Coblocks

What do they actually do

Coblocks is a YC Fall 2024 company that previously built an AI‑assisted app for creating and running data pipelines. As of now, they have discontinued development of Coblocks and direct visitors to their new product, Kira; the Coblocks site explicitly states the discontinuation and routes sign‑in/join/demo links to withkira.com coblocks.ai. YC still lists Coblocks with a two‑person team, but there’s no live Coblocks product for new users today YC.

When it was active, Coblocks offered a single place to connect data sources, write SQL/Python transformations with AI assistance, run and inspect jobs, and schedule outputs. They promoted one‑click integrations (e.g., Stripe, HubSpot, Postgres, Snowflake), an execution engine powered by DuckDB, schema‑aware AI suggestions, templates, and branching/version control; their demo showed connecting a source, using AI to generate/fix queries, running blocks, and sending scheduled outputs to destinations like Slack YC, demo. For anyone evaluating them today, the team has pivoted to Kira, an “automatic CRM” for real‑estate workflows, and there is no Coblocks product available to test coblocks.ai, withkira.com.

Who are their target customer(s)

• Early‑stage founder or solo engineer who owns data: They want one place to connect sources, run transforms, and deploy pipelines without stitching multiple tools together because maintaining a DIY stack slows product work.
• Analytics engineer at a small team: They spend time fixing SQL/Python pipelines and infrastructure instead of shipping analyses; they need AI help and better runtime visibility to speed up authoring and debugging.
• Product or growth manager needing reliable, timely metrics: They get conflicting numbers from different tools and depend on engineers for scheduled reports; they want automated delivery (e.g., to Slack) with predictable freshness.
• Data analyst not fluent in SQL/Python: They rely on engineers for transformations and iterate slowly on fragile queries; schema‑aware, AI‑assisted query generation would help them produce correct queries faster.
• Platform/security lead at a growing company: They need permissions, sharing controls, and secure embedding so data work is auditable and safely shared across teams; lacking these blocks company‑wide adoption.

How would they acquire their first 10, 50, and 100 customers

• First 10: Run hands‑on pilots with YC founders and early startups, building their first pipeline live and guaranteeing a working outcome; use the demoable integrations and AI assistance to produce quick wins and case studies YC, demo.
• First 50: Offer a limited self‑serve trial with ready‑made templates (Stripe, Postgres, HubSpot) and short walkthroughs seeded in developer communities; convert via low‑friction paid pilots and referrals using early case studies YC.
• First 100: Add channel partners (small SaaS vendors, analytics consultancies) and ship permissions/embedding to win security‑minded teams; target ads to roles that converted early and optimize onboarding to get time‑to‑first‑pipeline under a day coblocks.ai.

What is the rough total addressable market

Top-down context:

The data‑integration/ETL/pipeline tooling market is commonly estimated at roughly $15–18B today, with some forecasts projecting growth toward the low‑$30Bs by 2030 Grand View Research, MarketsandMarkets, TBRC.

Bottom-up calculation:

Focusing on startups and SMBs (about 30–35% of current spend) yields a SAM of roughly $4.5–6.3B; capturing even 0.1–0.5% of that would translate to about $5–25M in ARR for an early vendor Grand View Research, TBRC, Matillion/industry commentary.

Assumptions:

• Market definitions vary (iPaaS vs. ELT vs. replication), so we use conservative mid‑teens estimates for TAM.
• SMBs account for ~30–35% of current spend, with faster growth than large enterprises.
• Illustrative penetration math (0.1–0.5% of SAM) is used to show revenue scale; actual conversion depends on product‑market fit and distribution.

Who are some of their notable competitors

• Fivetran: Managed connectors that replicate data from SaaS apps and databases into warehouses; overlaps on ingestion and reliability, but focuses on replication rather than an integrated AI authoring and execution canvas.
• Airbyte: Open‑source and cloud connector platform; strong for custom connectors and control over hosting, but lacks Coblocks’ packaged UI for schema‑aware AI suggestions and single run/inspect surface.
• dbt (dbt Labs): SQL‑first transformation with modular models, testing, and versioning; the standard for managing SQL transforms, oriented around code and CI rather than an AI‑assisted block editor with an embedded engine.
• Hex: Collaborative analytics notebooks/workspace with AI help, scheduling, and sharing; close overlap on enabling non‑engineers to query and publish, but positioned more as app‑building than pipeline execution.
• Prefect: Workflow orchestration and observability for data pipelines; competes on scheduling, retries, and run visibility, but is primarily an orchestration layer rather than a connector + AI authoring product.