Ask Claude about your Tenderlake procurement notices in plain language. Once the Tenderlake connector is added to your Claude.ai workspace, you can have Claude summarise notices, surface deadlines, compare opportunities, or help you draft responses – all grounded in your own Tenderlake folders. Workgroups on Tenderlake Analyst additionally get market-wide analytics: competitor win/loss, buyer profiling, framework/DPS intelligence and semantic AI search across the whole procurement corpus.
The connector cannot change anything in your folders, notices, or settings. Nearly every tool is read-only; the only exceptions are two narrowly-scoped Analyst tools that create or update the analysis organisations used for competitor/buyer analytics (nothing can be deleted), plus the ability to send us product feedback.
No API keys are involved – you connect by signing in to Tenderlake and approving the connection, the same way you'd connect any modern app.
AI assistant access is granted per member by your workgroup's Tenderlake Admin:
If you're the Admin, you can grant yourself (and your colleagues) access on the same page. If you skip this step, the connection flow will show an "Ask your administrator" page instead of the approval screen.
Tenderlake is listed in Claude's connectors directory, so there is no URL to enter:
https://mcp.tenderlake.com/mcpThe sign-in and approval steps are the same either way.
Start a new chat in Claude.ai. Because connector tools are only available in new conversations after the connector is added, open a fresh chat rather than continuing an existing one.
Try one of these to confirm everything's working:
"Using the Tenderlake connector, which account am I connected as?"
Claude should tell you which Tenderlake workgroup the connection is bound to.
"List all my Tenderlake folders and tell me how many notices are in each."
"What AI searches do I have configured on Tenderlake?"
"Show me procurement notices from my Inbox folder published in the last 7 days. Group them by organisation."
"What's the deadline of every open notice in my Shortlist folder? Sort by deadline, earliest first."
"Summarise the top 3 notices in Inbox by their assessment score from my 'IT infrastructure' AI search."
"Find notices in Shortlist where the buyer is located in Germany or France, and draft a short pitch paragraph for each one based on what the notice actually asks for."
"Compare the scopes of work in the three most recent notices in Inbox. Where do they overlap, and where do they differ?"
"Translate the description of notice
<noticeId>into English and summarise it in 5 bullet points."
Workgroups with Tenderlake Analyst access see additional analytics tools alongside the basic ones. These work across the whole procurement corpus (within your plan's regions), not just your folders:
Try prompts like:
"We are <company>. We supply <what we supply>. We are considering an approach to <buyer>. Use Tenderlake for this and look at the last five years. Brief us on what they have procured in our space and around it, who their suppliers are and which of them we would be up against, what their notices and planning documents say they care about, and identify our possible wedges. Separate what the data shows from what you infer."
"How is Acme Ltd doing in UK public sector – what have they won in the last 12 months, and from which buyers?"
"Which frameworks are mainly used to buy IT services in the UK, and who operates them?"
"Run my 'lab equipment' AI search over the last 30 days for France and summarise the best matches."
Analyst tool calls are metered in credits (your workgroup's credits are topped up by agreement with us). Each metered response states what that call cost, and – if your admin has set you a monthly allowance – how much of your allowance is left, so your own spend is always visible. The workgroup's overall credit balance is shown to your admin in Tenderlake. If your workgroup isn't Analyst-entitled, these tools simply don't appear.
The connector can:
The connector cannot:
This is enforced on our side – not a convention. Every tool is checked against your workgroup's entitlements on every call.
When you chat with Claude using the Tenderlake connector, your conversation stays with Claude.ai. Tenderlake never sees it.
Think of Claude as a translator standing between you and Tenderlake. You ask Claude something in plain language; Claude figures out exactly what data it needs to answer you, and asks Tenderlake for only that. Tenderlake replies with the data, and Claude writes the answer back to you.
For example, you might ask Claude:
"Which notices in my Shortlist folder have a deadline before the end of the month?"
Claude reads your question, then sends Tenderlake a narrow request – something like "list notices in the Shortlist folder with deadline before 30 April." Tenderlake returns the matching notices. Claude uses those to write an answer for you.
What Tenderlake sees: the specific data lookups Claude makes on your behalf (folder names, date filters, notice ids) – exactly the same information you'd generate if you called the Tenderlake API yourself from your own software. We record these lookups in our standard request log so you can audit what was accessed, the same way we already do for every other API integration.
What Tenderlake does not see: the question you typed, anything else you discussed in the chat, Claude's reasoning, or the answer Claude wrote for you. That content stays inside Claude.ai and never reaches us.
Full details are in the Tenderlake AI Connector Privacy Policy, which supplements the main Tenderlake Privacy Policy.
To disconnect Claude.ai from Tenderlake, use either method:
https://mcp.tenderlake.com/mcp (no trailing slash) – or add it from the directory listing instead, which has no URL to mistype.