External Data Sources

External Data Sources are Collaboration.Ai's curated layer — data your team would otherwise have to research by hand across a dozen government sites and databases. We do that work upfront, curate it, and connect it to your Workspace, so it's already there, reasoning alongside Your Data, every time you ask a question. (For how this fits alongside Your Data, see The Knowledge Layer.)

External data arrives through three distinct services. They're often talked about interchangeably, but they work differently and answer different needs:

Data Packs — curated external datasets that Terminal draws on at question time as ambient knowledge. Nothing is written into your records; the knowledge is simply there whenever it's relevant.
Platform Organization Library — a shared library of organization records maintained at the platform level, extending discovery beyond the organizations your team already tracks.
Enrichment — a background service that fills in publicly sourced fields on your existing records, so profiles stay complete without manual data entry.

Data Packs

A Data Pack is a curated set of external data — award histories, entity registrations, open opportunities — built into your Workspace's knowledge layer as ambient intelligence. This is where a lot of the platform's real power lives: instead of Terminal summarizing what it can find, it's reasoning against real, curated datasets directly — the actual award record, the actual registration, not a search result pointing at one.

That's what makes an answer calculated rather than summarized: Terminal isn't guessing from a webpage, it's drawing a conclusion from the underlying data itself.

How that works in practice:

You don't go find a Data Pack — it finds your question. Ask Terminal something a pack can inform, and the system draws on it automatically. You never browse or search a pack directly, and it doesn't show up as records in your Network — it's knowledge the Workspace reasons with, not a place you visit.
Every answer still cites its source. You can see exactly which pack informed a given claim and follow it back — calculated doesn't mean opaque.
Pack data stays current at the source. Nothing is copied into your records; each pack is consulted at question time and refreshed on its own schedule — some near real time, others on a regular cadence — so you're always reasoning against the live dataset, not a stale copy of it.

What's in the catalog

DTIC Technical Reports Source: Defense Technical Information Center (DTIC).	DoW-funded technical reports and research. This is what lets Terminal answer "what defense research has already been done" by reasoning against the actual report — the real substance of past work, not a citation pointing at one.
Federal Hierarchy Source: SAM.gov.	Agency, department, and sub-tier office relationships across the federal organizational structure. This is what lets Terminal place an opportunity, an award, or an office within the actual structure of the federal government, rather than treating agency names as flat, unrelated labels.
Grants.gov Source: Grants.gov.	Federal funding opportunity announcements (NOFOs) that organizations can apply for. This pack is how Terminal answers "what grants are open right now" from the actual announcement, not a stale summary of one.
SAM.gov Organizations Source: SAM.gov.	Verified organization names, registration status, and identifiers. This is the official identity behind a name — when Terminal answers "who is this organization, officially," it's reasoning against the real registration record.
SAM.gov Funding Opportunities Source: SAM.gov.	Federal contract solicitations, RFPs, RFIs, and procurement notices. This is what lets Terminal answer "what's open right now" with a real, current opportunity — not out-of-date information.
SBIR / STTR Funding Opportunities Source: SBIR.gov, America's Seed Fund.	Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) awards and open solicitations. Ask whether an organization has received SBIR/STTR funding, and Terminal draws on the actual award record — phases, abstracts, and what the funded work was actually about.
USAspending Source: USAspending.gov.	Federal contract and grant spending activity. This is how Terminal answers "how much federal work has this organization done, and with whom" from real spending records, not an organization's own claims.

Note: Availability varies by Workspace: your CAI team pre-configures the packs that fit your program, and your Admin can adjust what's enabled at any time (see Managing External Data). If an Agent Skill depends on a pack that's turned off, the skill pauses until the pack is re-enabled — you're told exactly what's required; nothing fails silently.

Platform Library

The Platform Library is a shared layer of enriched organization records maintained by Collaboration.Ai at the platform level — a cross-sector network spanning industry, academia, and government, from anchor institutions and national labs to emerging companies. Every Workspace can draw from it, so discovery extends beyond your own rolodex from day one.

How it shows up in your work

You don't browse the library as a separate place — it surfaces where you work:

In answers and discovery. When Terminal researches a landscape, library organizations appear alongside your own records — extending results beyond the organizations your team already tracks.
When you save results. Library organizations in research results show as New; organizations already in your Workspace show as In-Network. Select and save New ones, and they become workspace records with their enriched details already filled in.
When you create organizations. If an organization you're adding matches a library record, its available details fill in automatically.

What's shared back

The library grows carefully, and sharing is controlled by your Workspace:

With sharing enabled, organizations your team adds contribute name and website URL only to the shared library. No files, no people, no attributes, no context.
Everything else about your records stays private to your Workspace, always.
Sharing is off unless your Workspace explicitly opts in, and the setting can be changed with your CAI team at any time.

Contributed organizations become eligible for enrichment from trusted public sources — which benefits your own copy of the record too. See the next section.

Enrichment

Enrichment automatically fills in publicly sourced facts on your records — for organizations, details like industry, headquarters location, employee count, founding year, and logo; for people, professional details like current role and work history — so profiles are useful from day one without manual data entry.

It runs in the background at two levels: on the Platform Library, keeping the shared organization layer complete, and in your Workspace, keeping your own records current. You don't trigger it — it's a managed service, quietly working.

What enrichment needs

Enrichment matches your record against trusted public sources, and it needs an identity anchor to match on:

Organizations — name plus Website URL.
People — name plus Email.

That's why the website field matters so much on organization records: without it, there's nothing reliable to match against.

Editing enriched values

Enriched data is yours to control. If an enriched value is wrong or your team knows better, edit it like any other field — you have full edit control over enriched fields on your records.