Nebius bought Inferize, a 10-month-old Israeli startup with 17 staff, for an estimated $100 million to $150 million, according to Calcalist. The money goes on software that keeps inference GPUs from sitting idle, and Nebius is still adding data center capacity alongside it.
Perspective Coverage
3 publishers
- Builder
- Builder 38%
- Operator
- Operator 32%
- Investor
- Investor 30%
Reality
- Evidence55
- Adoption12
- Hype gap+30
- Incentives70
- Confidence60
buildOne report1 publisher Callers of Google's Custom Search JSON API must pick a new web search provider before January 1, 2027
Google stops serving the Custom Search JSON API on January 1, 2027 and has already closed it to new customers. Teams using it for general web search now pick between rewriting for another provider and running a shim that copies the old contract.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+20
- Incentives65
- Confidence45
European agentic-AI startups pass last year's funding total with 7.9bn euros raised
European agentic-AI startups have raised 7.9bn euros this year, already more than last year's 7bn, according to Sifted data. The founders who explain how customers deploy agents each sell one layer of the stack and disagree about which one has to come first.
Reality
- Evidence35
- Adoption
- Not scored
- Hype gap+30
- Incentives75
- Confidence40
Keenable left stealth with $26 million and says AI labs already query its index during training and at runtime. The scarcity argument comes from the investor who led the round.
Reality
- Evidence38
- Adoption22
- Hype gap+35
- Incentives70
- Confidence55
buildOne report1 publisher Microsoft's Agent Harness ends an autonomous run when the model types RESEARCH_COMPLETE
One extension method turns an IChatClient into a bounded agent loop with tool approval and history persistence already wired. The default exit condition is an exact marker string the model has to emit.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+12
- Incentives30
- Confidence58
buildOne report1 publisher searxng-gateway starts its paid providers before it knows whether SearXNG failed
The Go proxy keeps SearXNG's JSON shape, so callers need no change. Starting the paid work before SearXNG answers costs one billed provider call on every cache miss, plus a five-minute breaker on any 4xx.
Reality
- Evidence48
- Adoption8
- Hype gap0
- Incentives70
- Confidence52
buildOne report1 publisher Seven buyers put $82.06m of preferred stock into Firecrawl at 5.7 times its Series A
The Form D filed on September 14th dates the first sale to August 31st, 2026. The case for that much money rests on figures Firecrawl reports itself: 180,300 GitHub stars and an ARR claim on its Y Combinator profile.
Reality
- Evidence64
- Adoption58
- Hype gap+15
- Incentives68
- Confidence52
buildOne report1 publisher Retries inflated one agent turn to 47,004 input tokens for a 280-token answer
A hosted-skills agent that let the model call four tools in whatever order it liked pushed a single webshop question past the context window of a frontier model. The fix was to gather the evidence in code first.
Reality
- Evidence58
- Adoption20
- Hype gap+12
- Incentives35
- Confidence55
buildOne report1 publisher Rewriting four MCP adapters in Go cut 23 seconds of cold start to two
The Python adapters worked; they just paid interpreter startup and an import walk on every agent tool call, on a cluster that scales to zero because one person funds it. The Go control was already an order of magnitude quicker.
Reality
- Evidence45
- Adoption20
- Hype gap+20
- Incentives25
- Confidence55
buildOne report1 publisher Keenable raised $26M to sell what Microsoft stopped giving away
The volume discount on its search API starts at 100 requests per second, which works out to roughly $3.2m a year. That tells you who the index is really for.
Reality
- Evidence34
- Adoption24
- Hype gap+26
- Incentives62
- Confidence41
buildOne report1 publisher Exa hires a CTO for the part of search that has a power bill
Jeff Pinner arrives from Robinhood to run Exa's own index, GPU cluster and 323ms latency budget. Exa's stated targets imply roughly 97,000 requests in flight at once.
Reality
- Evidence42
- Adoption48
- Hype gap+30
- Incentives68
- Confidence45
The criminal AI market is a reseller business, and Grok's abuse desk is the chokepoint
ThreatDown says Kriminal, one of the newest crimeware AI tools, is a storefront and a jailbreak prompt on rented models, sold on the open web from $12.99 a month.
Reality
- Evidence58
- Adoption34
- Hype gap+12
- Incentives68
- Confidence52