Shiplog has raised a little over €807,600 in pre-seed funding, equivalent to $930,000, for an AI agent that decides how companies should respond to individual customers. In a market where early AI rounds can become exercises in valuation theatre, the modest size of the financing may help the Paris startup.
The company was founded in 2026 by Khushi Mehta and Mehdi Gribaa. Its backers include Kima Ventures, Project Europe, Purple, No Label Ventures, 100IN and Station F Fund. Shiplog is based at Station F and says it is beginning partner pilots, adding integrations and expanding its engineering team in Paris.
The round is large enough to test, not to hide
Less than €1 million does not fund a long enterprise sales cycle, a large engineering organisation and a broad marketing campaign at the same time. The founders will have to choose. That constraint can be productive if they concentrate on a narrow set of customers and measure whether the product changes retention, conversion or support costs.
Shiplog’s Ada product connects with systems such as Salesforce, HubSpot, Snowflake, Shopify and Stripe. It is intended to update a customer’s profile as behaviour changes and select the next action across marketing, onboarding, support and customer success. That ambition creates a wide integration surface before the company has many people.
The sensible use of the round is therefore not to support every platform. It is to make a few integrations dependable for customers with costly churn and enough data to evaluate the decisions. Shiplog has identified fintech, cybersecurity, software and consumer businesses as initial sectors. Even that list may need to narrow during pilots.
Customer evidence must be stronger than an AI demo
An agent can produce convincing recommendations in a demonstration. Production use is less forgiving. Customer data is incomplete, product events change and companies disagree about the action they consider acceptable. A decision engine also needs permissions, audit records and a safe way for staff to override it.
Early customers should be able to compare Ada’s actions with existing rules or human decisions. The relevant evidence includes incremental retention, response rates, revenue and hours saved, adjusted for the cost of integration and mistakes. A vague claim that engagement improved will not support an enterprise purchase.
The founders also need to decide how much autonomy to offer. A recommendation presented to a customer-success manager is easier to approve than an agent that changes pricing, sends messages or alters a product interface automatically. Greater autonomy may create more value, but it also increases the cost of a wrong decision.
Capital efficiency can become a product advantage
A lean round forces Shiplog to reuse the customer’s existing data stack rather than replace it. That fits the product’s position above current systems. If deployments require months of custom data engineering, however, the startup will behave like a consultancy and consume its financing quickly.
The investor group offers networks across French and European startups, but a long list of backers can also create noise. The company needs operating help with pilot customers, security and senior engineering more than broad exposure.
Shiplog’s youth should not be romanticised. A team founded this year has limited evidence about renewal cycles, data drift and how its decisions perform during a customer’s difficult quarter. The advantage is that it can design around current AI economics without defending an older software model.
The €807,600 round buys a period of focused learning. If Shiplog can prove measurable outcomes with a small team, it will have a stronger basis for the next financing. If it scales headcount before the pilots answer that question, the small round will disappear without resolving the main risk.
