S26 mid-batch survey

N=38
01

How would you categorize your startup?

Agentic B2B SaaS
1642%
Hard tech
821%
AI-native B2C
616%
Devtools/infrastructure
616%
AI-native service
13%
Defense
13%
02

How many hours a week do you work?

>80
2668%
70-80
718%
50-60
25%
60-70
25%
<50
13%
03

Have you pivoted at any point during the batch yet?

No
2361%
Yes
1539%
04

What’s your favorite coding agent tool (i.e. which one would you be most disappointed to lose access to)?

Claude Code
2258%
Codex
718%
Conductor
38%
Hoplite (our own tool)
13%
JCode / Claude Code / Paseo
13%
OpenCode
13%
Orca
13%
Our own tool
13%
Superset
13%
05

Which LLM providers are you using? (select all that apply)

Claude
3695%
OpenAI
3182%
Gemini
411%
Fireworks
13%
Lilac
13%
Poolside
13%
Together.ai
13%
Wafer
13%
06

What hosting platforms are you using? (select all that apply)

AWS
1745%
Vercel
1642%
Google Cloud
1129%
Cloudflare
1026%
Railway
1026%
Azure
38%
Not deployed yet
38%
Bare metal
13%
Hetzner
13%
Modal
13%
Render
13%
07

What sandbox providers are you using? (select all that apply)

We don't use sandboxes
2258%
Modal
718%
We roll our own
513%
Daytona
25%
E2B
25%
Cloudflare
13%
Smol Machine
13%
08

What database and storage providers are you using? (select all that apply)

Supabase
1232%
S3
1129%
Hyperscaler Postgres (e.g., RDS and Aurora)
616%
Neon
616%
R2
616%
No DB yet
38%
PlanetScale
38%
ClickHouse
25%
Firebase
25%
MongoDB
25%
Convex
13%
DynamoDB
13%
09

Do you use GPUs?

No
2463%
Yes
1437%
10

Where do you run your own internal agents for coding and other tasks? (select all that apply)

On my laptop
3387%
In the cloud
2258%
11

Who do you use for SOC 2?

No need for compliance (yet)
1950%
Oneleet
718%
Vanta
718%
Undecided
25%
Comp AI
13%
Not yet
13%
Rippling
13%
12

Preferred corporate card/banking platform?

Mercury
1334%
Brex
821%
Rho
821%
Every
38%
Ramp
38%