Callback

What do they actually do

Callback combines a software platform and services to turn manual operations into repeatable AI workflows called “Blueprints.” Teams define processes as small, validated steps, run them with orchestration and audit trails, and refine logic over time. Forward‑deployed engineers handle process discovery and implementation, wiring in retrieval and function calls; execution blends automated LLM steps with human quality checks where needed (site, blog, YC).

Customers use Callback today for tasks like invoice extraction, PDF report parsing, and image/data labeling—turning messy inputs into structured outputs for downstream systems. Engagements typically start with a demo and hands‑on onboarding rather than pure self‑serve (YC, site, blog).

Who are their target customer(s)

• Finance / accounts‑payable teams: They spend time manually extracting and validating line items from invoices and PDFs. They need a reliable, auditable way to convert documents into structured data for ERP and payment workflows.
• Data‑labeling / ML ops teams: Human labeling is slow and inconsistent, making it hard to scale dataset creation. They need higher‑throughput, quality‑controlled outputs with traceability.
• Operations analysts / reporting teams: They pull structured facts from disparate PDFs and reports by hand, which is error‑prone and delays reporting. They need repeatable extraction that feeds KPIs and reconciliations reliably.
• Compliance, legal, or regulated‑business teams: They must prove consistent decisions with audit trails and strict access controls. Ad‑hoc manual processes make it hard to satisfy audits and regulator reviews.
• Product or process owners with bespoke manual workflows: They lack engineering capacity to automate edge‑case‑heavy processes. They need a partner to map the process, ship an initial automation, and maintain quality when self‑serve tools fall short.

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

• First 10: Leverage founder networks and targeted outreach to AP, labeling, and reporting teams to run 4–6 week paid pilots delivered by forward‑deployed engineers, showing measurable accuracy/time savings and clear audit logs.
• First 50: Package the early wins into standard Blueprints (invoices, PDF parsing, labeling) sold by a small BD/sales team and boutique partners; publish short case studies and use referral credits to convert pilots into multi‑month contracts while reducing engineering time per onboarding.
• First 100: Productize onboarding with a library of editable Blueprints and lightweight integrations so common use cases can self‑deploy; add a partner channel and modest CS/SE team to drive land‑and‑expand, reserving on‑site engineering for complex or high‑risk workflows.

What is the rough total addressable market

Top-down context:

Near‑term categories include accounts‑payable automation (~$3B), intelligent document processing (~$2–3B), and data labeling (~$3–4B), with broader BPO spend around ~$300B globally (AP automation, IDP, data labeling, BPO).

Bottom-up calculation:

Summing AP, IDP, and labeling yields roughly $8–9B, but AP overlaps with IDP; a conservative de‑duplicated core TAM for Callback’s current workflows is about $7–9B today (AP, IDP, data labeling).

Assumptions:

• AP automation sits within the broader IDP/document automation category, so we de‑duplicate overlap.
• Scope limited to document‑to‑structured‑data and labeling workflows that Callback delivers today (platform + human QA).
• Near‑term addressability constrained by services‑heavy onboarding; full BPO upside depends on further productization.

Who are some of their notable competitors

• Hyperscience: Enterprise IDP for extracting and validating data from invoices, forms, and other documents with built‑in human review; overlaps on high‑accuracy, auditable document workflows for regulated customers.
• UiPath: Broad RPA and process‑orchestration platform with document understanding and strong governance; overlaps where customers need large‑scale orchestration, integrations, and audit controls.
• Ocrolus: Financial document automation focused on lending/finance (bank statements, paystubs, tax forms) with human‑in‑the‑loop checks; direct overlap for finance/AP document extraction needs.
• Scale AI: Data‑labeling and data‑operations platform with human‑in‑the‑loop pipelines and evaluation tools; competes for ML/Ops and labeling budgets rather than end‑to‑end enterprise workflows.
• Appen: Large provider of human annotation and data collection with managed QA and multilingual coverage; better for raw labeling capacity, while Callback pairs labeling with process discovery and workflow automation.