Your agents are only as good as what they’ve read.
A document library (TeamAI calls them datastores) is the source of truth your agents read from. Upload your brand guidelines, case studies, pricing sheets, and past campaigns.
Part of Agent Garden, the self-serve path: build your own agent team on TeamAI.
Four steps from a folder of PDFs to agents that cite their sources.
No vector databases to stand up, no chunking strategy to tune. You organize documents the way you already organize your team, and retrieval is handled underneath.
Create a datastore
A datastore is a document collection scoped to a department or a project: a Brand datastore, a Case Study datastore, a Q3 Launch datastore. Plans include 10, 25, or unlimited datastores by tier, so the marketing library can stay separate from the sales library.
Load it
Dozens of file types plus web addresses: PDFs, catalogs, whitepapers, spreadsheets, even an entire website by URL. A workspace holds thousands of pages. Point a datastore at your own site and your agents have read every service page you’ve ever published.
Attach it to agents
Any custom agent can read from one or more datastores. Your email agent gets the brand library. Your proposal agent gets the case-study library. Datastores are shareable across the team, and workspace permissions control who works with what.
Get answers with citations
Agents retrieve the relevant passages for each request and annotate the answer with sources. Before a stat goes in a proposal, you can check it against the original document. Retrieval works in 50+ languages, in chat and inside any agent.
Four libraries that pay for themselves.
The pattern is the same every time: the documents your team keeps re-explaining to new hires, freelancers, and vendors become documents your agents have already read.
Brand guidelines that actually get followed
Your voice rules, approved claims, and banned phrases live in a Brand datastore. Every agent that writes anything (ad variants, emails, blog drafts, social posts) reads them before it writes. One source of truth instead of a style guide nobody opens, and fewer review cycles before copy ships.
Case studies and proof on tap
Sales asks for “that number from the HVAC case study” and the answer arrives with the source page attached. Proposals and outreach pull real results from real documents instead of a model’s invention, which is the difference between proof and a claim a prospect can’t verify.
Pricing and product facts in every ad
Spec sheets, service-area lists, and current pricing live in a datastore, so landing pages and ad copy claim what you actually sell at the price you actually charge. An ad with last quarter’s price costs real money and real trust. Update the document once and every agent works from the new version.
Campaign history that answers questions
Performance reports, creative briefs, and post-mortems go in as they’re written. Six months later, “what did we learn from the spring promo” gets a cited answer in seconds instead of an afternoon in the shared drive. Institutional memory stops leaving when people do.
Build the library yourself, or have WebFX run it.
TeamAI is the agent runtime WebFX built in-house. Production agents and production-tested skills run on it, and they run on curated document collections, the same way yours will.
You build the agents, load the datastores, and curate what goes in them. Full control, your pace, your team doing the upkeep. The right fit when you have someone who owns the library and the time to keep it current.
A TeamAI paid marketing agent team runs on this same infrastructure, reading from libraries WebFX curates: your account’s knowledge curated by a named Revenue Marketing Specialist, your industry’s refreshed quarterly, fed by real WebFX marketing work compounded over decades.
Same datastores, same retrieval, same citations. The difference is who does the curating. Start in Agent Garden and you can hand the marketing library to us later. Same datastores, same platform, so nothing has to be rebuilt.
Put your documents where your agents can read them.
Tell us what your team writes every week and we’ll show you which documents belong in the library first, and which agents should read them.