ProspectingConnector
Serper
Run fast, structured Google searches in Claude or ChatGPT through Tulina's own MCP server — the same engine this site's own processes default to.
Official site: serper.dev
What it can do
1 read
- 01Read
Run web searches and get structured, parseable results back
Fast, structured Google results, and the default choice for general web search across this site's own processes. Defaults to French results unless a query specifies otherwise — worth setting the country explicitly for anything outside that market.
How do you connect Serper to Claude and ChatGPT?
Through Tulina's own MCP server. Run fast, structured Google searches from the same chat as Claude or ChatGPT — the default engine behind this site's own processes. Defaults to French results unless a query sets the country explicitly.
Why through Tulina
The connection was never the hard part.
Most systems now publish their own way for a model to reach them, and that is good: it is what Tulina connects to. The hard part is deciding who may use it, what it is allowed to return, and what the model knows before it asks.
The keys stay in one place
YouTeamOrganisation
Connect Serper once. The credential is encrypted in a vault, never held in clear, and never pasted into a chat or a workflow file. A person uses their own key where they have one, their team's where they do not, the organisation's after that, so nobody is blocked and nobody holds more access than they need. Revoke it once and everything running on it stops.
What leaves is filtered before it leaves
fields droppedmasked or hashedpseudonymised
Roles decide who can reach Serper at all. Beyond that, a policy on the connector itself decides what comes back. The data is minimised at the source rather than after it has already reached a model.
Every call is on the record
whatwhowhen
Runs are logged for the organisation. That is what makes an automated action reviewable after the fact, and it is what a security review asks for first.
The run starts knowing your company
namingkey accountsdecisions
A direct connection gives a model access. It does not tell it how your company names things, which accounts matter, or what was decided last quarter. Tulina carries that across every tool, so the same request against Serper returns something usable instead of something generic.
Processes that run on Serper
Brief every cold call before it starts
A cold calling script opener for every account on the list, written from a dated signal and the lead's public words, saved on the record before anyone dials.
5 steps →Build a target list from a sentence
AI lead generation from one sentence: companies sourced and scored on dated signals, contacts verified, checked against your CRM and staged as a campaign.
6 steps →Find doctors and clinics
A census of every doctor and clinic in a French city, from map listings cross-checked against the practitioner registry and facility directory, with gaps named.
6 steps →Find local points of sale
Every target city is searched three different ways for shops that fit your positioning, each one ranked on fit and reviews before the first email goes out.
5 steps →Outbound from buying signals
Each morning, companies with a new stack or hiring signal are scored on your rubric, checked against the CRM and live sequences, and given one opener per buyer.
7 steps →Outreach based on government data
Every candidate's website is verified before a credit is spent, tiered by signal strength, then written one grounded opener that is checked for repetition.
7 steps →Outreach based on Slack membership
Membership in a Slack community is the qualification, so everyone in it is in scope, resolved by name and corroborated against their own words.
6 steps →Prepare for a demo from the last call
The last calls, the deal's state and fresh research become one prep brief, plus a deck built on that company's own use cases, posted before the demo.
5 steps →Qualify accounts by number of local merchants
Multi-location businesses qualified on a real location count: a published figure checked for scope and date, then a map count confirmed by a second source.
4 steps →
