Clew.

The problem

Small business owners know something is broken in how their business runs, but can't articulate what — and a $50K consulting engagement isn't an option for a 3–15 person shop. The result: the journey is a loop, not a line. SMB owners cycle through "notice → tolerate → improvise → seek → defer" for years, across multiple broken workflows simultaneously, never reaching the diagnosis stage that would let them prioritize fixes.

The pain stack is specific: multi-system manual data entry (data re-entered into 2–3 systems with errors), no time or capacity to find/evaluate solutions (the owner is also doing the operational work), trust gap with AI-generated advice for business-critical decisions, and employee resistance to new technology when changes do happen.

The solution

Clew solves the meta-problem: not any single workflow pain, but the diagnosis that lets the owner see how their business actually runs and decide what to fix first.

A structured AI conversation surfaces workflow steps in plain language; the system generates a current-state process map without requiring the owner to know any consulting or technical framework; automation opportunities are scored; a prioritized first-fix recommendation plus a "cost of doing nothing" calculator give the owner a concrete next step rather than another vague suggestion.

How it works

Conversational flow → structured deliverable. The AI runs a structured interview adapted to the owner's tolerance for technical detail, holds the conversation on-topic (rejects requests for software recommendations or legal/tax advice), and handles vague users ("everything is fine") with re-engagement prompts.

Four-step deliverable: (A) Transition message signals the diagnosis is ready; (B) Current-state process map visualizes the workflow as it actually runs today; (C) Automation opportunities are scored on impact and effort; (D) Prioritized first-fix plus the cost-of-doing-nothing calculation gives the owner a concrete next step.

Primary model: Anthropic Claude Sonnet 4.5 (claude-sonnet-4-20250514), chosen for conversational depth, structured-output reliability, and tone fit. System prompt v2 at ~14,875 characters encodes the conversation rules, scope guardrails, and four-step deliverable structure. Replit-hosted prototype with hidden demo mode (press D or tap logo 3×) for live walkthroughs.

Who it's for

Primary persona: overwhelmed owner-operators of service businesses, 3–15 employees, who handle admin themselves. They are the bookkeeper, HR department, operations manager, and top IC — all in one person. Often non-tech-savvy, not consulting buyers, and explicitly service-based and operations-heavy rather than software-led businesses.

Customer segment: businesses with fewer than 50 employees, sweet spot 3–15. Engagement model: B2B, sold directly to the owner — no procurement layer, no IT department, no consulting RFP. The product has to feel like a consultant who actually listens in the first five minutes or the owner closes the tab.

Why it matters

The SMB software market is at 7–9% CAGR through 2035, the AI integration segment growing faster, and the global SMB IT consulting market is projected to exceed $400B by 2026 — but the AI-powered business process consulting niche for SMBs is essentially uncreated. Incumbents either serve enterprise (BCG, McKinsey) or sell generic AI assistants that don't do diagnosis.

Clew's structural bet: AI handles the discovery bottleneck. Traditional consulting can't economically serve SMBs because human discovery time is too expensive; AI conversational discovery, process mapping, and opportunity scoring collapses the diagnosis cost to a fraction of traditional consulting — opening a category that didn't exist because the unit economics never worked.

At a glance

Project

Clew

Built by

Manny Severino

One-liner

AI workflow diagnosis for overwhelmed SMB owners — conversational discovery, process maps, and prioritized automation opportunities.