The startup search company Neeva promised users an ad-free search experience. It delivered, but then circumstances changed. A tale from the AI shakeout.
Neeva seems to be yet another case where consumer behavior is almost directly at odds with supposed preferences: they always say they value privacy and are willing to pay, but there's a long list of startups who - even with more creative user acquisition - ended up proving that behavior doesn't match their survey responses. With each case, it becomes harder for others in the future to convince investors to make that bet.
What a blast to reread "Living With A Computer." I will never forget the single most exciting thing about first using a word processor (in 1985): footnotes. The program automatically fit notes on the pages, and when one added or deleted a footnote, all the subsequent notes were automatically renumbered!!
Thank you. Yes, it is almost impossible to convey what a revolutionary breakthrough these changes were.
Closest comparison I can think of: When Jimmy Carter was on the campaign trail in 1976, he got cheers from audiences all across the South when he talked about the way FDR's Rural Electrification Administration, in the 1930s, had for the first time brought electric power to small communities like his in Plains. Everyone could remember the before-and-after difference that **lighting at night** had meant.
I too tried Neeva, on your recommendation, but for the way I use search terms and the things I look for, I didn't see a dramatic difference. Your example of "T-shirts" isn't a thing I'd search for.
The most common way I search is via Google Maps (e.g. restaurants, locations, addresses, routes) or through Google Patents (for my day job), and Neeva had nothing like either of these.
In the rare searches where "sponsored" links show up in Google, they are (or seem to be) marked, so scrolling down isn't that big a chore.
So, no compelling Neeva use case for me - I signed on but haven't touched it in months and months. Sort of sad they couldn't make a go of it, but I'm not surprised.
I haven't looked at coverage elsewhere, but would imagine the leap in AI progress altered venture investors' appetites for betting against the incumbents.
Thanks. Understand your point. People's habits and use-cases differ.
For me, there was a huge difference in ad burden when I changed from Google to Neeva. I'm seeing it once more, as I switch back again. (Neeva appears to have disabled its Chrome extension.) I do realize that searches for *products* — "T-shirts" are practically begging for Google ads, whereas a lot of other searches can't be monetized in the same way. (Talked about this in the original piece.) But I kept running into more ads than I wanted, on Google.
As real-world evidence seems to indicate, your use-case may be more common than mine. Thanks for having giving Neeva a try, and for reporting your results now.
Neeva was terrific. If I had known they needed recruitment, I would have praised and linked to it on Twitter. In any case, I now do most of my initial searches on Bing Chat Creative. (Which, I see, is acquiring ads, so far under the main response.) Thanks for the update, Jim.
Very sad indeed. I tried to help them a bit with infosec because I believed so much in their vision, and assess they were doing very well in that area compared to typical startups.
Why do you say this was a consequence of "AI shakeout"? I very much look forward to the future essay after your further conversation with the founders.
Thanks — I was doing a shorthand on the AI front. I mentioned it because I know that Neeva was going all-out early this year to incorporate AI in its results — and because the Neeva statement refers to that as one oif the changed-conditions that made it unrealistic for them to go on. But I'll ask Sridhar R more about that when I talk with him.
I concur with your disappointment and so many of the comments offered thus far. I am among those who tried Neeva initially after your suggestion, didn't really get hooked enough to become a paying user, but then tried again when "Neeva Gist" became available as their AI/LLM-enhanced Android app.
That said, I was a bit surprised that neither you nor your other readers referenced the elephant in the room: Twitter. The alternatives - from Notes here on Substack to the various flavors of Mastodon to Post.news and Spoutible and others that I have not tried yet all seem more attractive in the first degree and yet... and yet Twitter still seems to rule the arena independently of the twists and turns in the new proprietor's machinations.
I do think "Notes" is different because of the preexisting foundation offered by contributors like yourself, Jim... and it also is much more attractive to me than Medium even though I find interesting things on the latter platform.
I wish I had a bit of inspiration to share as to how to take things to the next level so that Twitter will become less of a primary point of referencel for open-ended information and perspective sharing... alas, I still think it's a tough call (and with the Bruins eliminated brutally and the Celtics down 3-0 after two close games lost in Boston and a horrid performance in Miami, I'm not in a mood to look for miracles at the moment...)
Ed, good points. I don't have any sufficient response either on the Twitter front nor on the full possibilities of Notes, which I'm still getting a feel for.
Sorry for the travails of the Hub-area sporting dynasties. DC-area sports fans have plenty of woe, although the (apparent) end of the Snyder regime offers potential hope.
This saddens me. I, too, signed up for Neeva on your recommendation, and changed all of my defaults. I mostly preferred it to the Google results, though there were times, not often, that Google would find something Neeva did not. But it was wonderful being out from under the ads. I wonder how many people would buy in if Google offered a subscription model at a reasonable fee to be free of the ad-driven results...
I will say that, even ignoring the ads, the results I get from web search over the years have dramatically declined in usefulness. I suspect (I'm not even close to being an expert) that it has to do with SEO manipulations. The internet overall, like the products of the increasingly behemoth titans of IT, feels less and less useful, and more and more like hands inside my wallet.
Glad you tried it; like you I'm said it is going away.
I agree with the general long-term decline in sheer utility of search. You'd think that there would be some market for an approach that used ever-improving tech to create ever-more-useful search results. But that was what the Neeva founders were thinking too...
I just, today, received my refund from Neeva. This wasn't referenced in your initial post, but I think it bears mentioning that these people, who are closing their enterprise and could easily just disappear, had the integrity to return the unused portion of their customers' subscription. In this wild and wooly internet environment, that bears mentioning with a hearty foot-stomp.
My 21 year old grandson has been giving me a tutorial on AI. [As a kid I recall communicating with two tin cans and a string.]
What I found truly frightening was Ross Andersen’s article in The Atlantic in which he described, in excruciating detail, how AI could reduce the response time to a possible nuclear attack to nanoseconds. Also, because of counter cyber developments, how such a response might be left to machines and/or some person at a nuclear base or on a nuclear sub.
As a former professor my concern about student AI essays is of a lower timbre. In the pre-AI days I could sense student plagiarism (I maintained student portfolios). I would speak to the student and then ask him/her to pronounce and define a fancy word in the essay. BINGO!
We may be due for a revival on the 1950s/60s tech-dystopia novels about nuke war, notably the truly chilling Fail Safe. (My parents wouldn't even let me see that movie when it came to the local theater when I was in high school.) I'd call it an "unforgettable" film, except that its existence seems to be largely forgotten now.
Nice trick for sussing out plagiarism. My wife worked for quite a while as an admissions officer and a well-known private university. She said you could always sense a fake essay — but, again, that was before the age of AI. I *imagine* that some similar tell-tales will emerge.
Sorry to see this. I subscribed after you mentioned it, but I ended up going back to Google after a while (sorry!) because the lag time was just too slow--sometimes several seconds. Am I a victim of our "I want it NOW NOW NOW" culture? Maaaaaybe. But there's one data point for you.
Justin, it's fine! And you may have had a clearer view toward the inevitable future than I did.
I didn't really notice much of a response-time difference with Neeva. But what I really wonder now is whether *anyone* can make a viable go of a search engine that's not groaning in ads. I mean, in principle Google surely "could" do that, with a separate-tier service. (They might even call it "blue check"! Just a thought.) And they've already broken the glass wall of presenting all Google products as "free."
It may be too much a niche for anyone to spend time and money on — that would seem to be one lesson of Neeva's experience. Still I'll keep my eyes peeled for anything in this realm.
In 1979 we were writing and re-writing operations and training manuals. My Darlene was named Lynn, a full timer, and she could make an IBM Selectric do magic tricks. What she couldn't do was produce pages in Jeppesen size and format, or afterward come up with an index by key word or adjust pages for graphics after the fact and still have time to eat, sleep or leave her desk. So we pitched the Lanier word processor advertised in a trade mag at our department budget meeting. No dice. And for thirty five years every one of us humped a leather case full of books and charts all over the world; today pilots carry laptops. How far we've come since then.
Ah yes. I remember the ritual of going through the fat books of Jepp or FAA charts every 28 days, and page-by-page replacing the ones with changes. Having all of this on the iPad is a huge step, as you know better than I.
I acquired my first personal computer in 1984: an Osborne Executive (with a whopping 5” yellow CRT and dual floppy drives). The company was already going bankrupt by then, due to its disastrous forward marketing campaign now known as the ‘Osborne Effect’).
A fellow Purdue grad student, a member of the local Hoosier Osborne Group, (‘HOG’; get it?) scored a deal to drive to Chicago and buy as many of the OE-2 Executives as he could acquire for the incredible price of $700 each. My wife & I calculated the $ and pain to be saved in being able to write and “typeset” our own PhD theses would be worth eating beans for 2 months, so we went for it... and we’ve never been without a portable computer since then.
Yes, I well remember the Osborne. I actually went on a reporting visit to their HQ just as the machine was coming onto the market. In that era, I preferred the KayPro, with its "bigger" screen. But being around at that time was like being in Detroit in the 1910s (as I think I mention in the story), and seeing all the little startup companies — without foreknowledge of which was going to survive.
Reid My first computer was a Leading Edge (Korean?). I bought it because the instruction manual was in common sense English rather than computer gibberish.
As I was writing a book, I was terrified that my floppy disc would do something untoward—so I printed out what I wrote as ‘Back up.’ One day my floppy flopped and all i had to do was retype the pages that I had run off.
I now gladly embrace my IPad—-and the cloud (wherever it may be)/
Yes, the fragility of "back up" in those days is astonishing in retrospect — as so many things about early computing are.
From the days before computing of any sort, we know the stories (apocryphal or real) of authors who left a full manuscript in a cab and never saw it again — or had it burned in a fire, etc. From the early computing days, we know about chapters or articles that evaporated from tape storage for floppy disks. Even in the earliest days of the blogging world, sometimes the blogging software would mysteriously vaporize what you had written.
The cloud era brings its complication, but at least it has minimized these fears.
May 22, 2023·edited May 22, 2023Liked by James Fallows
I think part of the challenge start ups face is to communicate what people are losing by staying with what they’ve got … with staying with “the familiar”. Saying your product is better… gives people something new matters. But saying what people are losing by staying with “what already is” matters too.
I fear we don’t realize the societal cliff we’re approaching… we can’t see it… and won’t see it until it’s too late.
I am eager to talk with Sridhar R when he's ready in a couple of days, about this and other issues. I have found him very thoughtful about underlying issues in individual behavior and civic response that are being revealed / changed / played-upon in the tech era. Will report on what I learn.
Thanks for replying! I look forward to reading your report on your conversation with him! It isn't easy for people to see the loss of something they don't already have. My mentor in systems thinking (Russell L. Ackoff) taught that corporations never measure the profit they don't gain when they fail to do something new. They never consider that a loss to their bottom line. But if you don't do something new that would create a future benefit it is definitely a loss... of something new not something old. Good luck with your conversation!
Neeva seems to be yet another case where consumer behavior is almost directly at odds with supposed preferences: they always say they value privacy and are willing to pay, but there's a long list of startups who - even with more creative user acquisition - ended up proving that behavior doesn't match their survey responses. With each case, it becomes harder for others in the future to convince investors to make that bet.
What a blast to reread "Living With A Computer." I will never forget the single most exciting thing about first using a word processor (in 1985): footnotes. The program automatically fit notes on the pages, and when one added or deleted a footnote, all the subsequent notes were automatically renumbered!!
Thank you. Yes, it is almost impossible to convey what a revolutionary breakthrough these changes were.
Closest comparison I can think of: When Jimmy Carter was on the campaign trail in 1976, he got cheers from audiences all across the South when he talked about the way FDR's Rural Electrification Administration, in the 1930s, had for the first time brought electric power to small communities like his in Plains. Everyone could remember the before-and-after difference that **lighting at night** had meant.
I too tried Neeva, on your recommendation, but for the way I use search terms and the things I look for, I didn't see a dramatic difference. Your example of "T-shirts" isn't a thing I'd search for.
The most common way I search is via Google Maps (e.g. restaurants, locations, addresses, routes) or through Google Patents (for my day job), and Neeva had nothing like either of these.
In the rare searches where "sponsored" links show up in Google, they are (or seem to be) marked, so scrolling down isn't that big a chore.
So, no compelling Neeva use case for me - I signed on but haven't touched it in months and months. Sort of sad they couldn't make a go of it, but I'm not surprised.
I haven't looked at coverage elsewhere, but would imagine the leap in AI progress altered venture investors' appetites for betting against the incumbents.
Thanks. Understand your point. People's habits and use-cases differ.
For me, there was a huge difference in ad burden when I changed from Google to Neeva. I'm seeing it once more, as I switch back again. (Neeva appears to have disabled its Chrome extension.) I do realize that searches for *products* — "T-shirts" are practically begging for Google ads, whereas a lot of other searches can't be monetized in the same way. (Talked about this in the original piece.) But I kept running into more ads than I wanted, on Google.
As real-world evidence seems to indicate, your use-case may be more common than mine. Thanks for having giving Neeva a try, and for reporting your results now.
Neeva was terrific. If I had known they needed recruitment, I would have praised and linked to it on Twitter. In any case, I now do most of my initial searches on Bing Chat Creative. (Which, I see, is acquiring ads, so far under the main response.) Thanks for the update, Jim.
Stewart, thank you. I am sure the Neeva people will be glad to know that you tried and liked their product.
I will give Bing Chat Creative a try.
Very sad indeed. I tried to help them a bit with infosec because I believed so much in their vision, and assess they were doing very well in that area compared to typical startups.
Why do you say this was a consequence of "AI shakeout"? I very much look forward to the future essay after your further conversation with the founders.
Thanks — I was doing a shorthand on the AI front. I mentioned it because I know that Neeva was going all-out early this year to incorporate AI in its results — and because the Neeva statement refers to that as one oif the changed-conditions that made it unrealistic for them to go on. But I'll ask Sridhar R more about that when I talk with him.
I concur with your disappointment and so many of the comments offered thus far. I am among those who tried Neeva initially after your suggestion, didn't really get hooked enough to become a paying user, but then tried again when "Neeva Gist" became available as their AI/LLM-enhanced Android app.
That said, I was a bit surprised that neither you nor your other readers referenced the elephant in the room: Twitter. The alternatives - from Notes here on Substack to the various flavors of Mastodon to Post.news and Spoutible and others that I have not tried yet all seem more attractive in the first degree and yet... and yet Twitter still seems to rule the arena independently of the twists and turns in the new proprietor's machinations.
I do think "Notes" is different because of the preexisting foundation offered by contributors like yourself, Jim... and it also is much more attractive to me than Medium even though I find interesting things on the latter platform.
I wish I had a bit of inspiration to share as to how to take things to the next level so that Twitter will become less of a primary point of referencel for open-ended information and perspective sharing... alas, I still think it's a tough call (and with the Bruins eliminated brutally and the Celtics down 3-0 after two close games lost in Boston and a horrid performance in Miami, I'm not in a mood to look for miracles at the moment...)
Ed, good points. I don't have any sufficient response either on the Twitter front nor on the full possibilities of Notes, which I'm still getting a feel for.
Sorry for the travails of the Hub-area sporting dynasties. DC-area sports fans have plenty of woe, although the (apparent) end of the Snyder regime offers potential hope.
The news that Neeva is ceasing operations breaks my heart.
Agree
This saddens me. I, too, signed up for Neeva on your recommendation, and changed all of my defaults. I mostly preferred it to the Google results, though there were times, not often, that Google would find something Neeva did not. But it was wonderful being out from under the ads. I wonder how many people would buy in if Google offered a subscription model at a reasonable fee to be free of the ad-driven results...
I will say that, even ignoring the ads, the results I get from web search over the years have dramatically declined in usefulness. I suspect (I'm not even close to being an expert) that it has to do with SEO manipulations. The internet overall, like the products of the increasingly behemoth titans of IT, feels less and less useful, and more and more like hands inside my wallet.
Glad you tried it; like you I'm said it is going away.
I agree with the general long-term decline in sheer utility of search. You'd think that there would be some market for an approach that used ever-improving tech to create ever-more-useful search results. But that was what the Neeva founders were thinking too...
I just, today, received my refund from Neeva. This wasn't referenced in your initial post, but I think it bears mentioning that these people, who are closing their enterprise and could easily just disappear, had the integrity to return the unused portion of their customers' subscription. In this wild and wooly internet environment, that bears mentioning with a hearty foot-stomp.
My 21 year old grandson has been giving me a tutorial on AI. [As a kid I recall communicating with two tin cans and a string.]
What I found truly frightening was Ross Andersen’s article in The Atlantic in which he described, in excruciating detail, how AI could reduce the response time to a possible nuclear attack to nanoseconds. Also, because of counter cyber developments, how such a response might be left to machines and/or some person at a nuclear base or on a nuclear sub.
As a former professor my concern about student AI essays is of a lower timbre. In the pre-AI days I could sense student plagiarism (I maintained student portfolios). I would speak to the student and then ask him/her to pronounce and define a fancy word in the essay. BINGO!
Yes, re Ross's article.
We may be due for a revival on the 1950s/60s tech-dystopia novels about nuke war, notably the truly chilling Fail Safe. (My parents wouldn't even let me see that movie when it came to the local theater when I was in high school.) I'd call it an "unforgettable" film, except that its existence seems to be largely forgotten now.
Nice trick for sussing out plagiarism. My wife worked for quite a while as an admissions officer and a well-known private university. She said you could always sense a fake essay — but, again, that was before the age of AI. I *imagine* that some similar tell-tales will emerge.
Another overlooked tech-dystopia movie from 1970, "Colossus: The Forbin Project" (Wikipedia entry here: https://en.wikipedia.org/wiki/Colossus:_The_Forbin_Project).
Sorry to see this. I subscribed after you mentioned it, but I ended up going back to Google after a while (sorry!) because the lag time was just too slow--sometimes several seconds. Am I a victim of our "I want it NOW NOW NOW" culture? Maaaaaybe. But there's one data point for you.
Justin, it's fine! And you may have had a clearer view toward the inevitable future than I did.
I didn't really notice much of a response-time difference with Neeva. But what I really wonder now is whether *anyone* can make a viable go of a search engine that's not groaning in ads. I mean, in principle Google surely "could" do that, with a separate-tier service. (They might even call it "blue check"! Just a thought.) And they've already broken the glass wall of presenting all Google products as "free."
It may be too much a niche for anyone to spend time and money on — that would seem to be one lesson of Neeva's experience. Still I'll keep my eyes peeled for anything in this realm.
In 1979 we were writing and re-writing operations and training manuals. My Darlene was named Lynn, a full timer, and she could make an IBM Selectric do magic tricks. What she couldn't do was produce pages in Jeppesen size and format, or afterward come up with an index by key word or adjust pages for graphics after the fact and still have time to eat, sleep or leave her desk. So we pitched the Lanier word processor advertised in a trade mag at our department budget meeting. No dice. And for thirty five years every one of us humped a leather case full of books and charts all over the world; today pilots carry laptops. How far we've come since then.
Ah yes. I remember the ritual of going through the fat books of Jepp or FAA charts every 28 days, and page-by-page replacing the ones with changes. Having all of this on the iPad is a huge step, as you know better than I.
Thanks for the link to your 1982 article!
I acquired my first personal computer in 1984: an Osborne Executive (with a whopping 5” yellow CRT and dual floppy drives). The company was already going bankrupt by then, due to its disastrous forward marketing campaign now known as the ‘Osborne Effect’).
A fellow Purdue grad student, a member of the local Hoosier Osborne Group, (‘HOG’; get it?) scored a deal to drive to Chicago and buy as many of the OE-2 Executives as he could acquire for the incredible price of $700 each. My wife & I calculated the $ and pain to be saved in being able to write and “typeset” our own PhD theses would be worth eating beans for 2 months, so we went for it... and we’ve never been without a portable computer since then.
Yes, I well remember the Osborne. I actually went on a reporting visit to their HQ just as the machine was coming onto the market. In that era, I preferred the KayPro, with its "bigger" screen. But being around at that time was like being in Detroit in the 1910s (as I think I mention in the story), and seeing all the little startup companies — without foreknowledge of which was going to survive.
Reid My first computer was a Leading Edge (Korean?). I bought it because the instruction manual was in common sense English rather than computer gibberish.
As I was writing a book, I was terrified that my floppy disc would do something untoward—so I printed out what I wrote as ‘Back up.’ One day my floppy flopped and all i had to do was retype the pages that I had run off.
I now gladly embrace my IPad—-and the cloud (wherever it may be)/
Yes, the fragility of "back up" in those days is astonishing in retrospect — as so many things about early computing are.
From the days before computing of any sort, we know the stories (apocryphal or real) of authors who left a full manuscript in a cab and never saw it again — or had it burned in a fire, etc. From the early computing days, we know about chapters or articles that evaporated from tape storage for floppy disks. Even in the earliest days of the blogging world, sometimes the blogging software would mysteriously vaporize what you had written.
The cloud era brings its complication, but at least it has minimized these fears.
I think part of the challenge start ups face is to communicate what people are losing by staying with what they’ve got … with staying with “the familiar”. Saying your product is better… gives people something new matters. But saying what people are losing by staying with “what already is” matters too.
I fear we don’t realize the societal cliff we’re approaching… we can’t see it… and won’t see it until it’s too late.
Yes, I agree.
I am eager to talk with Sridhar R when he's ready in a couple of days, about this and other issues. I have found him very thoughtful about underlying issues in individual behavior and civic response that are being revealed / changed / played-upon in the tech era. Will report on what I learn.
Thanks for replying! I look forward to reading your report on your conversation with him! It isn't easy for people to see the loss of something they don't already have. My mentor in systems thinking (Russell L. Ackoff) taught that corporations never measure the profit they don't gain when they fail to do something new. They never consider that a loss to their bottom line. But if you don't do something new that would create a future benefit it is definitely a loss... of something new not something old. Good luck with your conversation!