Paris-based Shiplog has raised about €807,600, reported elsewhere as $1 million, in pre-seed financing for a customer intelligence platform built around an AI agent called Ada. The round is small by the inflated standards of generative AI, yet the problem Shiplog has chosen is large: companies collect detailed customer signals across sales, payments, product usage and support, then routinely reduce those people to broad segments.
The company was founded in 2026 by Khushi Mehta and Mehdi Gribaa. Kima Ventures and Project Europe backed the round, alongside Purple, No Label Ventures, 100IN, Station F Fund and angel investors, according to Vestbee and The SaaS News. Shiplog plans to put the money into product development, its Paris engineering team, integrations and early pilots in fintech, cybersecurity, software and consumer businesses.
From dashboards to decisions
Shiplog’s pitch is that Ada creates a live profile for each customer and selects a next action across the relationship, including onboarding, marketing, product experience, support, retention and expansion. The platform is designed to connect with existing systems such as Salesforce, HubSpot, Snowflake, Shopify and Stripe rather than asking customers to replace their core software.
That distinction matters. Most companies do not suffer from a shortage of customer data. They suffer from incompatible records, delayed analysis and unclear ownership of the next step. A sales team sees one account history, a customer-success manager another, while billing and product systems hold signals neither group checks regularly. A useful agent would need to reconcile those streams quickly enough to recommend an action while the information still has value.
Static segmentation remains popular because it is understandable and relatively easy to audit. A company can define an enterprise customer, a churn risk or a high-value shopper with visible rules. Individual AI recommendations may be more precise, but they create new questions: which data caused the decision, whether the recommendation is discriminatory, and who is accountable when an automated intervention irritates a customer.
The integration burden is the real product challenge
Shiplog therefore has to prove more than the quality of its model. Its integrations must be dependable, permissions must reflect the customer’s own access controls, and identity matching must avoid combining records that belong to different people or companies. In B2B software, where one corporate account can contain many users with different roles, an individual action can conflict with an account-level commercial strategy.
The startup’s position as an intelligence layer could help adoption because buyers can test it without a wholesale migration. It also exposes Shiplog to competition from the platforms it connects. Salesforce, HubSpot, Snowflake, Shopify and Stripe all have incentives to add more intelligence inside their own products. Shiplog needs to demonstrate that a neutral layer spanning several systems makes better decisions than each incumbent can make from its own data.
Privacy will be part of the sales case, especially in Europe. A system that assembles granular behavioral profiles needs clear data-retention rules, lawful processing and controls that let customers correct or delete records. Enterprise buyers will also ask whether their data is used to train shared models. Answers that can survive legal and security review may become a stronger differentiator than a marginal gain in recommendation accuracy.
The pre-seed round buys time to answer that question through pilots. Early customers will be valuable less for headline revenue than for evidence that Ada improves measurable outcomes such as activation, renewal, expansion or support resolution. Recommendations that look sophisticated but do not change those metrics will be difficult to defend in corporate budgets already crowded with AI experiments.
A focused French entry into agentic software
For France’s AI sector, Shiplog represents a pragmatic layer of the market. It is not training a foundation model or building costly computing infrastructure. It is applying agentic software to a specific operational bottleneck and relying on established business systems for distribution. That can require less capital, although enterprise security reviews and long sales cycles can still consume a young company’s runway.
The reported euro value also deserves context. The $1 million figure converts to roughly €807,600 at the rate used in European coverage, so the two headlines describe the same financing rather than separate rounds. At this stage, precision about valuation or ownership would be misplaced because those terms were not disclosed.
Shiplog’s opportunity rests on a genuine frustration with customer tooling: dashboards explain what has happened, while operating teams still have to decide what to do. Turning that gap into a trusted automated system is harder than producing another chart. If its pilots show that individual recommendations are accurate, explainable and commercially useful, this modest Paris round could establish a defensible place between data warehouses and customer-facing teams.
