Shiplog raises $1M to build AI customer intelligence for B2B SaaS

Shiplog, a Paris-based startup working out of Station F, has closed a $1 million pre-seed round from Kima Ventures, Project Europe, Purple, No Label Ventures, 100IN and Station F Fund. The pitch is straightforward: most B2B SaaS companies still market to customer segments drawn up weeks ago, even though usage-based pricing and AI have made every customer behave differently by the day. Shiplog's answer is an AI agent called Ada, which watches customer behavior in real time and decides what happens next for each individual, rather than for a bucket of thousands.

Co-founder and CEO Khushi Mehta, 23, spent six years in go-to-market roles before landing on the idea, most recently running marketing for an agentic commerce infrastructure company. She teamed up with Mehdi Gribaa, 25, a computer engineer who'd shipped AI products in finance, healthcare and retail, after meeting through the Entrepreneurs First program. Their shared observation: engineering teams ship features fast, especially with AI, but marketing tools haven't kept pace, turning marketing into what Mehta calls a bottleneck instead of a growth engine.

The mechanics are fairly specific. Ada pulls signals from support tickets, emails, calls, location and product usage, then builds a living profile of each customer instead of relying on a static score a human has to interpret later. Mehta's go-to example is fintech KYC abandonment — instead of blasting everyone who drops out with the same three-day reminder, Ada tailors the follow-up based on that person's actual history with the company. The same logic extends to onboarding flows, paywalls and even figuring out who inside an organization is the actual buyer versus who's just clicking through a trial, by cross-referencing enrichment data with usage patterns.

Shiplog plugs into existing tools like Salesforce, HubSpot, Snowflake, Shopify and Stripe rather than replacing them, acting as a decision layer that gives marketing, customer success and account management a shared, live view of each account. Every AI action gets logged with sources attached, and nothing goes out to a customer without a human clicking approve first — Mehta was explicit that teams can edit any AI-drafted message before it ships. There's also a conversational interface, so someone can just ask Ada to surface expansion candidates for the week instead of digging through a dashboard.

The company has found early traction in fintech and cybersecurity, industries with huge customer bases where dedicating a human account manager to every account simply isn't economical. Before launching publicly, Shiplog ran a pilot processing more than four million events across roughly 1,000 customers to prove the system holds up at scale. Mehta's bet is that vertical focus — accumulating knowledge about exactly where fintech onboarding funnels leak, for instance — becomes the moat, rather than trying to be a generic AI layer for every industry at once.