Do we have to change how we work?

No. fmcg.network appears inside the assistant your team already uses. There is no dashboard, no second login and no new interface to train anyone on.

How do we know the answer is right?

Every record carries a data_quality label of beta, reviewed or gold, and every capability has a confidence threshold. Below it, results are flagged for review rather than guessed. Today's supplier dataset is labelled beta, and we say so on every result.

Who actually writes these capabilities?

Named industry specialists: regulatory consultants, sourcing agencies and data vendors, each publishing under their own name. Every submission is reviewed before it goes live, so an answer carries both an expert author and a network check. If you want to take a question further, the assistant can point you to the person who wrote it.

What happens to our data?

Telemetry records duration, capability and status code only. Question and answer payloads are not retained. No third-party trackers and no analytics cookies unless you accept them.

What if a capability does not know?

It says so. Unknown ingredients go to a review queue and lower the reported confidence. An honest "unknown" is the product working, not failing.

For FMCG teams

Stop losing weeks to supplier research

Your team already asks an AI assistant questions. Give it verified FMCG knowledge to answer them with: sourced, labelled and honest about what it does not know.

Install in your AI client See what it can answer

The same question, two worlds

Without the network

Generic chatbot guesses, with no sources and no quality label
Trade-fair contacts and a spreadsheet from 2023
Regulatory check happens after the artwork is designed
Nobody can tell you how confident the answer is

With fmcg.network

Answers carry sources and a data_quality label
Suppliers across DACH, PL and NL, growing continuously
Novel food pre-flight before budget is committed
Confidence reported per answer, unknowns disclosed

What's in it for your role

Procurement lead

Today: Co-packer search runs on trade fairs, spreadsheets and personal networks. It takes 2 to 6 weeks.
With the network: Find co-packers in 2 minutes instead of 2 weeks, filtered by region, certification, MOQ and category.
First result: search_suppliers returns a curated list with data-quality labels, inside your own AI client.

Regulatory affairs manager

Today: Novel food status surfaces after sourcing and labelling budget is already spent. External advice is expensive.
With the network: Pre-flight every recipe in seconds, with the rule reference and a severity per ingredient.
First result: Paste an ingredient list and get violations, cleared, unknown, plus a review_required flag.

Innovation and product development lead

Today: Every recipe iteration needs a fresh manual regulatory and sourcing check.
With the network: Each iteration checked against regulatory rules and supplier availability as you go.
First result: A capability call in the same chat where the concept was born.

AI-forward founder or ops lead

Today: No consultant budget, but the same compliance obligations as the incumbents.
With the network: Enterprise domain knowledge at self-service price, with no tool to switch to.
First result: Connector installed in 5 minutes, free tier usable immediately.

Available today

Both capabilities run against the live endpoint right now. More are in the directory, marked honestly as planned.

Livedata_quality: beta

Supplier Discovery

Find a co-packer in 2 minutes instead of 2 weeks, filtered by region, certification, MOQ and category.

Tool: search_suppliers
Publisher: fmcg.network Core Team
GPC: 5000000053000000

Livedata_quality: beta

Novel Food Eligibility Check

Know before you spend sourcing and labelling budget whether an ingredient is an unauthorised novel food.

Tool: novel_food_check
Publisher: fmcg.network Core Team
GPC: 50000000

Written by people, not scraped

Capabilities are authored by named industry specialists and checked by us before they go live. You get both: a working answer now, and the person behind it if you want to go further.

You can see the author

Every capability carries one named publisher. No anonymous model output, no unsourced summary.

You can check the reasoning

The decision logic, input and output schemas and data sources are published in the open, before you rely on them.

You can reach them

When an answer needs a human, your assistant can hand you straight to the specialist who wrote it.

Know something the network is missing? Publish it yourself, for free

Fair questions

Do we have to change how we work?: No. fmcg.network appears inside the assistant your team already uses. There is no dashboard, no second login and no new interface to train anyone on.
How do we know the answer is right?: Every record carries a data_quality label of beta, reviewed or gold, and every capability has a confidence threshold. Below it, results are flagged for review rather than guessed. Today's supplier dataset is labelled beta, and we say so on every result.
Who actually writes these capabilities?: Named industry specialists: regulatory consultants, sourcing agencies and data vendors, each publishing under their own name. Every submission is reviewed before it goes live, so an answer carries both an expert author and a network check. If you want to take a question further, the assistant can point you to the person who wrote it.
What happens to our data?: Telemetry records duration, capability and status code only. Question and answer payloads are not retained. No third-party trackers and no analytics cookies unless you accept them.
What if a capability does not know?: It says so. Unknown ingredients go to a review queue and lower the reported confidence. An honest "unknown" is the product working, not failing.

Install in your AI client