I’m building Sift, a drop-in gateway that makes LLM tool use far more reliable when tools return large JSON payloads. The usual pattern is that agents paste raw tool outputs directly into the prompt, which quickly blows up context, causes truncation/compaction, and leads to incorrect answers once earlier results disappear. Sift sits between the model and its tools (MCP, APIs, CLIs), stores the full payload locally as an artifact (indexed in SQLite), and returns only a compact schema plus an artifact_id. When the model needs something from the data, it runs a tiny Python query against the stored artifact instead of reasoning over thousands of tokens of JSON. In benchmarks across 103 questions on real datasets, this approach cut input tokens by ~95% and improved answer accuracy from ~33% to ~99%. Repo: https://github.com/lourencomaciel/sift-gateway.
I was talking about part of this problem this morning, I keep having it with our large OpenAPI spec and the JSON parsing. But it occured to me that recently sometimes Claude will work around the context blowing problem by using a sub-agent to do the parsing. Have you seen that?
It's still a big time saver to have something like this, and stops it even taking the risk of doing it in the first place.
Happy to answer questions about the sandboxing, artifact format, or the benchmark setup.
The benchmark harness and datasets are in the repo if anyone wants to reproduce or extend the tests. Curious if others have run into the same context compaction issues with tool-heavy agents.
I fully understand that advertising enables many businesses.
But there have been businesses based on lead pipes for drinking water plumbing, asbestos for residential insulation, and so on. You could make an argument that these technologies enabled many businesses as well. That doesn’t mean we should allow lead pipes for drinking water or to use asbestos in residential homes.
Online advertising has leveled the playing field, allowing smaller brands to compete with big names. Platforms like Google make it easy to capture attention, which is why even giants like Nike are losing market share to newer players. This shift spans all non-regulated industries. Without online ads, launching a nationwide brand would require enormous budgets, leaving us stuck with the same old monopolies.
If you're right about that being the dominant effect we should see small businesses increase as a portion of GDP as online ads become more prevalent, but as best I can tell we aren't seeing that at all. For example this[0] chart from the US Chamber of Commerce shows their share of the economy actually shrinking significantly.
An alternative effect could be that online ads are an avenue for better resourced established companies to out compete and stifle upstarts. Startups are always pressed for resources and running an effective online ad campaign can take significant resources.
You're surely right that some small businesses have benefited from the online ad market, but I suspect that on average larger companies have benefited to a greater degree.
> if you're right about that being the dominant effect we should see small businesses increase as a portion of GDP as online ads become more prevalent
Ceteris paribus. Running a small business in most states involves more rules today than it did in 2000. (Common denominator: the cost of financial transactions due to post-9/11 anti-money laundering rules.)
How has it leveled the playing field? It's now become an arms race of bidding for the top ad spot, even for your own brand name. The big players can out spend the little guy and even be top ranking on searches for them.
Large companies spend quite a bit of money on online advertising, and also on research on that. They test their materials, they have data teams for comparing campaign results. And they can hijack other brand names if they pay enough. I wouldn't place my washing machine on your playing field.
After rereading a few times, I think I've parsed your argument:
That X enables many businesses to exist doesn't mean that X is a good thing on balance, because X itself can have harmful effects that outweigh the benefits.
This is of course true, and I can look past the inelegant phrasing.
But to make this a credible argument you need to argue for why the costs of advertising outweigh the benefits.
They're looking for depth where there isn't any. You're attempting to prove advertising is bad by claiming it's like lead pipes, on the basis of them both being bad. It's tautology. One could similarly "prove" ad blockers or puppies by this analogy to be whatever because they, too, can be good or bad.
i know what you're saying, but really now, you know what i mean. some scribbles on a pancard or some fish monger shouting above the crowd, it's not the same as the mass media ads that find you everywhere willingly and unwillingy, 24/7. so before 1900, why not, to make it easy. let's go back there.
I'll try to explain to the young people how bad things used to be.
Google ads was (and is) incredibly good for niche companies, since it makes it possible to advertise to people who are interested in your product instead of the general public.
So if you sell Warhammer paraphernalia, you can buy ads to be shown only to people who have searched for Warhammer related words, rather than "everyone in Wisconsin".
This lowers ad costs by many orders of magnitude, and makes a lot of businesses possible that simply couldn't exist before.
I'd want some damn good reasons to go back to the old ways!
> if you sell Warhammer paraphernalia, you can buy ads to be shown only to people who have searched for Warhammer related words, rather than "everyone in Wisconsin".
Tracking people and shoving your wares in their faces is not the only way to reach interested parties. You could go to a Warhammer convention, join a Warhammer forum and offer to send members samples, or just post images of your stuff in a sharing thread, whatever. Engage with people while they're searching for the thing you're offering.
Yes and you also don't need to reach all your customers directly. If you make a good product word of it will spread naturally. Ads actually inhibit that by taking over people's attention and pre-empting any interested customers from finding you by showing them your competitors first. So you end up paying for the reach that the advertisement industry took away from you in the first place.
Of course, word of mouth requires you to actually make a good product whereas with advertisment it's enough if your product looks good.
The old ways would have been to buy ads on warhammer.com, or the Warhammer magazine, sponsor the annual Warhammer convention, run a tournament, and so on. The money, in that case, may largely stay within the ecosystem, rather than going to some investment fund owning shares of google.
"Many" feels like it is doing a lot of heavy lifting here.
IME this is only true about drop shippers and similar business models. The vast majority of small businesses are, as a rule, awful at advertising. The few ads I see they are very poorly put together.
Even when they manage to get people to the business, small businesses are almost inevitably awful about maintaining their web presence, which makes it moot. Here's an example thread about such from the local reddit. Including some hostile responses from, charitably, overwhelmed small businesses about how you need to call to confirm a price https://old.reddit.com/r/Calgary/comments/1ewlsib/open_lette...)
> Even when they manage to get people to the business, small businesses are almost inevitably awful about maintaining their web presence, which makes it moot. Here's an example thread about such from the local reddit. Including some hostile responses from, charitably, overwhelmed small businesses about how you need to call to confirm a price
I’m assuming most of the places that redditor contacted to buy UPS batteries from are B2B shops that aren’t geared to selling to people off the street.
I’m assuming this because sometimes I buy replacement UPS battery strings, and I pay with a purchase order after talking to or emailing an inside sales person, not with a credit card at a register.
Places like this don’t even need to advertise, the professionals they’re selling to know where to find what they need.
In general advertising is low ROI, the tradeoff being that it's "easy."
I think a lot of these businesses could succeed using alternative promotional strategies. Some of them might suffer because the owners have more money than time and advertising is a good tradeoff in that case, but overall good products are still going to do well.
If you rely on advertising you should not be outsourcing it to google anyway. You should have more control over that part of your business which means it needs to be at least partially in house. Even if you do take some google ads, make sure you have other partners and make sure that ads meet your standards (not a scam, not for your competitors).
You don't rely on Google to handle everything, it's self-service. Whether you use Google or another platform depends entirely on where your audience is. For example, my father runs a small construction company, and 90% of his leads come from Google search. That’s where people are looking for services like his.
Emphasis on _advertising in its current form_, I think it's a valuable means to be able to a). monetize something and b). to spread awareness. But I agree with GP that as a society we're allowing companies to grossly over-engineer our lives around ads.
> Warfare is what happens when negotiations and treaties fail.
In the case of Israel, they are intentionally failing the negotiation part as they commit an active genocide. Their motto has always been "do 10x to your enemy as they've done to you" and at some point it completely fails.
If you're going to clutch pearls about literal war you're missing the point. Anything goes in war... that's the point. The concept of "war crimes" are, again, just part of game-theoretic calculations that agents make when they are deciding what lengths they are willing to go to in order to achieve their goals.
People think war has rules because most wars on television are completely lopsided affairs.
Arguable genocides in war are extremely common in war, and are probably the rule, not the exception. In recent years we've effectively seen an equivalent in the Sri Lankan Civil War, DRC, Darfur, Burma, I could go on. War is fucking terrible, and people should take every reasonable avenue to avoid them.
It was ironic that I lived in Brazil but because the entire office there spoke Portuguese they always sent me to the Spanish speaking countries. At that point I liked to travel and didn't mind staying in fancy hotels in Argentina and Chile. So all those conversations were in Spanish.
It isn't as funny to refer to ColdFusion as confusāo. I never heard a Brazilian make that joke.
It saves the payloads into SQLite, maps them, and exposes tools for the model to run python against them. Works very well.