Company Profile
1. What it is
The Company Profile is the platform’s one canonical store of facts about your company — what you do, where you operate, what you’re certified in, what makes you different, how you sound, and what your culture offers. Every app that needs company information reads it from here: AI agents ground their answers in it, RFP and bid responses cite it, recruiting pages describe your culture from it, generated emails and PDFs carry your identity from it. You confirm a fact once and stop being asked for it again — facts don’t change because a different app needs them.
- Access: requires the Administrator or Super Admin role, at Admin → Organization → Company Profile (directly above Business Profile in the sidebar).
- What it is not: it is not the Business Profile. Business Profile holds operational settings — work days, timezone, compliance and legal details like the Federal EIN. Company Profile holds the descriptive facts about your company that apps and AI reuse. The two link to each other.
- If you don’t see a “Research my company” button: AI research is enabled per tenant. Editing facts by hand is always available regardless.
The profile is organised into seven facets:
| Facet | What it holds |
|---|---|
| Identity | Legal name, headquarters, entity type, website, founding year, industry codes |
| Footprint | Industry, company size, markets and regions served |
| Offering | One-line description, products and services, who you serve |
| Proof | Certifications, differentiators, named customers, metrics you can cite |
| Voice | Tone and style guidance for anything written on your behalf |
| Culture | Values, perks, benefits — what recruiting and onboarding surfaces say |
| Posture | Competitive stance and positioning |
2. Standing it up
- Go to Admin → Organization → Company Profile. The page shows every facet, what’s filled, what’s confirmed, and what’s still missing.
- Click Research my company. An AI agent gathers facts from three directions: what the platform already knows (your business record, locations, company values, past RFP answers), your public web presence (researched from your website domain), and anything your own documents and contracts state. Research runs in the background — the button shows “Researching…” until the results are ready.
- If research reports it needs a website first, click Edit and fill in your website under Identity. Research works from your domain; a free email domain (gmail.com and similar) is deliberately not used.
- When research finishes, a Review N facts button appears. Click it to open the review queue — every proposed fact with its source, a confidence level, and a citation. Confident, low-stakes facts arrive pre-ticked; high-stakes facts (certifications, named customers, legal details) and anything the agent was unsure about arrive unticked so you decide explicitly.
- Tick what’s right, untick what isn’t, and apply. Everything you apply becomes an admin-confirmed fact — the strongest status a fact can have.
- Fill any remaining gaps by hand with Edit. Nothing requires research; typing a fact yourself confirms it just the same.
Try it with sample data first: Admin → Sample Data can load a demonstration profile — a partly-confirmed fabrication company with a certification about to expire and a ready research draft waiting in the review queue — so you can walk the whole flow before touching real facts. Clearing sample data removes only the demonstration facts; anything real is untouched.
3. How it fits together
Provenance — every fact knows where it came from. Each field carries its source: admin (a person confirmed it), internal (found in your own platform data, with a reference to where), web (found on a public page, with the URL), or derived (computed from other facts). The profile page shows this per field, so “why does it say that?” always has an answer.
Admin facts are locked. Research never overwrites something a person confirmed. A new research run can propose a change to a confirmed fact, but it lands in the review queue and takes effect only when you accept it. The queue is the only door back into confirmed facts.
The review queue is the anti-hallucination gate. The agent cannot write directly to your profile, and it cannot propose a web fact without a source URL or an internal fact without a reference. Anything it can’t cite is discarded before you ever see it.
Apps read, they don’t re-ask. AI agents across the platform receive a compact, confirmed-facts-first company context automatically. RFP Desk and Bid Fit draft from your proof and voice facets. Recruiting and candidate pages describe your culture. Generated emails, PDFs and portal pages carry your identity. When you correct a fact here, every one of those surfaces picks it up — there is no second copy to chase.
Sensitive fields stay home. A small set of fields (for example, competitive weaknesses you record for internal honesty) never leave admin surfaces — they are excluded from AI prompts, generated documents and the capability statement.
Write-through to Business Profile. A handful of facts live on both surfaces — website, founding year, industry, company size, description. Editing them here updates the business record too, so the two pages can’t drift apart.
Per-app AI instructions. When one app needs different wording — public-sector bids that must sound formal, recruiting copy that should sound warm — don’t fork the facts. Each consuming app can carry its own short AI instructions that style the output while reading the same profile underneath.
4. Running it
Reviewing research results
- Click Review N facts on the profile page (it appears whenever a research run is waiting).
- Proposals are grouped by facet. Each shows the proposed value, its source and citation, and the agent’s confidence.
- Tick or untick, then apply. Applied facts become admin-confirmed; everything else is discarded. Discarding is safe — a later run can propose again.
- If research found little (“your public pages didn’t give us much to work with”), click See what we tried — the usual causes are a missing or wrong website, or a site that says little about the company.
Re-running research
Research can be re-run at any time from the same button. Confirmed facts are never overwritten — new findings queue for review. The freshness digest (below) nudges you when a re-run is worth it, and after about 90 days without one.
Keeping it fresh
The profile is graded on three axes — how complete it is, how much of it a person has confirmed, and how current it is. Certifications carry expiry dates; anything expiring within 45 days flags the profile as ageing and appears in a freshness digest email to admins, so an expiring cert gets renewed in the profile before a bid or an agent cites it as current.
Generating a capability statement
Once enough of the profile is confirmed, a Capability statement (PDF) button appears — a one-page company summary built strictly from confirmed facts, suitable for attaching to bids and vendor registrations. Unconfirmed and sensitive facts never appear in it.
Fixing a wrong fact
Click Edit, correct the field, save. The correction is admin-confirmed immediately and every consuming surface reads the new value. There is nothing to update anywhere else.
5. Settings
The Company Profile has no toggles of its own — its controls are the page’s buttons.
| Control | Where | What it does |
|---|---|---|
| Research my company | Profile page | Starts a background research run; results queue for review |
| Review N facts | Profile page | Opens the review queue for the waiting run |
| Edit | Profile page | Direct editing of every facet; saves are admin-confirmed |
| Capability statement (PDF) | Profile page | One-page summary of confirmed facts; appears once the profile is ready |
| Per-app AI instructions | Each consuming app’s settings | Styles that app’s AI output without changing shared facts |
| Load / clear sample data | Admin → Sample Data | Demonstration profile and review draft; clearing removes only sample data |
6. More help
- Business Profile & General Settings — the operational twin: work days, timezone, compliance and legal fields.
- Sample Data — loading and clearing demonstration data across apps.
- AI Operations — platform-wide AI controls, including pausing agents.