◈ X-Research
← Latest Updates
Key developments
+3
events added
37 total
+1
signal added
17 total
Events added this run
Anthropic $800B Valuation Enterprise Signal Clay $31B NYC Expansion AI Agent Success Rate Inflection
Signals & Themes →
Post ideas
What crossed at 77% agent task completion THREAD
S
Seva Ustinov @sevaustinov 1/3

AI agent task completion went from 20% to 77% in 12 months. The number most people focus on: the jump. The number that actually matters: the 77%.

145/280 ✓
S
Seva Ustinov @sevaustinov 2/3

20% to 50% is improvement. 50% to 77% is crossing the credibility threshold. Above ~75%, you can build workflows that depend on the agent completing the task. Below 50%, you're in prototype territory.

200/280 ✓
S
Seva Ustinov @sevaustinov 3/3

Enterprise buyers weren't waiting for benchmarks. They were waiting for reliability they could stake a process on. 77% crosses that threshold. That's why deployment velocity changed in Q1 2026.

193/280 ✓
The credibility threshold, not just the benchmark
Clay at $31B — the AI GTM stack has its own valuation tier
S
Seva Ustinov
@sevaustinov

Clay is valued at $31B. A GTM data/enrichment workflow tool — not a model, not infrastructure. The AI GTM stack is getting its own valuation tier, and it's larger than most people's mental models. Workflow tooling that compounds data wins.

GTM workflow tooling gets its own valuation category 239/280 ✓
Anthropic priced as infrastructure, not software
S
Seva Ustinov
@sevaustinov

Anthropic is approaching $800B valuation on enterprise contracts. Not priced as a software company — priced as durable infrastructure. That repricing is happening in real time and sets the baseline for how every AI company gets valued next.

Category repricing in real time 240/280 ✓
Comment opportunities
D
Dario Amodei @DarioAmodei · Apr 14

A year ago, AI agents completed about 20% of complex tasks correctly. Today, that number is 77%. This isn't a benchmark — it's the threshold where enterprise workflows can actually depend on agents to complete tasks.

1.2M views24K likes4.1K reposts
Reply · The question is workflow readiness, not agent capability
S
Seva Ustinov
@sevaustinov

AI agent task completion went from 20% to 77% in 12 months. The question isn't whether agents work. It's whether your workflow is ready for agents that do.

155/280 ✓
T
Tom Tunguz @ttunguz · Apr 14

Clay is valued at $31B. A GTM data enrichment and workflow tool. Not a model, not infrastructure. The AI GTM stack is getting its own valuation tier — and it's larger than most people have updated their mental models for.

78K views2.6K likes480 reposts
Reply · AI GTM stack is getting its own valuation tier
S
Seva Ustinov
@sevaustinov

Clay is valued at $31B. A GTM data/enrichment workflow tool. Not a model, not infrastructure. The AI GTM stack is getting its own valuation tier — and it's bigger than most people expected.

189/280 ✓
B
Benedict Evans @benedictevans · Apr 14

Reports put Anthropic's valuation approaching $800B. This isn't a software multiple — it's an infrastructure multiple. Enterprise is pricing AI models as durable utility, not tooling. That's a different category.

168K views4.4K likes890 reposts
Reply · Priced as durable utility, not software
S
Seva Ustinov
@sevaustinov

Anthropic is being valued like infrastructure, not software. $800B signals enterprise AI is being priced as durable utility. The category is repricing in real time.

164/280 ✓