A sales copilot that knows what to recommend.
Your reps ask. It knows every product, policy and portal, and says what to recommend.
$20 of free credit to try it. No card.
- RepCustomer has a laptop, serial NX4-2291. What can I offer them?
- AgentIts warranty ends in five months (checked in the warranty portal). Offer the 3-year extension with accidental damage. Skip premium support: they’re a 12-seat account, under the 50-seat minimum.
The problem
What to recommend lives in your best reps’ heads. New reps dig through catalogs, policy PDFs and portals, then guess. Copilots that only search documents hand back facts, not a recommendation.
That’s where most sales copilot pilots stall. The answers are accurate, reps try it for a week, then go back to messaging the colleague who always knows. Fact retrieval was never the hard part. What’s missing is judgment: when to recommend, and when not to.
Your best reps carry that judgment: which question to ask first, which detail changes the answer, which option wins, and when the usual answer doesn’t apply. Little of it is written down, so a copilot that only reads documents can’t have it.
How it works
Hand it your catalogs, price sheets, policies and the questions reps really ask.
Files go in as they are: spreadsheets, PDFs, slide decks, whole folders. Tell it which source wins when two disagree, like this year’s price list over last year’s deck.
Coach it on how your best reps decide: when to recommend, and when not to. It writes those rules into procedures.
Your expert talks to it like a new hire, in writing or out loud. Each correction becomes a rule in a procedure, a plain-text playbook for one kind of request, with the file each step came from. You keep or undo every change.
A rep asks. It looks up what it needs in your portals (a warranty, stock, an account) and recommends.
When the deciding fact sits behind a login, it checks it in a cloud browser, the way a rep would. The answer is one recommendation, the reason, and what would change it, with each fact tied to where it came from.
A worked example
A rep at an office-equipment reseller is quoting laptops. The reseller, the customer and every number here are made up.
What it did
- Asked the question its procedure puts first: travel and offline files decide the model.
- Opened the partner portal in its cloud browser for the account’s docks, its contract price tier and stock.
- Applied the team’s rule: people who travel with client files get encryption and accidental-damage cover.
- Gave one recommendation, the reason and the source of each fact, and said what would change it.
- RepAcme Legal wants 40 laptops for new associates. Which model, and which support plan?
- AgentTwo things decide it: will they travel, and will they keep client files on the laptop?
- RepMostly travelling, and yes.
- AgentQuote the 14-inch Pro, 512 GB, with encryption on, and the 3-year onsite plan with accidental damage. Their offices already run Pro docks (partner portal), and 40 units puts them in contract tier B until November 30. One catch: the Pro is backordered three weeks. If they need them sooner, offer the Air with the dock adapter.
What Coach asks you
Coach reads your files first and sets itself up in one change you review. Then it asks about what the files leave unclear, in a few short questions. For a sales copilot, they sound like this.
Each answer becomes a rule in the procedure for that kind of request. You see the change in place before you keep it.
- AgentYour price list and the partner deck disagree on the Pro’s price. Which one wins?
- AgentWhen would you not recommend the extended warranty?
- AgentWhat do you need to know before you recommend a model: seats, travel, budget?
- AgentIs a backorder a reason to switch models, or to wait?
- AgentWhich requests should go to a sales engineer instead of me?
How evals check it
Before reps rely on it, test it on their real questions. A test case is a request and what a good answer does, in plain words. Each case runs three times, and a judge checks that the answer is right and grounded in what the agent found, not guessed.
| The request | A pass |
|---|---|
| “Which support plan for a 12-seat account?” | A pass: Doesn’t offer premium support (50-seat minimum), and says why. |
| “The customer wants the Pro, but needs it Friday.” | A pass: Checks stock in the portal and offers the model that’s in stock. |
| “Which laptop for a new hire?” | A pass: Asks about travel and client files before recommending. |
| “Can we match a competitor’s price?” | A pass: Hands it to a sales manager instead of promising a discount. |
Re-run the set after every correction, so a new rule doesn’t break an old answer. After Coach changes the agent, it can run the evals on its draft and tell you how it did against the last run.
Why Eager
- Judgment, not search: it follows your team’s decision rules, not just your documents.
- It checks the live answer in portals that have no API.
- Tested on your reps’ real questions before they rely on it.
- Where it answers: In your reps’ tools through the API, or in the Eager chat.
Questions
Read more
- How is it different from a copilot that searches documents?
- A document copilot finds facts and hands them back. This one is trained on how your team decides, so it asks for what’s missing, recommends one option and says what would change it. It also checks live facts in portals, which documents can’t hold.
- Can it recommend SKUs and configurations?
- Yes. Hand it your catalogs, compatibility rules and price sheets, and coach it on the trade-offs your sales engineers weigh. It recommends a configuration and shows the rule and the file behind each part of it.
- Does it need an API to our CRM or partner portal?
- No. Where there’s an API, connect it as a tool. Where there isn’t, it uses the portal in a cloud browser, as a rep would, and you can replay every step.
- How do we capture our best reps’ know-how without a documentation project?
- Collect 20 to 30 real requests with the answer your best rep gave. Let the agent try them in Coach, and have the rep correct the wrong ones in plain words. Each correction becomes a rule in a procedure.
- What happens when it doesn’t know?
- It asks for the missing detail, says it can’t recommend yet, or hands the request to a person, as you taught it. Teach it which requests need a specialist, and test that it hands them over.
- Where do reps use it?
- In the Eager chat, or inside the tools your reps already use, through the API and the JavaScript and Python SDKs.
- What does it cost?
- You pay for the tokens it uses, from a prepaid balance, at the price of the model it runs on. Your first team gets $20 of free credit, no card.
Other jobs it learns
Support agent
Answers your customers and fixes it in your systems, instead of linking an article.
Behind your voice agent
Your voice agent talks. Eager does the work in your systems while the caller holds.
On your customer’s computer
It does the fix on their Mac or PC while they watch, with Stop always in reach.
Patient intake
Finds the chart, checks coverage in the payer’s portal, and tells the patient what they owe.
Freight
Reads the truck’s live location, converts its time zone, and moves the pickup in your shipping system.
Claims
Catches the swapped digit in a case number, reads the right report, and keys the claim.
Payables
Checks a bank-change request against the supplier’s number on file before anything changes.
Start with one job.
Pick the request your team handles most. Hand it the files. Watch it work before you spend a cent.