TickerTain
$SPCX····$NVDA····$MSFT····$GOOGL····$ANTHROPIC·$TSLA····$CRWV····$NBIS····$HIMS····$TSM····$ASML····$FIGURE·$LYFT····$UBER····$SNOW····$META····$ELEVENLABS·$CURSOR·$COGNITION·$OPENAI·$AAPL····$RKLB····$XAI·$AMZN····$AMD····$000660.KS····$INTC····
←

gabe pereyra

T2 · manager / operator

Co-founder and president of Harvey, an AI platform for legal and professional services. Previously a research scientist at DeepMind and a machine-learning engineer at Meta.

6 calls·3 names·67% bull·last heard 2 months ago·Sourcery VC
track record

no scored calls yet — needs a stated position or a categorical verdict, with a matured window vs SPY

top calls

highest conviction · one per company
1sthigh conviction
$HARVEYHarveyposition

Harvey reaches roughly $300M ARR as product velocity compounds

Harvey is about four years old, near 900-960 employees, around 2,000 customers and roughly $300M ARR. The key signal is that product build cycles are compressing while the company expands into a broader legal AI platform.

Sourcery VC2026-06→
2ndhigh conviction
$OPENAIOpenAI

Pereyra: law firms can't commit to a single lab due to client conflict risk

Law firms can't depend on a single provider: if a firm uses only Anthropic, OpenAI won't trust it with sensitive data to represent OpenAI. A multi-model abstraction layer is structurally necessary — analogous to Snowflake/Databricks on top of cloud.

Sourcery VC2026-06→
3rdmedium conviction
$ANTHROPICAnthropic

Pereyra: Opus 4.7 costs 3x more than GPT-5.5 but is only 10-20% better — an opening for open source

Opus 4.7's premium price (3x more expensive than GPT-5.5) is not justified by its marginal 10-20% performance gain. This will pressure Anthropic and create room for post-trained open-source models to close the gap and win routing.

Sourcery VC2026-06→

most discussed · click a bar to filter

6 total
$HARVEY
Harvey
MEDgabe pereyra·Sourcery VC·2 months ago· position
Harvey is building enterprise SaaS and agentic consumption in parallel
Harvey believes it needs two businesses at once: a traditional enterprise SaaS motion for today's buyers and a consumption/agentic model as legal AI matures. That dual architecture is meant to bridge current enterprise purchasing and future AI-native workflows.
"we're going to need to build two companies in parallel... one company is this traditional enterprise... and then the second will be kind of the transition you're starting to see n…"
0:53
$HARVEY
Harvey
HIGHgabe pereyra·Sourcery VC·2 months ago· position
Harvey reaches roughly $300M ARR as product velocity compounds
Harvey is about four years old, near 900-960 employees, around 2,000 customers and roughly $300M ARR. The key signal is that product build cycles are compressing while the company expands into a broader legal AI platform.
"We're at around 2,000 customers, somewhere around 300 million ARR. So, we're going really fast."
0:29
$HARVEY
Harvey
HIGHgabe pereyra·Sourcery VC·2 months ago· position
Harvey: a single query can cost $20K — the usage curve is just starting
A brutal usage inflection: simple queries cost $20, while reviews of 100K contracts reach $20K per query. Harvey is already one of the labs' largest consumers of embeddings, and demand is only just ramping with long-horizon agents.
"we have simple like assistant type queries where you say, you know, draft me a document that a single query can cost $20. We have um like a review product where you can upload a 1…"
20:05
$ANTHROPIC
Anthropic
MEDgabe pereyra·Sourcery VC·2 months ago
Pereyra: Opus 4.7 costs 3x more than GPT-5.5 but is only 10-20% better — an opening for open source
Opus 4.7's premium price (3x more expensive than GPT-5.5) is not justified by its marginal 10-20% performance gain. This will pressure Anthropic and create room for post-trained open-source models to close the gap and win routing.
"if you look at Opus 4.7 and 5.5, Opus 4.7 is three times more expensive than 5.5 but it's you know 10 or 20% more performant. But like the these are going to cause pressure on eac…"
27:51
$OPENAI
OpenAI
HIGHgabe pereyra·Sourcery VC·2 months ago
Pereyra: law firms can't commit to a single lab due to client conflict risk
Law firms can't depend on a single provider: if a firm uses only Anthropic, OpenAI won't trust it with sensitive data to represent OpenAI. A multi-model abstraction layer is structurally necessary — analogous to Snowflake/Databricks on top of cloud.
"if you just use anthropic or you just use open AI um you run into conflict risk. So imagine you're using only anthropics models as a law firm and you want to represent Open AI. Op…"
7:17
$HARVEY
Harvey
HIGHgabe pereyra·Sourcery VC·2 months ago· position
Pereyra: clients will pay $10M token bills and demand billable-hour-style itemization
Harvey will see a massive wave of demand for the middle layer (benchmarking plus routing) once companies start receiving $10M+ bills for token consumption. The parallel to law's billable hour will converge toward fragmented pricing and multi-model orchestration.
"all these customers are going to start getting these consumption bills of like $10 million. And they're going to be like, 'What did my agent do that cost me $10 billion?' And if y…"
25:50