Dari

What do they actually do

Dari provides a hosted platform to build and run browser-based automations that keep working on authenticated sites. The system learns a workflow once, caches the DOM steps, and replays them deterministically so runs don’t rely on an LLM each time; when a site changes, it falls back to AI to adapt the path and re-cache it (YC profile, product site).

It exposes APIs and webhooks to trigger, pause, and resume workflows from your stack, supports human-in-the-loop steps, and offers controls for high concurrency. Higher tiers add proxy/stealth options and enterprise plans indicate HIPAA/SOC‑2 support (docs, pricing, product site).

Who are their target customer(s)

• Engineers building product integrations and automation: They need browser-driven tasks to run reliably in production, but agents often break on logins, 2FA, or UI changes and are costly when they invoke LLMs every run. Dari claims deterministic replay, 2FA handling, and API/webhook integration to address these issues (YC profile, product site, docs).
• Finance and procurement ops teams using supplier portals and multi‑page forms: Work is repetitive, error‑prone, and expensive by hand, while off‑the‑shelf automations fail on complex flows and authenticated pages. Dari highlights procurement and form‑filling use cases and emphasizes caching successful workflows to reduce brittleness and cost (product site).
• Insurance and back‑office operations running massive parallel uploads or claims workflows: They need high concurrency and reliability across many vendor portals, but current tools don’t scale safely or handle varied auth flows. Dari advertises parallelized workflows, concurrency controls, and proxy/stealth options on higher tiers (product site, pricing).
• Healthcare IT and clinical operations working with EHRs: They must automate authenticated, sensitive tasks without breaking workflows or violating compliance, so brittleness or credential leakage is unacceptable. Dari lists healthcare as a use case and offers enterprise plans that mention HIPAA/SOC‑2 for high‑volume customers (product site, pricing).
• Small AI startups or data teams needing reliable data entry/scraping: Manual extraction or flaky agents slow development and waste engineering time; invoking LLMs on every run is costly and inconsistent. Dari says it caches DOM steps so automations replay deterministically until a site actually changes (YC profile, product site).

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

• First 10: Run targeted, paid pilots with startup and mid‑market engineering/ops teams via YC intros, developer communities, and direct outreach; assign an engineer to capture one real authenticated workflow (including 2FA) and prove reliability (YC profile, product site).
• First 50: Package the pilot into a low‑effort self‑serve trial with templates for common portals; publish concise how‑to guides and example automations for organic developer signups, and use a referral/discount to convert pilots into paid accounts (product site).
• First 100: Hire 1–2 sellers to close mid‑market deals in finance, insurance, and healthcare; build SI/RPA channel partnerships and use HIPAA/SOC‑2 readiness and enterprise pricing to land multi‑workflow contracts (pricing, product site).

What is the rough total addressable market

Top-down context:

The clearest umbrella market is RPA software, estimated at roughly USD ~18B in 2024 and projected toward ~USD 22–23B in 2025; adjacent web scraping (~USD 0.7–0.8B) and procurement analytics (~USD 5.7B) add overlapping upside (Fortune Business Insights, Market.us, Mordor Intelligence).

Bottom-up calculation:

Using Dari’s published pricing ($200–$2,000/month plus enterprise), an initial beachhead of 10,000–20,000 relevant teams (procurement, insurance ops, healthcare IT, data/engineering) at an average $5k–$15k ARR implies a USD ~$50M–$300M bottom‑up SAM for early segments, with broader TAM expanding as additional verticals and regions adopt (pricing).

Assumptions:

• Browser‑level agents are a subset of overall RPA/automation spend; not all RPA budgets are addressable by Dari.
• Adjacent market figures overlap with RPA, so they are not additive without deduplication.
• Average ARR and the count of relevant teams are illustrative to size an initial serviceable market.

Who are some of their notable competitors

• Playwright: Open‑source browser automation for scripting tests and tasks. Gives engineers full control but leaves production reliability, logins/2FA handling, scaling, and monitoring to the user.
• Apify: Cloud platform for hosted scraping/automation "actors" with scheduling, storage, and proxies—good for end‑to‑end scrapers but you build and maintain complex login/replay logic yourself.
• UiPath: Enterprise RPA suite with visual bots and controls used by back‑office teams. Powerful, but browser UI changes, modern auth flows, and high‑volume concurrency can require heavy engineering.
• Zyte: Managed scraping and anti‑bot service that runs headless browsers with proxies and can handle authenticated extraction; optimized for large‑scale scraping rather than deterministic, replayable multi‑step workflows.
• Browserless: Headless‑browser infrastructure as a service (Chrome at scale with sessions and proxies). Useful execution layer, but you still implement robust session replay, 2FA handling, and workflow caching.