AI Foundation · Data Model

Most CRMs store data.
Ektie's agents act on it.

Traditional CRMs were built for humans to fill in and managers to report from. Ektie's data model was built for a different world — one where AI agents read every field before they act, and write every outcome back when they're done.

Define your objects. Your agents will know what to do with them.

Every field is agent-readable and agent-writable · Agents update your CRM automatically

Three objects. Everything your
GTM team runs on.

Contact, Company, and Deal come built in — pre-wired so your agents can start working from day one. Every standard field is already accessible to the team. Extend them or leave them as-is.

Contact

Every prospect, lead, and customer. Agents read contact fields to personalise outreach and write activity, scores, and stage back after every touch.

nameemaillinkedin_urlsignal_scorestageownerreply_intent

Company

Accounts, target firms, and market segments. Morgan reads company context before reaching out. Rex reads firmographics to calculate pipeline value.

namedomainindustryheadcountarr_potentialicp_fit_score

Deal

Opportunities in motion. Astra reads deal fields to tailor her close strategy and writes stage, probability, and next-step notes autonomously.

namevaluestageclose_dateprobabilitynext_stepblocker

Your business isn't generic.
Your data model shouldn't be.

Build custom objects for anything your team tracks — partnerships, trials, events, subscriptions, transactions. Define the fields, set the associations, and your agents will read and write them the same way they do with contacts and deals.

Unlike traditional CRMs that bolt custom fields onto rigid templates, Ektie treats every custom object as a first-class entity — fully queryable, fully agent-accessible, and fully associated with the rest of your data.

Custom object examples
CustomPartnership
partner_nametierjoint_pipelinecontact
CustomTrial
accountstart_dateproduct_usageconversion_score
CustomEvent
namedateattendeesleads_captured
CustomSubscription
companymrrrenewal_datehealth_score

Every field has a permission.
Agents respect it.

In Ektie, every field on every object has two permissions: agent-readable and agent-writable. Agents read fields to inform their decisions. They write fields when they take action.

Agent-readable

Morgan reads a contact's LinkedIn URL before visiting their profile. Orion reads deal stages before deciding who needs coaching. Agents only act on what they know.

Agent-writable

Morgan writes signal_score after each LinkedIn touch. Astra writes deal stage when she advances an opportunity. Rex writes health scores to company records. Your CRM updates itself.

Object
Deal
Agent-operable
FieldTypeReadWrite
nameText
valueNumber
stageEnum
close_dateDate
probabilityNumber
next_stepAI
blockerText

Your CRM doesn't wait for
a human to update it.

Every agent in Ektie is connected to the data model. They read before they act. They write when they're done. The result is a CRM that reflects your pipeline in real time — without anyone manually updating a field.

1
Agents read before acting

Before Morgan sends a connection request, he reads the Contact record — name, company, LinkedIn URL, signal score, last touch. Every action is informed by what's already known.

2
Every action writes back

After Morgan visits a profile, the contact's last_touch field updates. When Astra moves a deal forward, stage and probability write back automatically. The CRM reflects reality, always.

3
The Brain layers on top

Morgan's Personal Brain holds context that goes beyond structured fields — conversation history, signals, coaching from Orion. CRM objects hold facts. The Brain holds judgement.

4
Orion audits from the data

Orion — the Sales Director — reads aggregate deal data, signal scores, and activity patterns across all contacts. His coaching is grounded in what the CRM actually shows, not guesswork.

5
Rex runs the numbers

Rex — RevOps — queries objects directly to produce pipeline reports, forecast models, and health scores. No manual exports. The data is always current because agents write it.

Build your data model.
Let your agents run on it.

Standard objects ready on day one. Custom objects whenever you need them. Every field readable and writable by the agents doing the work.