the economics of the internet lead to its creation anyway.”

Mattew Barker who was the one who alerted me to the article summed it up well “Wait until the slop starts training on the slop. It's the internet's Mad Cow Disease moment.”

So is all SLOP created equal? We’ve seen Brennen Bliss from Propellic recommend “There's no reason you can't use AI, but make sure you give it the unique information about. Your property about your experience, about your airline, whatever industry you are in travel with your cruise, your, your media company, you need to, you need to have something. Above and beyond what's available out there.” And I think that is about right. What looks like slop from a distance but could actually be a delicious cassoulet.

Like most things ai - I suggest you forget about the tool and focus on the problem or opportunity. Define your content strategy. Know what it is you DO want to say and how you want to represent yourself. Talk to your personas, understand what they do and don’t resonate with (unlikely to be SLOP). Add value in some way. If ai can help with any of that - and for sure it can, I know many people and brands interacting with their personas which they load into LLM’s to talk to to help refine messages and messaging, use it for THAT.

How loveholidays build ai product

Fascinating video/podcast this week if you want to climb inside the mind of a couple of blokes actually building with ai. VUX World interviewed devs Eugene and Tom on how they built the virtual assistant “Sandy”. We don’t normally hear from the builders, usually it is the marketers, the PR or the CEO, so this is a refreshing and pragmatic conversation.

It is all about customer service at scale. It’s a little long but definitely a good one for your next walk.

Thanks to Alex Bainbridge for the share in the Slack group.

Road trips with the kids - solved!

Sam Keller who owns and runs Working without Borders and previously hosted our early education sessions for the community here, has been back in the garage, this time with his son Kaz, building the ultimate conversation starter for well, anyone.

Their starting point for Cazzy was naturally enough themselves as father and son. Kaz “pointed out that if we expected him and his sister to prefer to engage in conversation rather than consume highly compelling digital content, we needed to up our game in terms of thinking of things to talk about.”

Cazzy generates conversation kickstarters tailored to any family’s interests. I wonder what this might spark in the minds of car and campervan rental companies? How could it be tuned in to a destination so the conversations follow the journey? Lots of tangential opportunities here for those who want to attract families. The big lesson here is ai just giving you the answer, is often not the answer………

Want to have a conversation that both you and the kids will enjoy?

You can find Cazzy at www.cazzy.ai

Can ai predict turbulence?

Shocking news this week of a turbulence event causing a fatality alongside lots of injured passengers on a flight from London to Singapore. I wondered aloud if ai could predict turbulence - as it seemed like something that might be in the ai wheelhouse. I then asked Perplexity.

The short answer is “Yes, AI can predict turbulence ahead of passenger flights, providing valuable lead time for airlines to take proactive measures and enhance safety. AI systems leverage advanced weather models, machine learning algorithms, and vast data sources like aircraft sensor readings, satellite imagery, and meteorological data to forecast turbulence with increasing accuracy.”

The references are here, here & here.

Or is it? The first of these is a directory of tools and a no doubt ai written article (albeit on a very very very long tail search term), the second is a Linked In post by someone saying it should be possible in the future. The final piece of evidence is from a tool that is built for weather prediction and in service with South West Airlines. The tool seems to mainly be used to predict when and where cancellations are most likely because of weather such as snow. It doesn’t categorically say using this tool will absolutely avoid all turbulence events.

If we are moving to a world of one sentence paragraphs from which we distill our understanding of topics we want knowledge on, I expect we are not going to get the right knowledge or the full picture. Continue to dig in!

How to work with Tony

The calendar is now very full. I am unfortunately no longer able to do the board room workshops for the time being unless they are in Melbourne. I am however still open for speaking at events.

MARKETPLACE: The Everything ai in Travel brand is currently being given a full makeover and will soon launch as a marketplace for ai powered tools available B2B for the travel industry. We’ve had about a dozen companies already pre-commit to their spot. I can’t thank them enough for the faith they are showing but I’m going to try - with grandfathered forever lowest listing pricing. If you also want to get in on that - shoot me a message.

Consultancy:

Please email me to deep dive into the specific opportunities within your business you might wish to explore. By interviewing key internal stakeholders we can identify which of your bottlenecks are most ripe for an ai powered fix and the approach to take to fix those across a month long project. The earliest I am available for this is now July with very limited availability.

For the fully committed business who now understands the transformative power of this technology, the final phase is to move to build your own internal “AI centre of excellence” which is combination of building an ai culture in your business by taking a human centric approach as well as building out or buying in the best solution to each identified issue. Please email me for more details on any or all of these phases.

Want to follow in Propellic’s footsteps and get in front of a highly engaged audience of travel decision makers by sponsoring the newsletter? We are booking Q3 & Q4 sponsorships now. Also email me on that one for rates and details.

Always happy to chat to anyone looking to engage either of the two travel related startups in which I’m involved

• HandbookFM.com for those looking to up their training and onboarding game such as DMC’s who want to show prospective customers how they will train their local teams on the customer brand values and safety criteria

• Customised Trip which is an ai that mimics the human travel agent to build out a bespoke itinerary for a client before the human sales team gets involved.

  • Post building the itinerary and before bringing the human into the loop, we are able to offer the (tyre kicker) customer the chance to leave with their itinerary if they are never going to book. This massively improves the conversion rates on those who do reach the human sales agent allowing your people to focus on highest intent guests.

  • It can also come also with an full trip fulfillment option so the whole process from conversation to travel experience is taken care of or it can hand off to your team for pricing and fulfillment.

Most clicked last week was the link to the custom GPT I built for businesses to build their own strategy around ai. Actually it was a new most clicked ever! Two weeks in a row. Strangely I received zero feedback from anyone on it. So how was it? Let me know as I’m happy to continue to improve it or if it worked well, do a post about it yourself and tag me in. That’s it - you’ve made it to the end of this edition. I’ll be putting the result of the most clicked post in next week’s edition so you can see where others are focussing. If I’ve missed something, you’ve got a tip or any feedback at all - you can simply reply to this email and it will come straight to me. I’m doing this for You so please don’t be shy to tell me what you think