My opinion on the new API company...
TL;DR: There is nothing wrong with this, it just doesn't solve the real problem.
TL;DR: WHAT DO I MEAN? The industry has a relevance problem. Building more industry coalitions does not resolve this. The reason for network APIs failing is the same reason 5G has failed and the fear 6G follows in the same footsteps. We industrialize our wanted answers before validating any market interest in what we are doing. We scale then ask why...
Stop here if that is all you need. Otherwise read on...
Introduction
The challenge we have as an industry is we enjoy doing what we already know how to do.
Don't get me wrong I am excited by all new industry initiatives especially those who attempt to make something work that hasn't before. But we need to learn from the past, create materially different outcomes and generate growth not cost. This time it has to be different.
WHAT IS GOOD: This delivers better coverage coordination. Developers want total population coverage for the segment they are targeting. For many applications this is fine as long as they have as close to 100% in each country. Many applications are country specific.
In the end appendix, you can find an "announcement country coverage estimate" (subscribers not technology), estimating the impact of excluded operators. Full disclosure, I have used ChatGTP 4.o1-preview to generate this table. I appreciate as many local people checking and saying if the estimations are about correct. This is a good crowdsourcing test of the new 4.o1-preview accuracy.
Ok.. so now what?
WHAT IS STILL NEEDED: Everything else.
APIs can be cheap but...
APIs you don't use are a cost
Standardized APIs you don't use are a massive cost!
I am going to walk through what is more important than creating this company.
Find real need - where to look?
Having universal access to irrelevance is not valuable. We have to start with honesty.
We start with technology and what we believe it theoretically can do. The challenge is the technology is not fully deployed and not fully operational. It never is. Unless an application developer is building a bespoke app for a unique network deployment scenario (private network) network coverage cannot be guaranteed. Historically there have always been three G's deployed in overlapping lifecycle. Indoor 80% of the traffic goes over Wi-Fi. And in rural areas the coverage is often missing, with the likelihood of satellite and High Altitude Platforms (HAPs) being more cost effective solutions.
As an industry we are too excited by our own technical cleverness and we forget the composition of reality and the environment those who we target live in.
In Rakuten we view APIs in a similar way to Facebook (see appendix). APIs are the result of effective solution building, not the start. We find areas where we believe there are growth opportunities and we decide if we treat
The pure API business happens as a result of democratizing the value that has already been proven. Facebook did not start with API's, it started with a social graph, then it gave access to the graph once it realized the value.
We are in the business of building growth companies, not justifying technology rollout.
What are two examples of industry opportunity?
Where is the industry unique and what big problems are growing?
Focus on real time identity and fraud
The most interesting part of our business is the exhaust not the primary deliverable. We deliver connectivity but the side effect is we have validated identities and behavioural awareness of massive percentages of the population. We have a better real time idea of who is in a country at any given time than any government or other official authority. This is invaluable. We need to open our minds to beyond the traditional definition of fraud and to the even bigger problems of AI generated fake content, fake identity, and the destruction of future trust. If we as a society can no longer trust someone or something is real, even when hearing and seeing the person then we have massive problems. This is currently more realistic than less. Telecom is unique in its continuous situational awareness.
Is this a problem we can solve for in a way no other industry can solve?
I argue yes. I argue that is the kind of solution we should be tackling. Tomorrow's unsolved problems not today's already solved ones. We have a start with fraud traction. Where we have collaboration that represents large enough populations then banks are already integrating and proving value, and creating further opportunities for more behavioral signals. Telecom is unique since it is very hard to fake subconscious behavior with device spoofing or similar.
For all people saying this is not something we can do, I ask why?
The World’s leading End to End Unmanned AI based Space Solution
In many aging societies there will be a large shortage of workers. We even see this as a problem in traditional management of telecom networks where unless we automate operations, irrespective of savings, we will not a workforce in 10 years time that is large enough to manage networks the way we do today.
This is universally true in many industries and has led to the launch of Rakuten Neo.
The service is already launched and is rapidly growing its customer base. We see a
This business could not exist without network APIs!
That is completely untrue. We as an industry would like that to be true but today this business uses ZERO network APIs and we are really happy. If we don't have to use network APIs we don't want to, because network APIs are a cost if they are not a benefit!
The question we should ask ourselves as an industry is what should be Rakuten Neo be using in the future and WHY?
It is mission critical, requires real time responses and needs high levels of security, privacy and trust. What should it be using in the future? For example can we remove on-premise IT and make everything an off premise managed service? What else?
The nice thing about this business example is it puts the API opportunity into perspective. APIs are as powerful as the business they enable - what is the value network APIs create in this scenario and how does that deliver 4x value to cost.
For more information on Rakuten Neo, please reach out directly to Ryan (Seung-Hyun) Son , the brains behind the whole operation and business model.
Ensure coverage meets needs
To come back to what the announcement does help to address. For any solution offered it has to be possible to deliver, according to the requirements of the customer not the industry.
That means the solution needs to work where there customers are. If they need it on Wi-Fi it needs to work on Wi-Fi, if they need it in the countryside, it needs to work in the countryside. And it has to be independent of any operator subscription dependancy. If the customer changes operators it still has to work.
The market has zero interest in the number of operators taking part, they assume all do.
If the business is a growth business they all will want to be a part. We then get the same network effects that led to mobile voice ubiquity originally.
Understanding next best alternatives - be honest!
Being able to do something is challenge one, being able to do it better than the next best alternative is harder. There are always next best alternatives. Often they are not what you ever consider. Doing nothing is quite often the biggest next best alternative the world chooses. This is when you think the value you create is much bigger than the value your target audience thinks you create.
Next best alternatives are not a technical specification competition. They are a real world beauty contest. Good enough, cheap enough, easy enough, ALWAYS WINs.
领英推荐
Telecom has to be very careful. Good enough, cheap enough, easy enough is where the innovators dilemma is born. It is where paradigms shift. Telecom traditionally architects systems to deny failure, cloud architects systems to accept failure as inevitable, but designs to remove fatality from occurrence. in 2008, I designed a global multimedia service that had a memory leak on the web server side. Over a number of days the responses would become slower and slower until responses were timing out. We could not afford redundancy and we could not afford to delay launch. We automated a health fetch every 5 minutes, if the response took longer than 3 seconds we restarted the backend, restart was fast because the implementation was so simple. We won the award for the highest uptime award of any service that year. Perception gave us 100% availability.
The true competition for network slicing, network quality on demand is nothing to do with the network but rather carrier bonding from above, plus any other available network like Wi-Fi. The best available market solution can be used, the collective capacity from all can be used, it is the most universally available solution.
The BBC chose this solution for broadcasting the UK general election results this year and covered the announcements live from each seat for the first time ever.
To learn more see here.
If the BBC is trusting this solution who else will? How big is the market for a different solution? Why is the different solution better? These are the really interesting and harder questions to answer. NOTE: no need to coordinate all operators to offer this...
Removing Developer Breakage
Hypothesis: We have designed the absolute AI criminal anti-virus API that everybody can embed in whatever app or service they have, to make sure nobody can pretend to be somebody they are not and nobody can post fake content without it being traced back to them. This keeps all the app customers safe and removes threat of compromise, lawsuits and legal bills.
Perfect: What else needs to happen to make this a success?
There are four decision criteria that potentially can cause adoption breakage:
Note #4 is #4 not number 1. Until I have moved through the first 3x steps I do not get to the fourth. I admit this is an oversimplification and generalizing all developers into the same bucket of adoption is a slight oversimplification. But not as much as we think.
Developers find solutions to their problems by googling. Make sure any solution is findable.
Developers want to try things, it is faster than speaking to somebody. We never want to speak to people, until we are successful and want a massive discount you cannot say no to.
How much will it cost me? Is the cost larger than the problem? Is the implementation cost larger than the problem? Think to the BBC example above. Simple, practical, worked.
Finally it has to work, the definition of it working is the customers to define. And it combines the cost of all the previous answers. APIs are a SaaS business.
Removing Delivery Cost
Everything costs something. People selling API platforms are doing so to make money from selling them, since it costs to make them. Every API costs money to deliver, the cost of each API is relative to the approach taken to discover the API value.
By coincidence I just saw Dean Bubley post on the same announcement I am commenting on.
These statements struck me from Dean
"Basically it’s an “intermediate aggregator” platform for standardised public network APIs, that feeds up to other platforms and integrators that pitch APIs / capabilities to enterprises or other classes of developer"
Vonage, Google and other cPaaS / hyperscale providers will be channels for the network APIs. This makes sense too, although there are risks from "aggregator stacking", especially in terms of operations, security etc.
It’s not intended to be profit-making (maybe reinvestment funds, but not large dividends), so API pricing should be reasonable. It's essentially a mission to increase relevance, not (direct) revenues
This is not communism. Your production costs have to be more effective than the next best alternative and the value has to be multiples higher than the cost. I have not heard of the term "aggregator stacking" before.
Beware the problem of "Tom Smykowski"...
In Summary
I did not intend this article to be this long. I would like to thank my sponsor Mosaic beer for helping me close it out in the Sunday evening.... Unlike network apis it only takes one Mosaic beer...
On one hand APIs are really easy because they are how everything works. On the other hand network APIs are really hard because they represent massive opportunity. The key learning I would like everybody to take away.
Network APIs are as valuable as the promise they represent
Because we have built an expensive network does not mean there is value directly associated to that expense. There is value to the people building and running those networks but that is very different. And there might be value you have not even considered, that has appeared as an exhaust fume from the core mission (validated identities).
I urge the industry to be brave and discover through as much direct and immediate market engagement as possible.
Mosaic out - drops the mike...
Appendix 1: Facebook - what can we learn?
When Zuckerberg started Facebook he didn't collect all the USA colleges together so they could find something valuable to do together. He recognized that college students want to be connected with other college students and he made it as easy as possible for them to do that.
He did at one college then when he knew what people wanted, he made it as easy as possible for other colleges to do the same. And then he made it as easy as possible for people in general to do the same.
To summarize:
In 2007 they launched their API platform. I was one of the first users when we embedded Facebook into our Volvo Ocean Race Mobile Experience.
We used Facebook APIs for two unique reasons:
It met our other criteria:
APIs came after value.
Appendix 2: Announcement country coverage estimations
This table was generated by ChatGTP OpenAI o1-preview logical thinking model.
Please highlight any erroneous estimations in the comments
VP Sales Tr3Dent
2 个月Geoff... Great article and perspectives that surely took more than a weekend to compile!! Well done. I always go back to the Great One.. Steve Jobs. He said you must always start with the customer experience and work back to the technology. These API coalitions of the willing are great, but what Enterprise problem are they truly solving? Cart before the horse me thinks. Let's connect soon. Lot's of interesting action at Tr3Dent.
Real problem is that cellular operators are fragmented because spectrum allocation is tied to geography, by laws of physics, and distributed across organizations in each geography according to some policies (adapted by governments to each geography). It is incorrect to compare this with "no-borders" internet companies, as this is completely different environment. I have some ideas how to make telecom API usable, but I think these ideas will be hard to achieve due to reasons above, so we better focus on something else.
Thanks, Geoff. Extremely insightful
Chief Business Officer | LinkedIn Top Voice 2024 | Telecom Media Hi-Tech | #VPspeak
2 个月Excellent commentary Geoff! Some great insights and learnings for everyone. I remember you and I discussing some of these at one of the MWC’s.